Daily innovation for marketers

When a webhook fires, the post includes the related records you selected for that webhook.

Toll-free SMS and MMS credit rates applied to Automation sends

Automations that send from a toll-free number deduct toll-free credit rates for US and Canadian recipients:

• SMS: 1.5 credits per 160-character segment for US recipients; 2.5 credits per segment for Canadian recipients.
• MMS: 4.5 credits per 1,600-character segment for US recipients; 7.5 credits per segment for Canadian recipients.

Totals are rounded up once across the full send. Recipients in any other country continue to use the standard country price. Automations using a 10DLC number are unchanged.

SMS reply metrics in block-level reporting are attributed to the most recent delivered send of the same block to the same Contact before the reply, rather than by the reply's own date.

A reply that arrives after a reporting window closes is counted in the window when the send that prompted it occurred, not the window when the reply arrived.

Automation SMS per-block reporting attributes clicks, replies, and opt-outs to the most recent successfully delivered send that preceded each event. Previously, a block sent twice to the same contact that received one click showed a 100% click rate; it shows the correct 50%. This correction applies to all three event types.

Tags in the path, query string, and fragment all work. Tags in the hostname are still rejected to prevent open redirects, and the http/https restriction is preserved.

Automation Canvas if steps with multiple branches can route Contacts through field, tag, list-status, and segment criteria.

Field and tag branches. Contacts are routed to the first branch whose field value or tag condition matches, evaluated in branch order.

List-status branches. Contacts are routed by their subscription status on the named list — subscribed, unsubscribed, unconfirmed, bounced, added, never added, or not subscribed. Membership is checked directly without building a segment per branch.

Segment branches. Contacts are routed to the first segment branch they match (in or not-in).

If no branch matches, the Contact takes the No Match path. Classic if steps continue to route as before, unchanged.

In the campaign editor's Link Actions panel, tag names and other link action labels with no spaces — for example, names joined with underscores — wrap onto multiple lines and stay inside their container. Short names and hyphenated names are visually unchanged. This applies to all link action chip types: tags, lists, and Automations.

Labels can store a description of up to 500 characters, editable when creating or updating a label — omitting the field on update leaves the existing description unchanged, and sending an empty string clears it.

When two reseller documents in your knowledge base give conflicting information, Active Intelligence uses the one uploaded more recently. The upload time is shown in knowledge base search results. Every other source — Brand Kit, voice profile, and so on — still takes priority over documents.

The Enrich Contact Automation action can write or overwrite a contact's email address using data from your enrichment vendor.

Two modes are available:

• Fill-empty writes the vendor email only when the contact's email field is blank.
• Overwrite replaces the stored address after the lookup.

If the vendor returns no email, the current address is left unchanged. If the vendor email is already associated with a different contact, the field is also left unchanged.

The email editor's authentication step previously had no timeout and no retry logic, causing load failures on transient connection problems. The auth request now has a timeout and retries once on a connection error, so a brief network hiccup no longer prevents the editor from loading.

When deleting a contact also removes their deal, the deal's notes, tasks, and activity are deleted along with it.

When you delete an account, any deals that belonged only to that account are deleted along with it, including their notes, tasks, and custom field data. If the account had a secondary contact, the deal moves to that contact instead.

When you save a deal that already has custom field data, the save path reuses the records it already fetched rather than fetching them a second time. The practical result: those deals save faster and are less likely to time out during busy periods.

The GET /api/3/contacts endpoint with include=contactAutomations no longer times out or runs out of memory on accounts with large numbers of Automation entries.

Large imports and contact syncs that previously stopped partway through with a server error finish instead. The API response shape and values are unchanged.

In long-running processes, Automation block counts can be up to 60 seconds stale for display purposes only — this is never saved to your account.

The description field is writable via the labels API (POST and PUT /api/3/labels). On update, omitting the field leaves the existing description unchanged; sending an empty string clears it. The description field in the label modal ships as part of the same update.

When the enrichment vendor returns a ranked list of languages, the Enrich Contact action compares it case-insensitively against the options you've set up in your preferred-language custom field and writes the first match. If the field is empty, it fills it in; if you've set the action to overwrite, it replaces whatever was there. When the vendor returns no language data, the field is left untouched. If none of the vendor's languages match any of your field options, the rest of the enrichment run still completes normally.

This applies to the preferred-language field specifically. All other field-level controls — choosing which fields to fill and whether to fill or overwrite — work as described in the Enrich Contact action settings.

Brand monitoring previously discarded pages that lived on a sibling or parent subdomain of a brand's website — for example, pages on a subdomain sharing the same registered domain as your brand's site would be lost.

Pages on any host sharing the brand's registered domain are retained when the site map returned them. Pages on unrelated domains are still excluded. Duplicate www. variants of the same page are deduplicated to a single entry.

Bounce webhooks always carry complete data — contact, campaign, bounce type, code, and description — and the payload shape is unchanged.

Segment by Automation — filter Contacts by the automations they've completed

You can use Automations as a condition when building segments. A prompt like "contacts who completed the Welcome Series" resolves to that specific Automation. Up to 2,000 Automations appear in the catalog, including search results; hidden Automations are excluded.

The Automations Performance report gives you a faster, richer view of how your Automations are performing.

KPI summary bar. A new summary bar at the top of the report shows the count of active Automations and aggregate send, open, click, bounce, and unsubscribe rates across all Automations matching your current filters. The summary updates with your filters and is independent of sorting and paging, so it always reflects the full filtered set — not just the page you're on.

Name search. A search field filters the table, the summary bar, and CSV exports simultaneously, so what you see on screen matches what you export.

New table columns. Each Automation row includes labels, created date, last triggered date, and a screenshot (the latest captured image of the Automation). Columns can be sorted by date.

When Active Intelligence generates a segment inside the Automation Canvas, it resolves names in your prompt against your account's actual lists, tags, and Automations before building the segment. Previously, names were passed through unresolved, which could produce segments that never matched any contacts.

As a safeguard, any condition that cannot be resolved to a valid identifier is rejected before the segment is saved, so a broken segment is never persisted. The segment builder and other callers are unaffected.

If a trigger or action isn't available on your plan, it still appears in Active Intelligence's view with a note explaining why it can't be used, so Active Intelligence can tell you what's available and why some options are restricted.

The "Wait until a specific day" step matches its scheduled time within a 15-minute window. An Automation set to fire at a precise minute still triggers even if that exact minute is narrowly missed.

When building an Automation email series, Active Intelligence preserves email subjects you explicitly named, applies them on every email completion path, and correctly looks up saved templates by name before generating a layout. A template chip forwarded on create is no longer dropped.

Relative Day N + time waits are understood without requiring calendar dates in the prompt. Asking Active Intelligence to "write the emails for me" on a new series is treated as a create operation.

Active Intelligence no longer attempts to restyle the "Marketing by ActiveCampaign" branding on Forms. Previously, it would write CSS targeting the branding element, which had no visible effect and produced a false confirmation. Active Intelligence now declines these requests and explains that branding can be shown or hidden but not restyled.

Forms that already have branding CSS in their custom CSS field will have it removed when the form is visually edited.

When editing a Campaign through the MCP route, requests that include AI image generation are declined before anything is changed. The response states that nothing was changed and suggests using a content-manager image instead. Edits and creates that don't request image generation are unaffected. A mixed edit — for example a subject change combined with a generated image request — is declined as a whole.

The warning appears at the moment you make the format switch, before it takes effect. This reduces the risk of breaking integrations that depend on the existing format.

The email editor's interface has been updated to match ActiveCampaign's design system. Controls in the side panel — toggles, tabs, dropdowns, and buttons — follow consistent component styles.

The canvas uses a unified green color for container, structure, and stripe frames, replacing the previous mix of blue, red, and indigo. The scrollbar on the blocks panel appears only when you hover over it.

When starting a new toll-free SMS number registration, the Business Location step opens with a country already selected.

The country is taken from your account's default address when it is US or Canada. If there is no default address, or the default address is outside the US or Canada, the country defaults to US. A country you've already chosen or a stored answer is never overwritten.

Cards are pinned to a consistent 158px width. Previously, cards on a short last row could grow up to 220px, making them visibly wider than cards on full rows above them. Design templates, Brand Kit cards, AI brand cards, and their loading skeletons are all affected. Layout template thumbnails retain their 168px width and also no longer stretch on short rows.

The Tag ID column appears after the Tag Name column in the Manage Tags list. You can read and copy a tag's ID directly from there. The column is not sortable.

The "Edit email" button is hidden in the send block modal when the underlying Campaign is a split Campaign, so the option no longer appears where it can't complete.

Clicking a series card that has no associated message opens a validation check before the sender details modal appears. Previously, the modal opened and only surfaced the problem after you filled in the sender details. The check runs first, so you get immediate feedback.

The segment contact preview tab shows a question-mark icon next to each contact. Hovering or focusing it loads a plain-English explanation of why that contact does or does not match the saved segment, broken down condition by condition. The explanation includes loading and error states with a retry option.

The tip shows "2 of 2" in the footer with a "Done" button as the only dismiss action. Closing with Done or sending a task ends the tour.

Keyboard focus triggers the same animation. If you have reduced motion enabled in your system settings, the illustration appears in place without movement.

Two visual fixes to the Scheduled Tasks create and edit form make it easier to fill out correctly.

The task name field shows a visible label reading "Scheduled task name." The description text under the New task, Schedule, and Notify by email sections uses the default text color instead of muted gray, making the copy legible against the form background.

Hovering the Scheduled Tasks clock icon in the navigation bar triggers the full animation, including the upward jump and the circle stroke drawing with the arrow-dot pulse. The resting appearance of the icon is unchanged.

On the US/Canada toll-free SMS registration flow, the "How contacts opt in" question on the Opt-In Information step pre-selects "On a web form" when no answer has been saved yet. A previously stored answer still takes precedence.

In the SMS inbox, you can draft a reply using Active Intelligence alongside the existing manual draft option. The composer shows a "Draft reply with AI" button next to "Draft reply", plus labeled options for Personalization, Quick responses, and Uncategorized. The conversation summary caption reads "Generated with Active Intelligence" so you can see at a glance how a summary was produced.

When a contact has a linked deal, the inbox can now pull that deal's owner and assign them to the conversation automatically — so the right person is handling the thread without anyone having to do it manually.

To activate it, go to your inbox settings, enable assignment, then check the new Auto assign deal owner to the conversation checkbox that appears beneath the existing responder assignment toggle.

Toll-free sending bills 1.5 credits per segment (US) and 2.5 credits per segment (Canada), producing decimal balances. The balance footer formats to at most one decimal place using your account's locale, plural labels ("credit" vs. "credits") resolve correctly at display precision, and the low-credits warning banner triggers for any balance below 1 credit — the minimum needed to send a single segment.

SMS composer shows a credit range estimate for toll-free sends

The credit range estimate reflects each recipient's country rate. A Tool Tip lists each toll-free country rate and notes that the final cost depends on each recipient's country. Current rates are 1.5 credits for US recipients and 2.5 credits for Canadian recipients.

The send and schedule buttons require that your account balance can cover the lowest possible cost of the send. The low-credits banner in the audience step uses the same figure.

MMS sends from a toll-free number show no estimate.

A sender type selector appears in the SMS Campaign builder when your account has both a standard approved registration (10DLC, UK, AU, or DK) and a toll-free registration (approved or pending). You can choose between your default sender and your toll-free number.

A few things to know:

• While a toll-free registration is pending, helper text explains that sends start after approval and the number stays hidden. Once approved, the number is shown.
• Sending, scheduling, and test sends are gated on the sender your Campaign uses: toll-free requires an approved toll-free registration; the default sender requires an approved standard registration. Saving is always available.
• Toll-free reaches US and Canada only. If your account has another approved country registration plus toll-free, they're treated as two distinct senders.
• The Automation Canvas SMS block is not included in this change.

Picking, deselecting, and re-picking a list in "Subscribed to SMS list(s)" or a segment in "Send to" in the SMS builder works without locking up the page.

The revision history menu remains anchored below the page header when opened.

Refreshed SMS inbox conversation summary and reply composer

The conversation summary in the SMS inbox is a collapsible accordion. It's collapsed by default and labeled "Conversation summary" — expand it to see the generated content, which carries a "Generated with Active Intelligence" attribution. Long summaries scroll inside the panel rather than pushing the rest of the page down.

The draft-reply composer is restyled to match. The AI option is labeled "Draft reply with AI", and Personalization and Quick response controls are icon buttons sized to match the Send button, so the composer feels consistent and uncluttered.

A failed feature-flag network request could previously cause the app to become unusable for your entire session. A failed or malformed response errors cleanly without affecting the rest of the app.

The Active Intelligence permissions customization grid has a new structure: a Plan Mode autonomy heading with three autonomy tiers and descriptions, a full-autonomy confirmation dialog, an advanced per-resource panel, a resource-level caveat note, and nine group headings that organize the advanced permissions grid.

Any previously saved rules tied to renamed permission keys will appear as unlisted on the customize page. No automatic migration is provided — review and re-apply any affected rules.

When you act on an account through the Partner Portal, the top-bar chip displays the account's level in your hierarchy — shown as "Level N" after the account name (for example, "… | Level 2"). Level 0 is the top-level anchor account.

This makes it clear which rung of the account hierarchy you're currently acting on, so there's no guesswork when navigating between accounts during impersonation. When the backend does not report a level, the chip displays as before.

Previously, tabs that were slow to start because they were queued behind other tabs loading at the same time could miss a deadline and show a top-level error, even when the server itself responded quickly. Tabs that are queued or throttled load late rather than failing. A tab that is genuinely waiting on a hung request still times out.

The manage user seats modal has been updated. The quantity card spans the full width of the modal. If you submit a change and hit a rate limit, the modal surfaces that message directly. The "Review purchase" title is correctly translated.

When an MCP connector fails to connect, you see a plain-language explanation rather than a raw error code. Three specific failures are covered: authorization sign-in refused by the provider, the connector could not be reached, and the connector URL was rejected. Any other failure shows "Could not connect [server name]."

If you try to create a custom connector with a name that already exists, the form stays open and asks for a different name rather than showing a generic error. Other create or preview failures show the detail returned by the server.

The dashboard builder opens with an "Add KPIs" slot. Selecting it opens a widget gallery where you can browse KPI cards by category, search by name, preview values, and select up to five KPIs to add as a row to your dashboard. The footer shows your current selection count and the five-KPI limit; saving adds the row to the draft in the order you selected.

The partner tab can be renamed inline — press Enter or click away to save, or Escape to cancel without closing the builder.

Inline field editing on the Contact detail page uses clearer icons with Tool Tips and shows a save confirmation toast in the top-right corner after a field is saved.

The fix applies in both the Content Manager modal and the Content Manager page, so the size limit is consistently enforced whether you're working in either location.

The label on an Ideas card that previously read "Update from ActiveCampaign" reads "Latest update" instead, keeping the label brand-neutral for accounts running under a reseller or partner brand.

The Accounts page in ActiveCampaign HQ (ACHQ) replaces the static Type filter — which offered fixed Client, Partner, and HQ options — with a Hierarchy filter built from your actual account structure.

The filter options are generated from the real depth of your partner account subtree. You get one entry per depth level present (Level 0, Level 1, and so on), a Leaves option that shows only accounts with no children, and an All option that restores the default direct-children view.

This change is not yet available to all accounts.

A guided tip on the navigation bar's clock icon shows you where Scheduled Tasks live. When you visit the Active Intelligence home page, a short tip card appears on the clock icon with a looping animation explaining that you can create and manage Scheduled Tasks directly from the navigation bar. Clicking "Next" opens the Scheduled Tasks page and queues the next step; clicking the clock icon itself takes you straight there. The tip appears once and clears after you interact with it.

Medium-weight text throughout the platform renders at the correct visual weight on macOS, so it no longer appears bold where it shouldn't.

The folder create and edit modal gains an optional Description field and a "Define success" section. When editing an existing folder that is not already a goal, you can attach a conversion event (searched from the server) and an optional end date. Saving with a conversion updates the folder's name, color, description, conversion, and end date in one step.

Folders created without a conversion use the standard create or edit flow with the new description field.

The Folders management page has moved from Settings to Business Goals > Folders. If you visit the old Settings > Folders path, you're redirected to the new location automatically, and the Settings sidebar link has been removed.

The Create and Edit folder modal includes two new additions: a Description field and a Define success section. In the Define success section, you can attach a conversion event to an existing folder — searched from the server — along with an optional end date. Saving with a conversion updates the folder's name, color, description, conversion, and end date in one step, and the folder's card and list view update accordingly. Folders created without a conversion use the standard create or edit flow with the new description field.

Run history redesigned: status chips, failure reasons, and a fix panel

The run history view in the flow builder gives you a clear picture of where every run stands and what to do when one stops short.

Status chips at a glance. Chips across the top — All, In progress, Completed, Timed out, Interrupted, Failed — each show a live count. Chips with no runs stay hidden unless you select them, so you're not wading through empty categories.

Failed runs, grouped by what you can do. Select the Failed chip and reason chips appear, grouping runs by their cause: Messaging window closed, Opted out, Can't be messaged, Sending limit reached, Connection needs attention, Step needs fixing, Contact not synced, Retry time ran out, or Something went wrong. Each run also shows a short reason line in the Status column so you can scan without clicking.

A fix panel that takes you straight to the problem. Selecting a run opens a panel beside the table showing the reason it stopped, the precise cause, and a direct link to act: Edit step opens the editor on the relevant step; Open Settings takes you to Settings → Channels for a connection that needs attention. The full path is shown, with unreached steps outlined.

Other changes in this view. A Duration column replaces the previous Steps completed column. The triggered, completed, and failed figures in the tab row are clickable and open the runs they count. Previous and next navigation works across page boundaries.

For flows with more than 200 runs, history is paged. Status chips show counts from a snapshot (counts are hidden while the snapshot loads), and reason chips are not shown — the newest 200 runs can't account for all failures across a long history.

The Edit with AI option is hidden on split-test send blocks in the Campaign builder, so you're no longer offered an action that can't complete on that block type.

The Campaign send modal includes an "Edit email" button that opens the mini editor inline — the same editor used by "Edit with AI" — so you can make last-minute changes without leaving the send flow.

In a multi-email Campaign, you can navigate between emails directly from the modal. If you have unsaved changes, email navigation is disabled and a Tool Tip explains why. The edit hover menu is hidden while the mini editor is open.

Custom connector setup shows its validation message on a single line, keeping the feedback in context so you can correct the issue without losing your place.

After creating a custom MCP connector, the Details tab shows the Server URL — the endpoint the connector points at — as plain text. Long addresses wrap. This field appears only for connectors you created that have a non-empty URL; curated connectors are unchanged.

When snoozing a conversation or a selection of conversations fails, the Conversations inbox displays a failure toast reading "We couldn't … Nothing has changed," so you always know when a snooze did not apply.

On the refreshed Contact detail page, the SMS and WhatsApp icons in the contact header are colored by channel: WhatsApp renders green and SMS renders pink. Channels that aren't available for a contact show no button. Tool Tips are unchanged. This is part of the ongoing Contact detail page refresh.

The contact limit banner — including all upgrade prompts — is visible to main admins on accounts resold through a partner that has a custom upgrade URL. Every prompt opens the partner's upgrade page. ActiveCampaign-specific upgrade options remain hidden for these accounts; all other account types are unchanged.

The activity filter on the Contact detail page has moved into a wider drawer, with a simplified detail view inside it. Empty and loading states within the drawer are centered. The activity feed in the side rail is sized to roughly half the viewport height, giving it a more proportional presence alongside other contact information.

Each tab in the tab strip includes a leading icon, making it easier to scan across tabs at a glance. Header channel icons render consistently.

On the Lists tab, channel headings (Email, SMS) are bold, and contacts with no list memberships show a pill-and-icon treatment rather than plain text.

These changes are part of the ongoing Contact detail page refresh.

Previously, opening multiple tabs could result in empty content areas with no error shown. Pages load reliably regardless of how many tabs are open at the same time.

On the refreshed contact page, the Actions menu in the contact header shows Archive only, matching the standard contact page. Previously, the refreshed page showed both Archive and Delete while the standard page showed only Archive.

The Campaigns page makes significantly fewer background requests while a send is in progress, so the page stays responsive without unnecessary network activity.

Status counts only refresh when the page's status set or total actually changes, rather than on every list fetch. Scheduled Campaigns begin polling two minutes before the earliest send time, then every five seconds. Campaigns pending review poll every 60 seconds. A sending row's progress polling stops once the send completes, then triggers one final list refresh. A 60-second list poll runs during a send so that a pause or stop from another session appears without a manual refresh.

The button is disabled until the campaign's message data has finished loading, then stays enabled through background refreshes — so clicking it always takes you where you expect.

Duplicate filters and misaligned controls on the Automations overview filter bar have been removed, leaving a clean, consistent set of filter options.

The Export button on the Automations Performance report opens a modal that starts the CSV export, polls its status, and offers the download link in place. The export includes your current filters and table sort order, and the button stays disabled while the modal is open. The CSV includes column headers for the Automations Performance report.

The Sales Performance CSV export can also be downloaded from the export status page, consistent with how other reports are handled.

Active Intelligence no longer allows editing Automation split Campaigns, and split Campaigns are hidden from the Automation emails preview. This prevents accidental edits to split Campaign configurations where the editor doesn't support them.

When adding an If/Else step in the Automation Canvas, a branch type picker appears before the condition editor. The picker offers If/Else and Multi-branch options. Choosing Multi-branch opens a criteria submenu with four options: field values, tags, list, and segment.

Classic If/Else continues to save without a criteria type, so existing Automations are unaffected. Editing an existing If/Else block skips the picker entirely.

A Campaign block in the Automation Canvas saves with the correct Campaign ID when field values are set in rapid succession.

Previously, batched state updates could cause the block to be saved with a Campaign ID of 0 — meaning the Campaign association was lost. All field writes complete and the form reflects the latest values before the save is submitted.

The Automation Canvas handles rapid node selection correctly. Previously, clicking an add-action node, closing the modal, and quickly clicking another add-action node before the close animation finished could cause the canvas to become unresponsive. That behavior is resolved.

The add-on management modal displays prices in the account's currency when the account uses a non-USD currency.

On the Appearance settings page, the Navigation Color option has been replaced by an Active Intelligence Theme picker with three options: Dark, Light, and Sync with system.

Picking a card applies the theme immediately. The choice saves automatically after two seconds — you don't need to click Save for the theme alone. Sync with system follows your operating system's color scheme live, switching between dark and light as your OS does.

The rest of ActiveCampaign stays in light mode regardless of which theme you pick.

New accounts default to Dark. Sync with system is opt-in. If your account was already set to Dark or Light, that setting carries over.

Active Intelligence responses that require structured output — including insights and analysis — are 30–60 seconds faster per call when the system is under load. Previously, high-demand periods caused calls to queue and stall. After this change, the backlog drops to near zero and responses arrive significantly sooner. The improvement applies automatically.

Active Intelligence picks the call-to-action button link in a Campaign from what it already knows about your account — your Brand Kit website, stored site pages, and non-social links — rather than leaving it blank or always asking you.

The preflight form asks about the link only when two options are equally strong. In that case, you'll see both candidates with their exact URLs, plus an "Other" option. Social profile links are never selected. Once a link is chosen, Active Intelligence tells you: "The main button links to {url}…"

Link priority order: a link you put in your prompt wins first, then a form or chat answer, then a pasted page, then Active Intelligence's own pick, then the service default.

Category labels on Active Intelligence home recommendation cards — Brand Kit, Automate, Email, and others — use consistent muted text in both light and dark themes, matching the intended visual hierarchy. Card titles are unchanged.

The summary is specific to the actual edits made and includes a verdict: done, partly done, or not done. A reason appears when relevant. Template edits state what changed immediately after confirming, ahead of any caveats. If a summary can't be generated, a short fallback sentence appears instead.

Three related accuracy improvements ship alongside this:

• Replacement phrases are no longer flagged as missing when you ask Active Intelligence to replace text.
• Blocks are labelled by what actually happened: "rewrote" when words changed, "updated" when only a link, image, or markup field changed, "restyled" when only presentation changed.
• The notice for unasked edits says an edit "changed" parts of the email rather than "moved" them.

This applies to direct edits. Confirm-gate edits will receive the same detailed summary in a follow-up release.

The Active Intelligence recommendation chip on the Segments index page moved into the shared prompt header row, matching the layout used on Campaigns. The chip's label, icon, and behavior are unchanged.

Active Intelligence no longer executes an older waiting approval plan when you say "yes" after the assistant has already replied to a different question in the meantime.

Previously, saying yes to an unrelated follow-up could inadvertently trigger an older pending action — for example, enrolling a Contact in an Automation you hadn't just confirmed. A turn-based gate now detects when the assistant spoke again after your first reply following the newest waiting card, clears the approval, and re-plans instead of executing the stale one.

Structured per-step card approvals are unaffected.

When Active Intelligence offers to bring back a finished or paused Scheduled Task, the task returns as a draft and does not run until you activate it. The offer and the confirmation reply both say so clearly.

If the task has a run limit that's already used up, or an end date that has passed, Active Intelligence tells you what to change before activating — including where to make each change (on the task page or in chat for the run limit; on the task page for the end date). A task past its end date is activated as a draft but won't fire until you update the end date.

When Active Intelligence brainstorms several ideas and one is a bring-back, selecting that specific idea correctly brings back that task.

Active Intelligence Automation reports and charts use a single, consistent label for each outcome throughout — in prose, tables, charts, and breakdowns. Previously, raw internal codes could appear alongside human-readable labels, causing the same outcome to be described differently in different parts of a report.

Turn the toggle on and the grid shows only direct leaf-level locations, regardless of where the Hierarchy Level filter is set. The two controls work together: the Hierarchy Level filter still applies, so you can narrow by depth and then further restrict to locations only.

The Hierarchy Level filter has also been broadened — it includes leaf accounts at every depth, and the default view returns your full account subtree rather than a shallow slice.

If you use the API to filter accounts: the filter parameter used to restrict results to direct locations has changed. The new parameter is stricter — it returns leaves only. There is no backward-compatibility fallback, so any integration using the previous parameter will need to be updated.

Opening several tabs at once no longer causes some of them to fail with an error. Previously, tabs that were slow to start — because they were queued behind other tabs loading at the same time — could miss a 20-second deadline and show a top-level error, even when the server itself responded quickly. Tabs that are queued or throttled load late rather than failing. A tab that is genuinely waiting on a hung request still gives up after 20 seconds.

Scheduled Tasks that fail with a retryable error after a read-only operation are retried as expected, rather than being skipped.

A task set to run at a specific local time — say, weekdays at 11:30 PM — stays at 11:30 PM local time in both winter and summer.

This applies to tasks scheduled daily, on weekdays, on chosen days of the week, or monthly. A schedule picks up the new behaviour the next time you edit it.

Editor design. The scheduled task editor heading, status chip, and layout match the design spec. The task's own name is the section heading. The status chip (Active, Paused, Draft) sits under the description. The green "This task is active" banner that previously appeared in the header is removed. Recent runs show a shorter date format, and the unread dot appears only on unread runs. When the Recent panel stacks under the form at narrow widths, its divider moves to the top with additional spacing above.

Navigation border. The left edge of the Scheduled Tasks page previously showed a double border. The icon rail skips its divider on Scheduled Tasks and task detail pages, matching the behavior already in place for the agent page. Top, right, and bottom edges are unchanged.

The Sales Performance report in Deals CRM is being rebuilt as a native page. The native page includes:

• Filters: pipeline, owner, date range, and currency
• Deal totals at a glance
• CSV export for the full report
• Deals by stage section with stacked bars per owner, count/value metric toggle, scaled and 100%-share bar scales, stage spotlight, a legend with deep links to the deals list, owner pagination, and an open/won/lost status switch that docks into the toolbar on scroll
• Paginated, sortable deals table for the selected status, with its own CSV export that follows the table sort

The Pages section displays a skeleton layout as soon as it begins rendering, so the page frame appears immediately rather than waiting for the full app to load. A spinner only appears when the editor itself is loading.

Deals CRM pipeline stage charts scale to screen width and show status tints

Wider screens display more stages per row; overflow stages collapse behind a shared toggle. Charts are tinted by deal status — won, lost, and open — using the same color frame as the deal overview. The deal overview status switch docks into the sticky toolbar when you scroll. A spotlight mode forces scaled bars for dense layouts.

In the conditional subject modal, the value input field renders at full width from the start, whether you're using a Tag Exists condition or a text condition such as Last Name / Equals.

Active Intelligence ranks the angles shown in your Campaign Variants carousel using your account's own click history. Tone, offer type, value proposition, and structure are all factored in, so the variant most likely to resonate with your audience appears in the first slot rather than a fixed default position.

Two related improvements ship alongside this:

• When you use Change look to generate a plain-build design, suggestions are informed by your account's industry vertical.
• Regenerate and Pick re-fetch your account context against the original brief, so occasion-based Campaigns (holiday sends, for example) resolve the correct date at send time rather than at draft time.

If your account doesn't yet have enough click history to drive the ranking, the carousel falls back to the standard angle order without interruption.

These improvements are available within the existing Campaign Variants feature.

When you open the add-to-lists modal for a contact who has no phone number or no email address, both the SMS and email sections are visible. The section for the unreachable channel shows a short explanation in place of list checkboxes, so it's clear why no lists appear there. The existing message for a reachable channel with no lists configured is unchanged.

The global navigation notifications panel and the Active Intelligence sidebar task history were failing for some accounts when the insights list response could not be parsed. The failure was concentrated on accounts with very large insights responses, where it occurred on about 1–3% of loads.

Both panels load reliably for those accounts.

What's better:

• Editing an automation's email from the Automation Canvas "Edit with AI" button reaches the automation correctly and applies the change in place. The reply includes the "This automation is active, so the change is live" notice when appropriate.
• When an automation email was built in an older editor, Active Intelligence names the email and offers to edit its subject line and preheader, rather than returning a generic refusal.
• New emails added to an existing automation are written for the business that automation already writes for — not the account's primary brand identity. Previously, adding an email to one business's automation could produce content written for a different brand entirely.
• When an automation email edit rebuilds the email's layout, the reply says so.
• When an image swap lands successfully, the reply says where the picture came from.
• When a send step is selected, only that step's email is edited. Previously, editing one of two same-named send steps could rewrite both.
• A subject-line change on a selected send step goes to that email in place, not to a full automation rebuild.
• When a new email's source layout couldn't be read, the reply says so clearly.
• When a source Campaign was sent by a different business, Active Intelligence uses only its design — none of its words, offer, or brand.
• When a multi-email edit changes some emails and then asks a question about another, the reply names all the emails that were changed.
• When a connection drops after a new send step is saved, Active Intelligence reads the automation back and either writes the email or marks the step failed with a clear message, rather than leaving the step loading indefinitely.

The AMP Form, AMP Accordion, and Carousel blocks have been removed from the email builder's block panel. Existing templates that already contain these blocks are not modified.

A "Show only Locations" toggle in the ACHQ Accounts grid filters the grid to direct leaf-level locations in the hierarchy. The Hierarchy Level filter has been updated to include leaf accounts at every depth, and the default view returns the full subtree.

The "Back to headquarters" button in the partner portal ends the active impersonation session when clicked.

Twelve API reference examples that previously displayed as raw broken text in the developer portal are corrected. These examples span accounts, deals, Campaigns, ecommerce orders, event tracking, and webhooks. Each one works with the interactive "Try it" console. A validation check is also in place to catch any invalid examples before they reach the portal in future.

Twelve API reference examples that contained invalid JSON are corrected. In the developer portal, these examples previously displayed as raw broken text and couldn't be used with the interactive "Try it" console. Fixed examples span accounts, deals, Campaigns, ecommerce orders, event tracking, and webhooks.

Spelling and wording across examples and descriptions — including field labels, query parameter names, and example account names — are also corrected. A validation check is in place to catch any invalid examples before they reach the portal in future.

The Settings › Webhooks drawer previously showed recent calls across all endpoints, with filtering happening in the browser. It now loads calls for exactly the endpoint you've selected, so what you see is always relevant to the webhook you're looking at. The "none of the latest 200" wording is gone because it no longer applies.

Webhook actions in Automations send the custom headers you configured, so every webhook call goes out with the authentication or routing headers your receiving system expects.

Parameters are URL-encoded before being appended to the request URL. An empty resolved value is still sent.

The "email marketing by ActiveCampaign" badge on unsubscribe and already-unsubscribed pages can no longer be forced on via a URL parameter for reseller child accounts or accounts with branding turned off. For directly branded accounts the badge continues to follow the form's own branding toggle as before.

Unsubscribe links in double opt-in confirmation emails — both the link in the email body and the one-click List-Unsubscribe header — unsubscribe the contact as expected. This applies to double opt-in emails already sent before this change.

One edge case: if a contact later subscribes to the same list through a different form, the original double opt-in email's unsubscribe link may not work as expected.

When your account has both a standard approved registration (10DLC, UK, AU, or DK) and a toll-free registration (approved or pending), a sender type choice — default sender or toll-free — appears under the sender name field in the Campaign builder.

• While a toll-free registration is pending, helper text explains that sends start after approval and the number stays hidden. Once approved, the number is shown.
• Sending, scheduling, and test-send are gated on the sender the Campaign uses: toll-free requires an approved toll-free registration; the default sender requires an approved standard registration. Save and save-and-exit remain available at any time.
• Toll-free reaches US and Canada only. If your account has another approved country registration plus toll-free, they're treated as two distinct senders.
• The "Not registered" banner is hidden for accounts with an approved toll-free registration.

Tags containing emoji or unusual character sequences sync correctly during ecommerce tag sync.

The "All campaigns performance" report returns a stable subject line for split Campaigns. The report shows the winning variant's subject, or the subject of the lowest message ID when there is no winner. Previously, the displayed subject could change between views depending on which variant the database happened to read first.

Every Engage route shows a loading stencil that matches the shape of its loaded layout, replacing the generic centred spinner or bare grid that previously caused layout shift when data arrived.

• Tables (Flows, Re-engage, Runs, Enrollments, Campaigns, Audience, Posts, Links, Surveys, Knowledge, Intents, Activity, Insights, Members, All channels) keep their toolbar and column names while loading, with stencilled rows capped at 25.
• Editors (flow builder, sequence editor, campaign editor/report) keep the breadcrumb, tabs, and rail heads while stencilling the board, phone, or report area.
• Inbox no longer flashes "No conversations yet" while loading; Channels no longer shows "0 connected" and "Pending review" before data arrives.
• Agent pages show real section titles over stencilled rows; Approvals stencils the queue and request.
• Settings: the nav and view head render immediately while each view stencils its body.

The segment contacts tab and trial summary display accurate contact counts for accounts using Portuguese (Brazil) and British English locales. Those errors no longer occur.

Contact matching reads contacts in batches of 1,000, so memory use stays bounded regardless of list size. Contacts added during a run are picked up, and existing entry-limit checks prevent duplicate enrollments.

Date range filtering is available across the key metrics on the SMS broadcast reports page, so the numbers you see in each view reflect the same window.

Failures widget: the failures widget uses the same date range selected on the page, so its numbers match the KPI metrics shown alongside it. The "view more" link carries the range through.

KPI card links: clicking a KPI card to drill into recipients carries the selected date range to the recipients page, so the detail view shows the same window as the number you clicked.

Recipients page date picker: a date range picker appears on the broadcast recipients page, and the selected range is preserved through breadcrumb navigation and the Back button to the details page.

Sending, scheduling, and test sends remain gated on the sender the Campaign actually uses: a toll-free Campaign requires an approved toll-free registration, and any other Campaign requires an approved registration. An account with only approved toll-free cannot send on the default sender.

A broad set of surfaces that were incorrectly showing ActiveCampaign branding to reseller children and accounts with branding turned off have been corrected.

Campaign footers and error notifications. Conditional-content error mail sender, scheduled-Campaign error mail sender display name, Campaign editor warnings, and the personalization empty state no longer carry ActiveCampaign branding for reseller accounts. Free-plan reseller accounts no longer see ActiveCampaign attribution in default Campaign footers: the reseller logo is shown unlinked with no alt text, and the "Marketing provided by: ActiveCampaign" line is removed from the plain-text version. Direct branded accounts and paid reseller accounts are unaffected.

Apps marketplace nav entry. The global nav's Apps marketplace entry is hidden for reseller children and accounts with links turned off.

Loading overlay. The classic designer's dual-preview pages no longer show the ActiveCampaign logo for reseller children and accounts with links turned off.

Admin header glyph. Direct accounts with branding on see the ActiveCampaign glyph as before; reseller children and accounts with links turned off see the neutral home glyph.

Shopify embed-status banner copy and Active Intelligence deep-link prompts. When branding is off, banners read "the form app embed" instead of "the ActiveCampaign form app embed," and deep-link prompts say "the platform" instead of "ActiveCampaign."

Clicking a column header or applying a filter in the All Campaigns Performance report or any Single Campaign report works without delay. All 19 report tables are affected.

For resold accounts whose partner has a custom upgrade URL, the Upgrade button, the Upgrade option in the account dropdown, and the Billing option in the account dropdown all open in a new tab. You stay on the page you were on in ActiveCampaign. Accounts without a partner upgrade URL are unaffected.

If your account is resold through a partner, the main admin sees the contact limit banner when approaching the contact limit. The banner includes a single "Increase contact limit" button that opens the partner's upgrade page. ActiveCampaign-specific upgrade options remain hidden for partner-resold accounts; all other account types are unchanged.

Main admins on partner-resold accounts that have a custom upgrade URL see the full set of contact limit prompts: the limit warning modal, the limit banner, the Contacts header Upgrade button, the add-contact call to action, the import limit modal, the CSV import modal banner, and the Campaign "past limit" banner.

Every upgrade call to action opens the partner's upgrade page in a new tab. The redirect applies only when the configured address is a valid HTTPS URL; all other cases behave as before. ActiveCampaign-specific elements — the auto-upgrade toggle, the AC help link, and the in-app purchase modal — remain hidden for these accounts. All other account types are unchanged.

Campaign initialization checks the size of each message before sending. If any message's HTML or text exceeds the configured limit, the Campaign returns to draft rather than being sent. Automations that trigger a send keep their status unchanged.

Gmail clips HTML over approximately 102 KB, and very large messages caused disk issues on mail servers. Previously, the sending infrastructure would refuse an oversized message on one route and fall back to another route that would send it anyway.

The size limit is currently off in production, so no Campaigns are blocked until the limit is activated. Instant autoresponders, send-last-broadcast, and immediate Automation sends are not yet covered by this check.

Previously, these notifications were silently dropped whenever a deal's pipeline name, stage name, or a task's type was absent. Where a pipeline, stage, or task-type name is genuinely missing, the notification still goes out — that field simply appears blank.

The edit_automation tool available through the API integration was previously marked as non-destructive, but it overwrites an existing Automation in place with no copy made. API integration clients now receive the correct signal for this tool, matching the behavior already flagged correctly for Campaign editing, so they can apply appropriate confirmation or safeguards before allowing the operation.

This applies to Forms in EU, AP, and legacy US regions. The contact is created and the confirmation email is sent as expected when the double opt-in message format is set to HTML.

Opt-in settings — including the opt-in list ID and send-opt-in flag — are preserved when options are missing from the request. Saves that omit action data no longer delete opt-in Campaign list links.

When a page loads with no valid session, the platform redirects to login immediately, before loading the application shell or firing any API requests. Previously, the app would render its shell and fire approximately 11 requests before the redirect happened.

The Sales Performance report can be exported to a CSV file, giving you a portable record of your deals data to share, archive, or analyze outside the platform.

The export applies the same filters and sort order visible in the report — date range, currency, deal status, pipeline, and owner — and includes every matching deal with no row limit. The file opens with a summary of deal totals (created, won, lost, and open counts, total values, and averages), followed by a row-level table with Pipeline, Stage, Deal Title, Deal Owner, Deal Value, Created date, and Currency. Dates appear in your account's date format and your time zone. Deals owned by a deleted user appear as "Unknown owner."

The export runs in the background; a separate download link lets you retrieve the finished file when it's ready.

Previously, the shell, navigation, and route components would mount and fire approximately 11 requests before the expired session was detected, resulting in a series of failed requests before the redirect occurred.

The Enrich Contact action in the Automation Canvas gives you control over which profile fields get updated and how. You can select exactly which fields to enrich, and choose whether to fill only empty fields or overwrite existing values entirely.

When you add a new Enrich Contact block, every field is selected by default with "only fill empty fields" set. You can narrow the selection to just the fields you want. Saving with no fields selected is blocked — at least one field must be chosen.

At runtime, your Automation writes only the fields you selected. "Fill empty" leaves existing values alone; "overwrite" replaces selected values when enrichment data is available, and leaves the current value in place when it isn't. Before any phone number is written, the platform checks that no other contact already has it. Account matching works by exact company name: a match relinks the contact to that account without editing the account record; no match creates a new account; if no company data is returned, the current account link stays and only the job title updates.

Email and language fields are excluded from writes in this release.

The endpoint supports limit (1–100, default 20), offset, search (list name contains), and sort (name or -name). Each list in the response includes its id, name, active contact count, and engagement rate. When no tracked Campaign has been sent to a list, engagementRate returns null. Access is restricted to admin-group users.

If you're using the legacy v1 engagement endpoints, plan to migrate to this endpoint.

The Engagement Management page has clearer navigation and more context about what the process does. A breadcrumb ("Lists / Engagement management") links back to the lists page. An informational banner explains that Contacts are unsubscribed from the selected list and tagged for later review, and notes that Apple Mail Privacy Protection opens are not counted as engagement. A "Learn more" button in the header opens the public help article in a new tab.

The page subtitle reads: "Find lists with low engagement and unsubscribe contacts who aren't opening or clicking."

When Engage is in account maintenance, pages previously gave no clear indication of what was happening — you'd see a generic error with no guidance. Engage detects when the service is temporarily unavailable and displays an "Engage is briefly unavailable" message with a Retry button, so you know what's happening and can try again when it's ready.

Images in open threads stay in place when the inbox refreshes in the background. Voice notes and video clips continue playing from where they are when their URLs rotate — they don't restart from the beginning.

If a media URL expires mid-play, the player picks up the refreshed URL, restores the playhead position, and resumes if you were playing. The inbox refreshes every 20 seconds in the background; those refreshes no longer disturb anything you're watching or listening to.

When you open Manage Contacts from a wait block's queue in the Automation Canvas, a new "End wait" action appears in the row menu and the selection toolbar. Selecting contacts and choosing End wait moves them past that wait step without ending their Automation run.

Each Contact's wait is skipped individually, so if a Contact has already moved past the wait block, that one is reported as failed while the rest continue processing. Requests with a missing Contact ID return a validation error and skip no one.

This action only appears when Manage Contacts is opened from a wait block queue — not from the global header.

Opening an email entry in the deal activity drawer shows the complete email — subject, From/To/Cc/Bcc addresses, attachments with file sizes, and the email body — for sent and received emails. Opened email entries show the subject and body. Link Clicked entries show the URL as a live link. This matches what was previously available only in the legacy view.

Four dropdowns that were opening behind modal overlays — including Engage table select fields inside Engage modals, a tier dropdown, two dropdowns in the tune-insights modal, and a dropdown in the automate modal — appear correctly on top of the modal.

When a stage contains deals, a dialog asks what to do with them before the stage is removed:

• Delete all deals — requires confirming three checkboxes before proceeding.
• Move all deals to any stage in any pipeline — available when you have the deal-edit permission.

Empty stages are deleted immediately as before. The delete option requires the deal-delete permission; without it, only the move option is offered.

The stage is hidden from the pipeline view while the action runs and reappears if it fails, so it can be retried.

URL-encoded date separators in due-date range filters are handled correctly. Malformed date ranges return a clear error message.

Deals previously stranded by an earlier stage deletion can be updated through the API again.

The bulk move on stage deletion does not fire per-deal events, so no webhooks, Automations, activity records, or update timestamps are triggered for the moved deals.

Deals CRM overview report loads faster

Load times on large datasets dropped from as long as 18.5 seconds to under 4 seconds across all report sections — totals, won-by-owner, value brackets, and deals over time. The report also supports async CSV export.

Filters include date range, currency, pipeline, owner, and status. Dates reflect your account's time zone.

A few deliberate differences from the previous version: won-ratio shows 100% for an owner with no lost deals (previously showed 0%); deals over time shows team and team average per user rather than one line per user; a deal is counted in the month of its last status change; open deals have no 24-month cap.

The delete-stage dialog shows an error banner and hides the Delete option when the pipelines query fails.

Lookup custom fields appear read-only — the same as win probability — and each field label carries an info icon Tool Tip naming the object the value is loaded from: the primary contact or the account.

This change also fixes a situation where the entire Overview tab became unavailable whenever a single lookup field was present. The tab loads correctly regardless of which custom fields are on the deal.

Win Probability scores continue to record their own timestamp separately. This applies to all accounts.

The first entry in a deal's activity log reads "Deal created" with an add icon, in both the activity row and the drawer.

A table in SMS settings lists your approved SMS registrations — covering US, Australia, Denmark, UK, and toll-free numbers — with a status of either Customized (reply saved) or Not customized (default reply). Clicking a row opens an editor where you can set the sender name from your address book and enter a support phone number. For Denmark registrations, a language switch lets you choose between Danish (default) and English.

Changes take effect immediately after saving — contacts who text HELP receive the updated reply right away.

The Custom Objects records list enforces that relationship access filters accept exactly one plain value or one [eq] value per request. All other forms return a clear error. Previously, unsupported forms returned an ambiguous success response for admin users or a server error for non-admin users, making it hard to tell what went wrong.

When you copy, share, or save an Automation as a template, the copy includes only the source-language email. Translated copies, the send block translations setting, and translation run records stay on the original Automation and are not carried into the new one.

Shares and templates that were already exported before this change are unaffected.

Contacts held at a Salesforce step in an Automation are released automatically after 24 hours. Any contact waiting longer than the timeout moves past the block, and their Automation history shows "Salesforce did not respond within Nh" so you can see exactly what happened.

Contacts in paused or inactive Automations are left as-is. Only the specific step that timed out is marked — nothing else about the contact's Salesforce record or other running Automations is changed. If Salesforce responds before the timeout fires, the contact continues normally.

The default timeout is 24 hours.

The API reference for POST /contacts and PUT /contacts/{id} responses now uses the same shared Contact schema already used by GET /contacts/{id}, so the definition is consistent across all three endpoints.

Two response behaviours are now explicitly documented:

• PUT response: The response reflects the stored record with your sent values applied, but is not read back after the save. Fields the database sets automatically — such as hash and email_domain after an email change — show their value from before the update. Custom field, tag, and list changes are saved after the contact record itself, so the returned udate can be earlier than the final stored value.
• POST response: The response includes the fields you sent plus id, cdate, udate, links, orgid, and organization. Database-set fields such as hash are absent from this response.

This is a documentation change only. No API behaviour changed.

An info icon appears beside each field label. Hovering or focusing it shows a tooltip: "This value is loaded from the account" for account-linked fields, or "This value is loaded from a related record" for other configurations.

These fields appear once the supporting changes are live and the feature is enabled for your account.

When part of a contact import's source data has been deleted from storage before processing completes, that portion is marked as failed rather than retried indefinitely. Only the affected portion fails; other parts of the same import continue normally.

The "Edit all" action on a Contacts view filtered by a specific list of contacts sends the correct contact IDs to the bulk edit action, so only the Contacts you can see are affected.

Previously, the IDs were passed incorrectly, which caused the bulk edit to apply to all contacts in the account instead of only the intended ones.

The per-run "View" action in the Contact Automation detail drawer is a button consistent with the rest of the interface. The automation link and status colors are updated, and spacing between the End and View buttons in each run row is corrected.

Feedback loop complaint counts are recorded for Campaigns regardless of whether the Campaign has a missing, empty, or unreadable send date. Previously, Campaigns in that state could silently drop complaint counts, meaning abuse metrics for those Campaigns were understated. Complaint counts for all affected Campaigns are recorded correctly going forward.

Change deal owner directly from the deal highlights card

The Highlights card at the top of the deal detail page uses the same click-to-change pattern as the Stage card. Click the owner name, search by name, select someone new, and the name updates immediately in both the Highlights card and the deal overview. The change is recorded as a "Deal owner changed" activity entry.

If you don't have permission to reassign deals, the picker is visible but read-only.

A conditional written with an unspaced < — for example, comparing a date field to a specific date — could cause the text version of the Campaign to be built incorrectly, preventing the send from completing. That issue is resolved.

Campaigns created by picking an AI variant get an "Initial" entry in the email editor's version history, so reset-to-original has a point to return to. Other creation paths already recorded this entry; the variant-pick path had been missed.

Campaigns created before this change remain without an initial version point and are not backfilled.

The Campaign summary and new Campaign creation page have a refreshed visual design. Both pages share a light background, consistent heading sizes, and a unified selection treatment.

Key changes include pill-button suggestion chips (report, edit, send) with Edit with AI as a secondary button; full-width toggle rows for tracking; updated dropdowns for list and address pickers; schedule cards with a focus outline and no pre-selection on page open; and single-line inputs for From name, From email, and Reply-to with a solid focus border.

On the new Campaign page, the Campaign name and campaign type headings are the same size and left-aligned, and type cards use the same outline selection as schedule cards. A translated Campaign explains why the Subject field is disabled. List search looks through lists already loaded, 100 at a time.

An internal metrics timing error can no longer interrupt or fail a Campaign send.

Previously, a timing error that ran alongside the actual send request could either stop an already-accepted send mid-way — sends could halt after approximately 10,000 recipients — or replace the real send error with a misleading one, hiding the true cause. Timing errors are logged as warnings and the send continues.

Campaign summary warnings and the email builder product block no longer show ActiveCampaign help links, logos, or product copy for reseller child accounts or accounts with branding turned off. Directly branded accounts see exactly what they saw before.

Goal folders show three buttons: Conversion Performance, Settings, and Add asset. Non-goal folders show Settings and Add asset. Conversion Performance and goal Settings both navigate to the goal detail page; non-goal Settings opens the Edit folder modal. The header wraps correctly on narrow screens, and any search filter you've applied survives navigating back.

When a Brand Kit's font block contained only fonts that can't be used in the email renderer, the kit was still influencing campaign generation. A Brand Kit whose fonts are all unusable is no longer treated as a style anchor. Color blocks were already handled correctly; fonts follow the same logic.

Brand Kit monitors your website and updates six fields automatically when changes are detected: primary color, accent colors, light and dark background colors, and title and body fonts. Logos are not refreshed.

Refresh runs on scheduled brand-monitoring passes — when a change is detected on your site, when the check can't complete, or when your site hasn't been scanned in 7 or more days. Changes are written directly; there is no approval step.

When setting up a date-based trigger, you can select a custom object date field and a fixed time zone. The Automation history shows the custom object field's name in the trigger description. Contacts matched by the custom object date condition flow through the existing Automation path — same-day deduplication, the Automation's own segment filter, and series start.

One limit to know: multi-entry "every year" matching is not supported for custom object date triggers.

The "Send a webhook" action in Automations can send its request body as JSON or as form-urlencoded fields. When you edit and save a webhook action with an output type selected, it posts in that format. A webhook action that has only a URL saved — with no output type — continues to send a form body, so existing webhooks keep working as before.

A warning appears before you switch a webhook from form-encoded to JSON format, giving you a chance to confirm the change before it takes effect.

The Automation runs view previously loaded your entire run history into the browser at once, which became slow for accounts with long histories.

The view reads only the newest 200 runs. For accounts with 200 or fewer runs, filtering, search (by contact, status, trigger, and reason), and sorting continue to work exactly as before. For longer histories, the view pages ten runs at a time, filters by status, searches contacts, and orders by start time newest-first.

A few additional details: an open run stays visible after paging away; a filtered page that shrinks past its end steps back to the last available page; run counts stay blank rather than showing zero while loading; and the empty-state message for a long-history search clarifies that it covers contacts.

When a paused Automation returns to active, contacts that were queued but excluded from the resumed run show a clear reason in their history. Previously, they showed the same status as contacts excluded when an Automation is turned off entirely, making the two situations impossible to tell apart.

Contacts excluded at reactivation show one of two distinct reasons: one for contacts who already had another active run in the Automation, and one for contacts who had been queued more than once and had the extra run discarded. Turning a paused Automation off still uses the existing status.

GET /api/3/automations accepts two new optional query parameters — entryDateStart and entryDateEnd (YYYY-MM-DD format) — to return entry counts for a specific date range.

When both parameters are provided: - Only Automations with at least one entry in the window are returned. - Each row includes entriesInWindow. - meta.entryWindowApplied is true and meta.total reflects the match count.

An empty window returns entryWindowApplied true, total 0, and an empty list. If the query times out, the full unfiltered list is returned with entryWindowApplied false and no entriesInWindow. Omitting both parameters leaves existing behavior unchanged.

When you edit and save an Automation email outside the Active Intelligence editor, the translation updates automatically to reflect the new content.

Removing a block from an Automation no longer leaves the remaining blocks with duplicate or missing position numbers. The issue could affect both the standard Automation Canvas and the AI-assisted automation editor. Automations continue to run correctly after a block is deleted.

Previously, archiving contacts via Automation left their list subscription records in place, meaning archived contacts continued to count toward your plan. The fix ensures the archive action works as expected end-to-end.

The report shows one row per automated 1:1 email with columns for Automation (linked), Automated 1:1 Email name, Emails sent, Unique opens, Unique clicks, Unique replies, and Bounces — each metric showing both rate and count. The table is server-sorted and paginated. Clicking "Emails sent" opens a contacts drill-down showing who received the email with the same metrics, linking each contact to their contact page.

Filters carry over from the previous experience: automation, sender, 1:1 email, label, email send date, and event date. CSV export is available on the report.

Only the icons actually used by admin pages are included. On the classic designer, the page HTML drops from 1.84 MB to 865 KB.

The popover pages through available assets until enough rows are available to fill the list. It shows correct empty states, an inline retry alert when some asset types fail to load, and toast copy that correctly reads as adding an asset to a folder. The popover opens directly from the folder contents "Add assets" button.

When Active Intelligence builds an Automation and a requested trigger isn't supported or available, the reply correctly identifies what was built instead.

Active Intelligence distinguishes two cases: a trigger that isn't a recognized Automation start at all (for example, a segment-based start), and a trigger that exists but can't be built for your account (for example, a removed tag trigger). Each case produces a different label in the reply, so the explanation matches what actually happened.

The Active Intelligence homepage "Worth a look this week" heading displays without a sparkle icon beside it.

When you ask Active Intelligence to add an AI-generated image to a Campaign draft, it responds with a clear explanation that AI image generation is not available and offers alternatives — using the content manager or attaching an image you provide. Requests to place an image you supply are unaffected.

When Active Intelligence queries deals that span more than one page of results, it counts and totals every matching deal rather than stopping at the first page. Two confirmed symptoms are resolved: 191 matching deals were reported as 147 with one stage reading 40 against 32 returned rows; and 130 seeded deals across three stages (43/44/43, $13,000) were reported as 44/44/42 due to page-ordering inconsistency.

Active Intelligence reads up to 1,000 matching deals. When the full set exceeds 1,000 deals or a page fails, the reply includes a disclosure: "Counts and totals cover 1000 of the N matching deals." When the displayed list is shorter than the full matching set, it states "The list shows K of N matching deals."

What Active Intelligence shows during a live conversation matches what you see after a reload, and stopped or checkpointed turns are stored correctly.

Error state consistent after reload. A tool step's error marker stays the same whether you're watching live or returning after a reload.

Files uploaded mid-turn available in later turns. When a turn was stopped by you or ended at the request checkpoint, files, chips, and slash-command markers are stored the same way a completed turn is, so they remain available in later turns.

Conversation name preserved at checkpoint. A turn that hit the request limit previously saved the conversation as "New Conversation." The conversation name is stored correctly at the checkpoint.

Image list shows accurate count. When Active Intelligence lists images from your content library and the library holds more than the page shown, the note correctly states how many are shown versus how many exist — for example, "Showing the latest 12 of 50 images."

Error messages show plain language. Errors from the content manager or Scheduled Tasks tools show plain-language messages.

Active Intelligence's narration — for example, "Let me look up your brand kit..." — appears as a transient status line on the working card rather than mixing into the reply text. It updates in place as the task progresses and clears automatically when the answer arrives, a tool call completes, or the run ends. Conversations loaded from history are unaffected.

Active Intelligence can supply the account's industry from its own context when onboarding has no industry on record. Values such as "Unknown," "N/A," "None," and similar variants are treated as no industry set, which may adjust the evidence tier for those accounts.

A new GET /api/3/lists/engagement/interactionRate endpoint returns the account-wide interaction rate: unique opens divided by recipients, averaged across the last 10 tracked, sent, non-text Campaigns. The response is { "interactionRate": 16.67 }, or null when no qualifying data exists. Main admins see every list; other users see only Campaigns sent to lists they can access.

What this means in practice:

• The abandoned-cart block's edit (pencil) button is removed, so products, sort order, and fields can no longer be changed. Existing abandoned-cart blocks still render, send, and can be styled, moved, and deleted.
• The image editor is removed from the picture toolbar's Edit button (visible only to accounts with content manager off) and from the Edit link on image gallery thumbnails. Choose, Preview, and Delete remain. Accounts with content manager on are unaffected — Edit there opens the content manager's own editor, which is unchanged.

When a test send goes to a recipient who isn't a contact in your account, the "View a web version of this email" link resolves correctly. The same fix covers the forward-to-a-friend, web-copy-no-social, and update-profile link types.

The "Has Downloaded File" condition no longer appears in the conversion event segment builder (Settings → Conversions). It has also been removed from contact search, the segment builder, the Active Intelligence segment preview, and the Campaigns custom conditions modal.

The "Has Downloaded File" condition has been removed from the list segment builder. It no longer appears as an option when building segments in that area.

The timing issue that prevented first-time abandoners from matching product and category filters is resolved.

Clicking the recommendation navigates to the suggested destination and closes the drawer. Dismissing it removes it until the next day.

At widths below 560px, the inline-CTA upsell banner stacks — the title spans the full width and the call-to-action button sits on its own row below it. Wide layouts are unchanged.

The Tags field in the contact inbox sidebar shows the "Enter tag" placeholder and accepts input as expected. Previously the placeholder and input area were invisible, though typing still added a tag.

When creating a new intent in the flow builder or the AI Agent modal, up to five intent suggestions drawn from your account's recent direct messages appear as chips under "Suggested from your messages." Selecting a chip fills in the intent name, description, and examples. Suggestions are fetched at most once every 30 minutes, and intent names already in your library are excluded. Suggestions do not appear when editing an existing intent.

The Intents tab also tracks AI Agent usage alongside flows: the "Used in" column reads "AI Agent" or "2 flows and AI Agent" as appropriate, and deleting an intent warns that the AI Agent will stop answering messages that match it.

Split tests are blocked on Campaign send blocks that use AI translation. The declare-winner action is also blocked on those blocks.

When creating a Scheduled Task through Active Intelligence, you can set it to run "Daily on weekdays." Tasks set to run on specific chosen days display as "Every {days}" — for example, "every Monday and Friday." The new frequency labels appear in task creation and in the repeating-schedule summary lines.

Polling stops immediately on the first non-pending response — success, failure, or cancellation. If you close the tab or navigate away, the backend detects the disconnection within 5 seconds and cancels the in-flight run.

When an account has no Scheduled Tasks, the heading, filters, and sort controls remain visible and an illustrated empty state appears above the message "No scheduled tasks yet." Previously the entire section disappeared, making the feature hard to find. A filter that matches nothing shows the toolbar and an empty grid rather than the empty state.

Scheduled Tasks — Active Intelligence's autonomous monitoring feature — classify and handle outcomes correctly, so the right things reach your inbox and nothing important gets dropped silently.

Blocker and tool-error reports are now delivered. Previously, when a Scheduled Task hit a blocker or a tool error, the outcome was dropped without being emailed or recorded. These outcomes are now delivered correctly.

Outcomes are clearly distinguished. A normal report is emailed and recorded. An empty monitoring period produces a normal report with activity checks and next steps. A task that needs editing sends its message as the report body. Transient errors are retried without storing or emailing. Internal infrastructure problems are retried silently — they never reach your inbox. Runs that may have already written data are skipped rather than retried, preventing duplicate assets from being created.

Error messages stay clean. Errors show plain-language messages instead of internal exception details.

Scheduled Tasks cards have been visually refreshed to make actions easier to find and understand at a glance.

• Action menu. The overflow menu on each card uses standard menu items with consistent icons, typography, spacing, and dark-mode colors. Active tasks show Edit, Pause, and Delete; draft or paused tasks show Activate instead of Pause.
• Delete is styled as a destructive action. The Delete option appears in red — label, icon, and hover background — so it reads as a dangerous action distinct from Activate, Edit, and Pause. The styling works correctly in both light and dark themes.
• Run date is visually de-emphasized. The run date line and pause-reason line are set in a smaller, muted style so the most important information stands out.
• Tighter, consistent padding. Cards are padded 16px on every side, and the gap between cards in the grid is reduced from 20px to 16px both horizontally and vertically.

Run and start times on Scheduled Tasks cards pin to the bottom of each card, so timestamps stay visually aligned even when neighboring card titles wrap to different line counts.

Hovering over a task card reveals the actions button and shifts the card background to a subtle highlight. The button also appears on keyboard focus and stays visible while its menu is open. Touch devices continue to show the button. Completed task cards keep their View button unchanged.

The overflow menu uses the standard dropdown menu component. Edit, Pause, and Delete actions are presented as proper menu items with consistent icons, typography, spacing, and dark-mode colors. Active tasks show Edit, Pause, and Delete; draft or paused tasks show Activate instead of Pause.

The Delete option uses red destructive styling — red label text, red trash icon, and a red hover background — so it reads as a dangerous action next to the other options.

When saving a custom audience fails due to a network error, timeout, or unexpected response, a general error message appears under the name field. The duplicate-name error message continues to appear when a name is already in use.

Selecting a different list in the SMS broadcast reports filter previously caused the KPI snapshot widgets to disappear entirely until the new data loaded. The widgets remain on screen and display a loading indicator while the updated numbers are fetched.

The Credits Consumed tile and the Credits Consumed column in the SMS Automations report display fractional values — "12.4" rather than "12", and "200" rather than "200.0".

The tile shows up to two decimal places with no trailing zeros. Totals up to 999,999 display in full; fractional totals of 1,000 or more are abbreviated (for example, 1,000.5 shows as "1K" with the full value in a tooltip when you hover).

The Automations table column is locale-aware: decimal and thousands separators match your account's locale — "1,604.3" in English, "1.604,5" in German.

The SMS recipients page — for both Automation and broadcast sends — stays stable when you clear a search or navigate back to it after previously viewing it. Previously, either action could cause the page to fail with an error.

On the Engagement Management page, a new modal lets you remove contacts who haven't engaged within a time window you choose — 3, 6, 12, or 24 months, with 6 months as the default. Open it from the row action on any list.

Before removal runs, you must tag the contacts being removed — the tag field is required and supports autocomplete over your existing tags. If you have tag-management permission, you can create a new tag on the spot; otherwise you can only select an existing one. A confirmation checkbox names the contacts to be removed and resets if you change the contacts option or time period. The submit button stays disabled until a tag is chosen and the box is ticked.

A "Remove connector" option is available on the detail view for custom connectors you have added to Active Intelligence. Selecting it asks for confirmation, then removes the connector and its connection. Built-in connectors do not show this option. If the removal fails, a notification appears and you stay on the detail view.

A banner on the Campaigns list can prompt you to have Active Intelligence draft your recurring campaigns, with prompts tailored to daily, weekly, monthly, and quarterly cadences and worded to match your campaign name.

Three visual issues in the QR code popover on a link's panel are fixed.

The file type menu (PNG, JPEG, WEBP, SVG) opens fully visible above the popover. The Download button's icon is white to match its label. The file type select and Download button are the same size — two equal columns at the same field height. The popover is wider to accommodate the Download label in longer locales, and the QR code is larger to match. The Copy button's icon beside it also follows its label color.

Level 0 is the top-level anchor account. When the backend does not report a level, the chip displays as before. This makes it straightforward to tell which account you're acting on when navigating between accounts during impersonation.

On the Opt-In Information step of the US/Canada toll-free number registration, selecting "On a paper form" as the opt-in type shows a file picker so you can upload the form directly. All other opt-in types continue to use a text URL input.

Previously, the navigation stayed fully interactive while a Pages modal was open.

When you review a migration from a previous contact database, the migration review screen and landing preview surface CRM objects including companies, deal pipelines, deals, and notes. Member notes and notes on companies are included in the review display. The "Couldn't bring over" section narrows to tasks for applicable migrations.

When importing contacts and a list limit prevents some lists from being created, the import keeps going rather than stopping. Once it completes, you'll see a warning that tells you how many lists were created versus attempted, which lists were refused, and how many contacts were affected.

Those warnings also appear in the "Needs your attention" section — on the review screen after import, and as banners on import-complete cards on the homepage, post-opportunity flow, and Contacts empty state. Warning text is formatted for your locale.

One thing to be aware of: the homepage import-complete card still shows "All assets are imported successfully!" even when a warning is present.

The list subscriptions picker in the Forms builder responds correctly when you hover an option and click before the dropdown finishes rendering. Previously, that sequence could cause the picker to enter a continuous re-render loop, leaving it unresponsive. The picker applies your selection once and stops re-rendering.

A toolbar on the Business Goals Folders page gives you three filter controls: a Folder filter that pins the asset grid to a selected folder, a Type filter scoped to Campaign, SMS campaign, Automation, and Form, and a Search filter that finds assets by name across types.

Filter state persists in the URL, so you can share or bookmark a filtered view. The grid's clear-search button resets all three filters. Archived folders are hidden from the Folder dropdown.

This change is behind a feature flag and is not yet generally available.

Each recorded voice note goes out as its own message. Image attachments have all ten available slots — none are held back.

The voice microphone is visible on Instagram and Messenger, and hidden on WhatsApp and Telegram, where recorded audio in this format isn't accepted.

In the segment Contacts table, the dash shown for a contact with a missing name, email, or phone is visible in dark mode. Previously it was invisible against the row background.

When previewing a Campaign that contains only images, the email client tab list — Gmail, Yahoo, Apple Mail, iPhone, iPad — stays visible regardless of the preview content.

The Delete button stays locked until the save completes. Opening a different note mid-save does not unlock the note still saving. Other notes' Delete buttons remain available.

This change removes the need to adjust filters after searching. Any deal you've ever created is now reachable directly from the header search bar, regardless of its current status.

On the deal details page, a clear (×) button appears next to the assigned account when the field is editable. Clicking it removes the account association immediately, without opening the editor. The page reflects the change right away.

Three improvements to note handling on the Deals CRM deal details page make it harder to lose work.

Note drafts are preserved. Unsaved note text is autosaved as a draft as you type. Drafts appear in the Notes tab with a "(Draft)" label. Saving a draft publishes it; cancelling a new note deletes its draft, while cancelling an existing draft keeps it. Drafts count toward the Notes tab total, and other users' drafts are visible.

Delete is disabled while a note is being edited or saved. The Delete button on a note is locked while that note is open in the editor, and stays locked until its save completes. Opening a different note mid-save does not unlock the note still saving. Other notes' Delete buttons remain available.

Radio custom fields display correctly. Radio-type custom fields on the deal details page render as radio buttons, matching the field type. Selecting an option autosaves after the usual pause; clicking outside or pressing Enter saves and closes; Escape closes without saving.

Deal description saves on blur, not on every typing pause. The Description field saves only when it loses focus. Previously it autosaved every 800 ms while you typed, logging a "Deal Description changed" activity entry for each pause. Other text fields keep the 800 ms autosave. A pending description edit still saves if you navigate away.

Stage changes show a loading state. Changing a deal's stage shows the previous stage and time with a loading indicator until the new stage and time are both ready. If the save fails, the indicator clears and the old stage is restored.

On the Deal Details page, the Activity tab's "Load more" button stays hidden when the selected activity type filter — Notes, Emails, Tasks, or Deal Updates — matches nothing in the loaded history. When you pick a filter with no matches on already-loaded pages, the tab pages through history automatically until it finds a match or reaches the end, then shows "No results" if nothing matches. "No results" and "Load more" are never shown at the same time.

The date-based Automation trigger can fire on account-level dates, including account custom date and datetime fields and standard account created and updated timestamps. Custom object date fields are also supported.

When setting up a date-based trigger, you can select a custom object date field along with a fixed time zone. The Automation history shows the custom object field's name in the trigger description. Contacts matched by the custom object date condition flow through the existing Automation path.

One caveat applies to custom object date triggers: the contact time zone option is not available when a custom object field is selected. A saved contact time zone or an empty zone is replaced with the browser zone (or UTC).

This capability appears in the trigger picker only when the feature flag is enabled for your account.

Three improvements shipped together on the SMS broadcast reports page:

• Failures widget: the failures widget on the broadcast details page uses the same date range selected on the page, so its numbers match the KPI metrics shown alongside it. The "view more" link carries the range through as well.
• KPI card links: clicking a KPI card to drill into recipients carries the selected date range to the recipients page, so the detail view shows the same window as the number you clicked.
• Recipients page date picker: a date range picker appears on the broadcast recipients page, and the selected range is preserved through breadcrumb navigation and the Back button to the details page.

These changes are behind a feature flag and are not yet available in production.

When you navigate back from an SMS send details page — using either the Back button or the breadcrumb — the date range you had selected is preserved in the list view.

The recipients page Back button is not affected by this change.

All existing actions are preserved. Admins see a ⋯ menu with Export (when permitted) and View archived contacts. Non-admins get Export as a standalone button. The Import menu still offers CSV import, import from an integration, and view import history. The Add a contact button — or an upgrade prompt when you're at your contact limit — remains in place. On a segment, the last-updated timestamp and Update now button still appear.

On the refreshed contact detail page, five secondary actions open in a right-side sidebar drawer instead of centered floating dialogs: Add to a List, Add Automation (and view automation details), Add a task, Add Deal, and Send a Campaign. The drawer is 33% wide and goes full width at 500px viewport, matching the pattern already used by Notes.

Only one drawer is open at a time. If you try to replace a form with unsaved changes, you'll be asked to discard them first. A form that's actively saving ignores a new open request. Inside a task or deal drawer, pressing Escape closes the drawer along with any open date picker or Stage menu.

This change applies to the refreshed contact detail page. The previous contact page, account page, and Reports keep their existing modals.

Import warnings surface in the "Needs your attention" section on the review screen after import, and as banners on import-complete cards on the homepage, post-opportunity flow, and Contacts empty state. Each warning tells you how many lists were created versus attempted, which lists were refused, and how many contacts were affected. Warning text is formatted for your locale.

One thing to be aware of: the homepage import-complete card still shows "All assets are imported successfully!" even when a warning is present.

A "Custom connector" tile at the end of the MCP Connectors list lets you connect a server that isn't in the catalog.

Clicking the tile opens a form where you enter the server details. The form validates your input before submitting: empty values, names or descriptions over 200 characters, non-URL values, and unsafe schemes are all caught without making a network request. A valid submission creates exactly one connection. Clicking Back discards any typed values and returns you to the connector list.

Before connecting, a consent view displays the server host, the authorization-server host, and the requested scopes. When the two hosts differ, a warning banner is shown. Clicking Continue starts the authorization flow; clicking Cancel returns you to the filled form with nothing written. If the server reports no authentication requirement, the consent step is skipped and the connection proceeds directly. After connecting, the server URL is visible in the connector's detail view.

The Campaigns calendar toolbar and scheduling dialog have been updated to align with the rest of the campaign experience.

The toolbar gains a "Go to today" button. The view dropdown and best-send toggle sit with the rest of the toolbar controls. Recommendation chips are labeled "Best" and "Industry best." "Campaign type" is sentence-cased across the calendar and the campaign page breadcrumb.

The schedule dialog stays on screen, uses the campaign name field style from the campaign page, and shows delivery date and time — including time zone — in a consistent format. "Learn more" is a link. The selected card uses a single 2px focus border, and schedule card styles are shared with the campaign flow schedule step.

Typing in the "To" field of the desktop test preview no longer triggers a new preview request on every keystroke or blanks the preview while you type. The preview updates only when you click the Preview button. If the address is unchanged from the last preview, clicking Preview refreshes it.

The "Has Downloaded File" condition no longer appears in the segment builder inside the Campaign create segment modal.

The suggestion chips on the Campaigns index page, under "Tell us what your next campaign is about," each carry an icon matched to their label type rather than showing the AI sparkle on every chip.

| Label type | Icon | |---|---| | Launch / new product / release | Lightning | | Welcome / onboard / new subscriber | Heart outline | | Announcement | Megaphone | | Promotion / sale / discount / offer / coupon | Tag | | Newsletter / digest / roundup / recap | Document | | Anything else | Email message |

The Campaign preview clears the spinner and shows a notification when the preview request fails, when a Campaign has no messages, or when your account has no email address on file. A Campaign with no messages stops the spinner and shows an empty body.

The loading indicator shown while Active Intelligence generates a Campaign displays only progress steps. Failure information no longer appears in a place meant for progress tracking.

A Starred toggle appears in the Business Goals toolbar, before the Status filter. Turning it on filters the view to your starred goals. The filter persists when you change the Status filter, and the state is preserved in the URL so you can bookmark or share a filtered view.

In the Archived view, stars are shown as read-only on cards and table rows. Starring is only available in the Active view.

This feature is behind a feature flag and is not yet available to all accounts.

Business Goals generate data-driven measures — charts derived from your goal statement — replacing the previous demo presentation overlay. Each goal produces up to five measures, rendered as charts in the Performance panel.

Measures are generated from your goal and replayed with fresh data at strategy time, so the Performance panel reflects your actual account data rather than a demo fixture.

Selecting Edit opens the shared labels modal; selecting Delete removes the folder.

Two display issues were also corrected: opening a folder previously showed the full total count but fetched only the first 20 assets per type; and folder contents ignored entitlement filtering that was already applied to the card count. The open folder neither fetches nor counts asset types your account can't use, so the count on the card matches what you see inside.

The folder grid spacing on Campaigns, Automations, SMS, and Forms pages was also corrected: folder cards previously sat flush against the filter toolbar. All four pages leave 28 px between the toolbar and the cards, consistent across light and dark modes.

The Contacts "Edit all" action applies only to the contacts on screen when your view is filtered to a specific list.

Previously, contact IDs were sent in a format the API ignored, which caused the bulk edit to apply to every contact in the account instead of only the intended ones.

The "End wait" action appears in the row menu and the selection toolbar. It shows up only when Manage Contacts is opened from a wait block queue — not from the global header — so it's available only where it applies.

Clicking a contact name in the Automation Canvas path view navigates to the contact page without a full page reload. Cmd/Ctrl/Shift clicks still open the contact in a new tab.

Opening the Contacts page from the Manage Contacts drawer on a wait block in the Automation Canvas filters by that specific wait, so you see only the contacts waiting at that particular step. Opening from the automation header still filters by the full Automation.

The email navigator in the Automation Canvas email preview stays fixed at the bottom of the panel while you scroll, so it remains accessible regardless of where you are in the preview.

A previous change had hidden the Apps marketplace nav entry for all reseller children, which affected Developer Sandboxes and other branded reseller clients that use ActiveCampaign branding and features. Both are visible again.

The caret on the Allow / Ask / Deny dropdowns on the Customize > Permissions page is visible in dark mode. Light mode appearance is unchanged.

Active Intelligence topic suggestions are filtered so that only ideas grounded in your account's actual data appear — fewer generic placeholders, more ideas you'd actually send.

Country-aware holiday suggestions. Holiday topics are based on your account's country, read from your default footer address. If the country can't be determined, no country-specific holidays are suggested rather than defaulting to US holidays.

Shopping-season topics only for retail and e-commerce accounts. Topics like Black Friday and Cyber Monday appear only for accounts in e-commerce or retail categories.

General fallback topics removed or capped. Broad catch-all suggestions appear only when no other relevant idea is available, capped at two, and are never used to fill slots alongside real suggestions. Accounts with no fitting ideas receive an empty list.

New ideas require real Campaign content. A maximum of two new ideas are returned only when your account has best Campaigns or best subject lines in its content library; otherwise no new ideas are returned.

Cleaner loading state. The suggestions area renders nothing while ideas load, then fades them in once real ideas arrive. If your account receives no ideas, you see a clean empty state rather than a spinner that disappears.

Topic ideas in Active Intelligence could previously include statistics pulled from a different campaign — for example, an idea labelled "niche topics, over 30% clicks" where that 30% figure came from an entirely separate campaign. The rationale you were reading to evaluate an idea wasn't actually about that idea.

Two safeguards now keep figures where they belong. The model is instructed that any statistic — a percentage, count, or amount — may appear only in the rationale of the candidate it came from, and that brand-new ideas (those with no source campaign) must state no statistics at all. A second, code-level check drops any new idea whose rationale contains a digit, so the rule holds even if the model doesn't follow the prompt. Rewritten candidates keep their own figures unchanged.

The rationale you see for each idea now reflects that idea's own data.

On the Customize > Homepage Experience page, when a coming-soon placeholder save fails, Active Intelligence keeps the idea suggestion pills visible so you have a clear next step. Previously, a failed save left only an error banner with no suggestions and no path forward. Suggestions appear for all states except unsupported data, which retains its own suggestion set.

The dividing line between the Active Intelligence sidebar and the agent page renders as a clean single line. When the sidebar is closed, the page panel keeps its own left border.

A relay improvement also prevents Active Intelligence from reporting an Automation as paused or activated when the status did not actually change.

One thing to know: typing the exact phrase a card button sends — for example, "Pause this automation." — triggers the action immediately without showing the card first.

Clicking an insight in the notifications drawer reopens the same chat that was started the first time, rather than opening a brand-new conversation on every click.

Previously, each click sent the "explain this insight, then create it" prompt into a fresh conversation, losing earlier context and causing Active Intelligence to create the Campaign or Automation again — piling up duplicates.

The fix remembers which chat each notification insight opened, per user, storing the conversation ID in memory, on the insight's metadata, and in local browser storage. If the saved chat has been deleted, it forgets the reference and starts a fresh one.

Text written before a tool call streams as progress, not stored as the answer — if a tool delivers less than promised, the reply says so.

What's more consistent: - Conversation history as real dialogue. Earlier turns arrive as native user/assistant messages. In testing, answering Active Intelligence's own question with a name previously caused the question to be asked again; after the fix, 6 of 6 runs acted on the answer correctly. - All Brand Kits named in context. Active Intelligence lists every Brand Kit by name, domain, and ID, and accepts a kit name, domain, or ID to repaint under. Previously only one kit was in context, so "move this to our other brand" prompted a generic question without naming the available kits. - Page address reaches Active Intelligence on every page. On a Contact's page, asking about "this contact" previously returned "which contact?" because the page address only reached the model for pages a context provider recognised. The address is passed on every page. - Tool routing stays within the request's actual menu. Active Intelligence no longer references or attempts to call tools removed from a given request, which was causing errors on mobile and other restricted clients.

When Active Intelligence creates a Campaign for an account whose Brand Kit was auto-created at signup or scraped from your website — and you've never reviewed or confirmed it — a disclosure line appears on your next new Campaign. Previously, this disclosure only appeared when a brand candidate was applied during that same session, so an auto-generated kit could go undisclosed indefinitely.

The disclosure is worded to match how the brand details were sourced (for example, website scrape, past campaigns, or inferred). It appears on Campaign creation and on variants styled under an existing kit. It does not appear on edits or before a style pick. If your Brand Kit fields were set by you directly, no disclosure appears. You can update your Brand Kit at any time.

When Active Intelligence edits the content of an Automation's emails, it updates emails built in the current editor and leaves older-editor emails alone. The reply names each email that was skipped and explains that those emails were built in an older editor, so their content can't be changed with AI — but their subject lines and preheaders still can.

Previously, the message said the whole Automation's content couldn't be changed, which was no longer accurate.

Active Intelligence contacts insight cards are more accurate and better structured. Cards no longer skip data tools or invent figures and chart points. Bullet counts match the expected range of 6–7, sub-headers the format does not call for are removed, the detailed summary is populated, and dashes no longer appear in titles.

The Active Intelligence chat composer accepts a broader set of file types and allows more files per message.

• Markdown (.md) and XML (.xml, including RSS and Atom variants) files are accepted. Previously these were rejected with "That file type is not supported."
• Up to five files can be attached per turn, via the paperclip menu or drag-and-drop. Previously only one file per turn was allowed.
• A turn containing only file attachments and no text can be sent.

If a file is refused or fails to upload, a notification appears and the remaining files continue uploading. All Active Intelligence composer surfaces share the same updated file-type list. Note that .txt files remain unsupported.

When you ask Active Intelligence to list your Scheduled Tasks, it names each task individually and shows its status and run setup.

Scheduled tasks show their next run time in your account's local time zone (for example, "Monday, January 7, 2030 at 10:00 AM"); draft, paused, and done tasks show their status and configuration but no next-run time, since they don't run until activated.

The load fetches all pages of your account's tasks in a single pass per turn, removing the previous fixed limit.

When you opened the Change look picker on a campaign you had just created, Active Intelligence could briefly show stock template previews for every look instead of your campaign's actual preview. This happened because the system hadn't yet finished syncing the new campaign before the picker tried to read it.

The picker retries the campaign read up to three times over 2.5 seconds when the campaign was created within the last minute, so you see your campaign's real preview instead of generic stock templates.

Custom CSS generated by Active Intelligence for a new form applies in the form preview immediately, before you save. You no longer have to save first just to see how the styling looks.

When you send a 1:1 email from a Contact's record or an Account's page, that email — along with its opens and link clicks — appears in the Deal's Recent Activity and Activity tab.

Previously, only emails composed directly from Deal Details showed up there; emails sent from Contact or Account pages were missing from the deal's timeline.

The Emails filter, Load More, and all other activity tab controls continue to work as expected. Recipients who are on the deal but are not the primary contact are not included, consistent with how the deal view has always worked.

The "New" badge on Reports → Your Dashboard in the left navigation uses the standard promotional chip style — solid blue — rather than the AI chip style. "New" is a promotional callout, not an AI surface.

The "Start from scratch" link in the Automations recipe modal reads "Start from scratch" — consistent with how it appears in Campaigns and Forms. The change is in English; other languages will update through the standard localization process.

The "Has Downloaded File" condition no longer appears as an option in contact search, the list segment builder, the Active Intelligence segment preview, or the Campaigns custom conditions modal. It has also been removed from the conversion event segment builder in Settings › Conversions.

The "Has Downloaded File" condition no longer appears as an option when editing a contact score rule's conditions or when building a custom audience for an SMS broadcast.

If you have existing score rules or SMS audience conditions that used this filter, review them to confirm they're still targeting the contacts you intend.

The button that opens the brand-look picker in the header of an AI-written email reads "Change look." Its behavior — applying the selected look to the open email — is unchanged.

Autonomous Insights insight cards that compare account performance against industry benchmarks are more reliable.

Previously, the quality review recomputed benchmark gaps independently and sometimes got the arithmetic wrong — for example, rejecting a valid card with a note like "0.0% cannot be 1.39 points below a 1.39% median." The benchmark gaps computed during production are saved directly into the insight's metadata, and the review reads those verified figures instead of recalculating.

In testing across 390 calls on 130 production-derived cards: 37 of 49 gap cards that were previously rejected incorrectly now publish, and no cards that were previously published are now rejected.

Two gaps are closed:

• Conversion attribution. A single combined call retrieves block events and conversion attribution together. If the attribution lookup fails, events are still returned with a note that attribution was unavailable rather than being dropped entirely.
• Email engagement. When an Automation sent email, the report result now carries a note directing the system to retrieve opens, clicks, unsubscribes, and bounces from the Campaign data source. Previously these metrics could be silently absent.

Tested across 120 runs against the final code: all 120 called the correct data sources and made the conversion lookup. Before the fix, approximately 1 in 4 runs failed.

When Active Intelligence generates a Campaign and the run fails after the Campaign has already been saved, the closing message tells you the Campaign is in your drafts.

Previously, any failure — including one that happened after a successful save — returned a generic "I encountered an issue trying to do that. Please try again in a moment" message, leaving you unaware that a draft had been created.

When you ask Active Intelligence to create several Automations in one message, it saves the first one and queues the rest. After each save, it asks — in your language — whether to build the next one. If you agree, the next Automation is built using the same context from your original request. If you decline, the queue closes.

A later message that isn't your immediate reply expires the offer, so short replies like "sí" confirm the queue rather than starting a new Automation. Each Automation gets its own session, so repeating the same request returns the existing Automation rather than creating a duplicate.

Previously, Active Intelligence read only the page URL and ignored the attached Campaign, so requests sent with an attachment were not routed to the Campaign editing tool. The fix means your attached Campaign is the target of any edit request, as expected.

The per-account cap on Scheduled Tasks has been raised from 25 to 100. Only active (scheduled) tasks count toward the limit — draft, paused, and completed tasks no longer occupy a slot.

If you hit the limit, the message tells you to pause or delete an active task to free a slot. Pausing is a valid way to make room without losing a task you want to keep.

When you open the brainstorm chip on the Scheduled Tasks page, Active Intelligence checks your existing tasks before making suggestions, so you don't get ideas for work already running.

Each suggestion is categorized based on what's already set up:

• New — no similar task exists.
• Complement — a similar task runs, but on a different day.
• Revisit — a similar task exists as a draft, paused, or completed task; the suggestion shows its saved schedule and won't carry a new cadence.

Suggestions that duplicate a task already running on the same day aren't shown. Picking a card creates one Draft for new and complement suggestions, or opens the existing task for revisit suggestions. If tasks can't be loaded, Active Intelligence will say the ideas weren't checked against your account.

The issue affected accounts in all regions. Sends proceed through every batch without interruption.

Segment generation and editing requests involving many tags, lists, or cities — which can take 61–135 seconds to process — complete instead of stopping partway through. The previous 60-second limit caused these requests to give up and retry at full cost. The time allowed for generate and edit calls is 180 seconds. All other segment calls are unchanged. If the service doesn't respond in time, you see the message: "The segment service did not respond in time. Please try again."

The date is resolved before the change is applied, so the wait step reflects exactly what you asked for.

When you ask Active Intelligence to show campaigns for a specific list — "Show me campaigns for my [list] list" — it returns a campaign list. Previously, it ran a campaign performance report instead, producing performance output when you only wanted a simple list.

When Active Intelligence creates an Automation and has to leave out one or more trigger starts, it tells you specifically why each one was omitted — whether the start doesn't exist, can't be built, or isn't available on your plan — rather than giving a generic message. The Automation's description also no longer lists the names of starts that were dropped.

When you ask Active Intelligence how an Automation performed over a specific period — for example, "How did automation 289 perform in the last 7 days? Include its email sends?" — the reply includes how many contacts entered and completed the Automation during that window. Previously, phrasing the question with "include" caused those contact counts to be dropped from the answer.

A related fix ensures that "last 14 days" consistently spans exactly 14 days. Together, these two improvements raised the accuracy of automation performance answers from 91.4% to 98.3%.

Ask Active Intelligence about your Engage conversations, flows, or contacts from any page and get back a rendered chart or table, not a plain text summary.

You can ask for conversations broken down by state, assignee, or channel; a conversations table with labels; a flows table showing status, channel, trigger, run counts, and last run; run outcomes for a single flow; contacts entered per step; and messaging volume per day. Active Intelligence reads your flow names at the start of each session (up to 50), so you can refer to a flow by name from any page and it will find the right one. If a filter matches nothing, you get a clear message rather than a blank result.

Automation performance charts generated by Active Intelligence label each series with plain outcome names — Success, Failed, and Waiting — instead of raw internal values. Multiple raw outcomes that map to the same label are merged into a single series, so charts stay within the eight-series display limit and are easier to read at a glance.

Active Intelligence also uses these plain labels in its replies, tables, and charts.

Outcome and action labels are in English only in this release.

The task-creation tool is excluded from autonomous scheduled runs and remains available in interactive chat sessions.

Text in the Scheduled Tasks create and edit view — including helper text, schedule description, and notification email disclaimer — wraps correctly within the left column at any width and no longer gets clipped when the preview panel opens alongside it.

The divider stays hidden while tasks are loading or if the request fails, so the Ideas and Inspiration section spacing is unaffected in those states.

When your account lacks the SMS Campaigns: Manage permission and a Campaign creation attempt is refused, the error message reads: "You don't have permission to create SMS campaigns. Ask an admin for the SMS Campaigns: Manage permission." You get a clear, actionable message rather than a generic error.

When searching Contacts, Deals, or Accounts, custom date fields now support relative date modes using the "is" and "is not" operators. You can filter by days, weeks, or months ago, and days, weeks, or months from now — so you can build searches like "contract date is within 30 days from now" without manually updating a fixed date each time.

This applies to custom date fields only. Standard date fields such as Date Created keep their existing date options unchanged.

The landing pages editor fills the full viewport when you open it. Previously, the editor retained the surrounding application shell in certain preview environments, reducing the available editing space. The full-viewport experience applies consistently across all editor URLs.

After an import finishes, the agent page shows a completion card with a "Review my data" button that opens the import review screen directly. If the export summary request fails, the card still appears so the review entry point is not lost. Closing the review screen returns you to the agent page.

The import review screen also carries updated copy for CRM data: companies, deal pipelines, deals, notes, and owners are labeled correctly in the review rows. The "Couldn't bring over" CRM row is relabeled "Tasks."

Starting a flow. Offers an AI-description path or manual build from a channel and trigger selection. The builder rail shows Channel, Run rules, and Performance sections, with the trigger in its own panel.

New triggers. Any comment, comments on a live video, story replies, reactions (any reaction or specific emoji picks from Instagram's tray or the emoji picker), clicked an ad (listed by name or added by ID), clicked a link (listed by account or created in place), and mention in a comment. Comment, story reply, and DM triggers can match by AI intent instead of keywords. Every trigger is listed in the picker; triggers the channel can't deliver are greyed with the channel's own explanation.

Story mention trigger. Shows who tagged you (anyone, followers, or accounts above a follower threshold) and displays the four most recent mentions as read-only examples, each marked "Would start" or "Wouldn't start."

AI intents. Each trigger's intent list has a test bar that judges a message against every intent and shows each verdict and why. New intents open in the rail with up to five examples and a "Check it before you add it" test. Testing an intent is free; only real runs use AI credits.

Links. A new link's UTM tags sit under its destination. A link can open a chat on Instagram or Messenger instead of a website.

React to message step. Sends one of six reactions on the message that started the flow (story mention, story reply, or DM), defaulting to a heart.

Webhook received trigger. Shows the webhook URL, who it is for, the data fields it brings (available as {{webhook.<key>}} tags), and recent calls with outcomes.

Rail resizing. The builder rail is resizable from its left edge; width is remembered per browser.

Selecting "Duplicate with AI" from the Campaigns list opens the original Campaign as a read-only preview alongside the Active Intelligence sidebar, pre-loaded with a greeting and topic ideas. No copy is created until you make your first change in the conversation — at that point, Active Intelligence creates a copy named for its new content and edits only the copy from then on. The original is never modified.

What carries over automatically: - Layout. A whole-email rewrite changes only the words and content you asked for. Photos and row padding stay put unless you specifically ask to change them. - Audience. The copied Campaign goes to the same lists and segment as the original. - Brand. On a multi-brand account, the copy uses the original email's own brand, not your account's default Brand Kit.

Additional details: - When a copy is made but the requested edit can't be applied, the reply says so clearly — it no longer also says the change was saved. - A copy of an older-editor Campaign whose layout can't be read is refused with a clear message rather than silently rebuilt into a different layout. - If the connection drops mid-turn after the copy is created, the copy's existence is reported right away. - After a remix, follow-up edits in the same conversation land on the copy, not the original. - If you reopen the conversation from the list after the session ends: if a copy already exists, edits continue from history; if only a question was asked, the session behaves like a normal chat.

Dropdowns that open inside modals are visible and usable.

When updating Contacts in bulk, the value input field receives focus automatically after you select a field, so you can start typing straight away without an extra click.

During contact migration, the key step previously showed a generic "invalid key" error when the private app token was missing required permissions — giving no indication of what actually needed fixing. The error message is now specific: it tells you the private app needs updated permissions, so you know exactly where to look.

The fallback row is displayed as Default: "{subject}". The count notes reflect rules only, excluding the default row.

In the Test and Preview modal, the "Test subject and email" field shows a plain summary for conditional email subjects — the number of rules plus the default subject, truncated to 100 characters — rather than the raw conditional syntax. The field is read-only for conditional subjects; the full conditional logic is still evaluated when the test send goes out.

Field labels on the campaign summary page and the new campaign page are sentence-cased. Affected labels include Campaign name, Delivery date, Delivery time, From name, From email, Subject line, Company name, Address line 1, Address line 2, and Start time. Because these are shared labels, the updated casing also appears in the calendar, editor, and SMS schedule. Three new descriptions for the open tracking, link tracking, and Google Analytics toggles are also added to the summary page.

When a Campaign has conditional subjects configured, the subject input field is disabled and visually distinct — matching the style used by the Calculate recipients block. The summary shows the number of conditional rules and identifies the default subject. This prevents accidental edits to the base subject when conditional logic is controlling what recipients see.

On the Business Goals strategies page, the feedback box in the drawer sends your input directly to the goal engine as a steering instruction. When you submit feedback, it ends the goal's current wait and starts the next action immediately — it's a stronger effect than leaving a comment.

While your feedback is being sent, the box is read-only and the send button is disabled. If the send fails, your draft is preserved and an error notification appears. If the goal has already ended, you'll see a specific message for that.

The page no longer opens the Active Intelligence chat automatically when you visit, and the chat no longer triggers a strategy refresh.

The Brand Kit picker's refinement controls display the correct label text. The back control reads "Back to the original looks" and the anchor fallback reads "Original — Look {take}", matching the assistant-composed card labels.

The error screen shown when the platform fails to load uses a branding value cached in the browser. When the cached value indicates branding is off, a neutral placeholder appears instead of the ActiveCampaign logo. When no cached value exists, the screen shows the neutral placeholder rather than the logo. Directly branded accounts on browsers with a valid cached value are unchanged.

Automation email cards display a Translate chip when translations are turned on for that send block. The chip uses the same AI sparkle and gradient border styling used elsewhere and is not interactive — clicking the card still opens the block as usual. The chip is absent when translations are off. Emails in a split test that are translating show both the in-progress chip and the Translate chip.

In the Automation Canvas path view, the run history header previously showed only "Completed" with no explanation of why a run ended. The header subtitle includes the end reason — covering manual cancellation, cancellation by another Automation, pause expiry, deactivation, duplicate detection, and other causes.

When a link to another Automation appears in the end reason, it's shown as the Automation's name. The subtitle truncates with an ellipsis in the fixed-height header and shows the full text on hover. When a run has no steps, the checkmark hover hint is omitted.

The de-duplication heading in the Automation Canvas pause confirmation modal uses consistent text styling, and an info icon next to it shows a Tool Tip explaining how the rules determine which contacts waiting to enter will or will not enter the Automation when it returns to active.

When more than one banner appears on the Automation Canvas at the same time, each one is visible — multiple banners no longer paint over each other.

The pause pill and the "modified since this contact entered" banner in the Automation Canvas path view no longer paint over each other — each occupies its own space in a shared column.

When Active Intelligence asks for a Campaign topic in chat, topic idea chips appear above the chat input. Clicking a chip fills the input with a suggested topic you can edit before sending. This works for both new Campaign creation and "make a campaign like [past campaign]" requests. A topic already stated in your message skips the question and goes straight to image selection.

The placeholder — "Tell me what to change about this email" — appears in the Active Intelligence sidebar so you know exactly where to start without having to figure out what to type.

Two new pickers are available in the Active Intelligence sidebar for campaign creation, both currently behind feature flags.

Brand look picker: A "Change look" button opens a carousel side panel showing up to three brand looks. You can pick one, keep the current look, or ask for more. A refinement carousel shows "More like this" results anchored on the chosen look.

Campaign options picker: Lets you pick from two or three Active Intelligence-generated campaign options or request new ones.

Both pickers are off by default and not yet available to all accounts.

Ask Active Intelligence to look up a segment by name and, when exactly one segment matches, it returns the segment's full card rather than a list of names. Previously, a name lookup always returned a list, even when there was a single unambiguous match.

When there are multiple matches, no match, or a single match with a slightly different name, the list is still returned.

On new accounts, selecting a category chip on the Active Intelligence homepage — for example, "Create Google Ads" — shows the matching idea cards for that category. Clearing a chip with × restores the default view.

When you asked Active Intelligence to act on many contacts at once — tagging 21 people by name, for example — it previously planned one search and one write per person. That pushed past the plan limit, the request was refused, and you were left to complete the work manually.

Active Intelligence plans these requests as a single lookup across all the names followed by a single write — six steps regardless of how many contacts are in the request. Large contact requests complete instead of stalling.

When Active Intelligence tried to edit an active Automation and the platform refused the edit, it previously misread the refusal as "Automation still being updated" and kept retrying. Active Intelligence correctly identifies a refused edit, surfaces the reason and a link to the relevant Automation, and stops retrying. Edits to inactive Automations are unaffected.

You can ask Active Intelligence to activate a paused Automation or pause an active one, and it walks you through the consequences before making any change.

Activating: A confirmation card appears in chat showing the Automation's title, the four consequences of activating, and two de-duplication rules as pre-checked checkboxes. Select "Activate & release queue" to proceed, or "Cancel" to leave the Automation paused.

Pausing: A pause card shows the three consequences, a schedule checkbox, the reactivation time, and the two de-duplication rules. The schedule starts off; turning it on reveals the date, a resume note, and both de-duplication rules. Pausing with one rule unchecked sends that rule as false along with a reactivation time.

Live queue counts are not shown in the card.

Ask Active Intelligence about any Brand Kit on your account by name and it looks up the real data. Previously, questions about a kit other than your account default returned the wrong kit's information, or incorrectly said the kit didn't exist.

Active Intelligence also handles requests to change which kit is set as your account default, and reads tone and competitor information for any kit. It won't report "no tone" or "no competitors" for a kit whose analysis is still in progress — it waits until the analysis is complete before drawing a conclusion.

Questions about the kit already loaded in your chat session still answer instantly, without an extra lookup.

Active Intelligence no longer treats an empty list membership condition as a required entry criterion when setting up Automations. If you haven't configured a list membership condition, it's no longer treated as a blocker.

On the agency overview page, the "Add new account" button is visible only for the top-level account in an agency. Accounts below that level no longer see the button. Non-agency accounts are unchanged.

WebP images sent through the inbox composer are converted to JPEG on channels that don't natively support WebP. PNG images stay PNG on channels that accept them — preserving transparency — and convert to JPEG only on JPEG-only channels. GIF stays GIF on channels that support it. JPEG images are unchanged. The same rules apply to Content Manager images and images that are dropped or pasted into the composer.

One thing to be aware of: the size check uses the source file size. A WebP file under a channel's size limit could be larger after conversion to JPEG — for example, a 600 KB WebP on SMS could exceed the channel cap as a JPEG.

The empty state on the SMS broadcasts page uses the current Messages illustration, matching the visual style of the rest of the product. The caption and action buttons are unchanged.

Previously, the translations list displayed explanatory text in place of a language selector. The list now opens with a "Select a language" prompt, and the languages available to choose from reflect the languages already present in your Contacts — so you see only the options relevant to your account.

The Scheduled Tasks composer suggests three prompts to help you get started:

• "Every Monday, recap last week's campaign performance."
• "Every month, draft a campaign based on content from my site."
• "What scheduled tasks should I create to help my marketing?"

Prompts rise into view one at a time. They disappear the moment you start typing, and if you click away while the field is still empty, they keep cycling.

When you scroll the SMS broadcast list table sideways, the action button column stays pinned to the right and the scroll shadow stays visible even while hovering or selecting a row. Column gaps are wider, dates are stacked, and your account's date format is applied throughout.

The SMS automation report details page includes a date range picker. Selecting a range scopes the KPIs, failures, and replies widgets to that period. When you drill into either widget using "View more," the selected range carries forward. When no start date is set and the end date is today, the page shows "All time."

The selected range also persists as you navigate: clicking a KPI card or going back from the recipients page keeps your range rather than resetting it.

Note that the replies widget filters by send time while the KPI reply counts use reply time, so the two figures can differ for the same date range.

Team members with the campaigns View permission can open SMS Campaigns in the builder to read them. Previously, View-only users were redirected away with no way to inspect a draft or scheduled campaign.

On the SMS Campaigns list, draft, scheduled, and sent campaigns show a "View SMS" action for View users. Sent campaigns with reporting show both "View report" and "View SMS." Edit, Delete, and Duplicate remain available only to users with Manage permission. Campaigns synced from HQ are unchanged.

In the builder, Save, Send, Schedule, and Test controls are not shown for View users. Message content, sender, schedule settings, audience, and all other editable fields are locked. Sent and sending campaigns remain read-only even for users with Manage permission.

Two permission gaps in SMS Campaigns are closed.

Creating an SMS Campaign is blocked for users without the SMS Campaigns: Manage permission. On the SMS Campaigns index, the "Start from scratch" link is visually inactive for those users and sends no request, so they see a clear signal rather than a confusing error.

The Duplicate button is removed from the read-only SMS builder header. Duplicating an SMS Campaign is still available from the SMS Campaigns index for users with the Manage permission, including for sent Campaigns. The read-only builder shows only Exit, reflecting that it's a view, not an editing surface.

View toggles across Campaigns, SMS, Forms, and Automations indexes use a unified style: the selected view shows a light grey background with a dark icon, and unselected views use a softer muted icon.

Corner radius on task run rows, the "See more" button in the Active Intelligence drawer, and view toggles across Campaigns, SMS, Website, and Automations pages — as well as the Business Goals toolbar toggle — has been updated to 8px, matching the "New chat" button already in the drawer.

Folder cards across Automations, Campaigns, SMS, and Forms — including the No Folder card and loading placeholders — use a slightly larger corner radius and a lighter border, reducing visual weight across all folder card states. Hover, press, and focus ring behavior are unchanged.

SMS broadcast list preview cards also use a softer border and a slightly larger corner radius. The hover overlay and Preview button are unchanged, and every row's preview thumbnail is affected.

These changes apply in light mode.

The preflight form used when creating, editing, duplicating, or reusing a Campaign design has been rebuilt with answer cards. Single-choice steps advance automatically when you click a card; the last step submits. Multi-choice steps show cards with checkboxes and a Continue button. An "Other…" card opens a text field. The header shows a progress percentage and a Skip option.

The intake questions are consolidated into five declared needs: campaign topic (required), campaign goal, CTA URL, image selection, and tone. When a Brand Kit already supplies tone, that question is skipped. The CTA URL question is skipped when fewer than two usable links exist on the account.

Typing a message in the chat while a form is on screen closes the form and sends the text as a normal turn.

WebP images sent through the inbox composer are converted to JPEG on channels that don't natively support WebP, so recipients see the image rather than an unsupported format. PNG images stay PNG on channels that accept them — preserving transparency — and convert to JPEG only on JPEG-only channels. GIF stays GIF on channels that support it. JPEG images are unchanged. The same rules apply to Content Manager images and images that are dropped or pasted into the composer.

One thing to be aware of: the size check uses the source file size. A WebP file under a channel's size limit could be larger after conversion to JPEG — for example, a 600 KB WebP on SMS could exceed the channel cap as a JPEG.

The opt-in settings panel in the Forms builder retains its correct padding regardless of stylesheet load order.

Filter bars on the Automations, Email Campaigns, Contacts, SMS, Website Forms, and Reports Campaigns index pages line up with the table columns below them. The gray background strip and side padding that pushed filters out of alignment have been removed.

Three fixes shipped for Engage:

• When creating a new flow, arrow keys keep the channel list open while you move through options. Enter, a click, Escape, or moving on settles the selection.
• A link deleted since it was ticked in a click trigger appears as "Deleted link," ticked, so it can be removed cleanly without affecting other selections.
• The "no account connected" message is held back until connections finish loading, preventing a false empty state on initial render.

The Engage camera icon in the primary navigation animates on hover, matching the motion style of the other navigation icons. The icon's color and resting appearance are unchanged.

Selecting a range narrows the recipients list to that window; changing the range resets the page. When no date range is set, the view defaults to the last 30 days. The filter is also available on the Automation recipients page.

A "Comment" link under each activity event opens an inline text box; save with the button or Cmd/Ctrl+Enter. Existing comments appear under the event, newest first, showing author, text, and relative time. Three comments are shown by default, with a toggle to view more. Authors can edit their own comments in place. Authors and admin users can delete comments, with an undo prompt before the deletion is committed.

Email thumbnails on the Campaigns list match the framing used by Automation thumbnails, with a light outer border and white padding. Status badges — Sent, Draft, Disabled, Active — are correctly sized. Row actions are pinned to the right column, empty values display an em dash, and the right scroll shadow stops at the action button rather than overlapping it.

Two situations are improved:

1. If a follow-up question does not produce a new Campaign version, the edit chips stay visible without re-running the analysis or showing a loading shimmer. Previously the chips disappeared after an 8–12 second shimmer.
2. When the conversation moves on to a different artifact — an Automation, SMS, Form, landing page, or segment — the chips hide. A later Campaign message brings them back.

The Business Goals empty state on the Business Goals and overview pages displays correctly. The prompt header background spans the full width as intended, and the "Or start with a recommendation" section is fully visible rather than hidden behind the header.

When a partner has configured a custom upgrade URL in the Partner Portal, their client accounts are redirected to that URL when they open the upgrade page. The tab is correctly hidden for single child accounts. The redirect applies only when the configured address is a valid HTTPS URL; all other cases behave as before. Expired accounts and whitelisted pages are unaffected. "Manage subscription" links are out of scope for this change.

Clearer naming. Each look is labeled "Look A," "Look B," "Look C," and so on, matching the naming convention already used in the campaign options picker. Previously, a look showed its Brand Kit's name, which read "Untitled brand" for any kit that had not been saved yet.

No duplicate looks. Proposals are compared against your existing brand before reaching the picker, so you no longer see a near-identical copy of your own brand alongside the new options. Color comparison uses perceptual distance rather than exact hex matching, so looks that are visually identical to a human eye are dropped — not just those that share the exact same hex code.

More varied seeded proposals. When a past campaign seeds the brand picker, proposals keep the seed's primary color exactly and differ in at least two other areas — accents, page backgrounds, or type pairing — so the options you see are meaningfully different from each other.

More visually distinct looks overall. The browser's default link color was previously being read as your primary brand color, pulling all proposals toward the same hue. That color is now excluded from brand color extraction. Generation prompts require proposals to use primaries from different color families with accents of different hues. The template preview applies the brand's accent color so proposals look as they would in a real email.

Lineage naming in the Campaign options picker. Refined looks are labeled by lineage: the anchor shows "Original — Look A," variations are labeled A1 through A4, and deeper refinements show labels like A1.1. The carousel is headed "More like Look A" with an always-visible back control, and the anchor wears a distinct gradient Original badge.

The branch picker in the global navigation displays its full list without scrolling. Previously, a height cap cut off the list.

Accounts with a single Automation see it in the list rather than the first-time welcome screen, which appears only when an account has zero Automations.

The confirmation presents both de-duplication checkboxes checked by default. Activating with a rule unchecked passes that rule as off while the other stays on. Canceling leaves the Automation paused. Replying in plain text to use standard rules activates the Automation with both rules on.

When working with Active Intelligence on an Automation, the chat composer shows how many automation emails are selected. Each email card in the Automation preview displays a "Not sent yet" label when the email has no associated campaign, or when its campaign is still in draft. Scheduled, sending, and completed campaigns do not receive the label.

Opening an animated GIF in the content manager detail view plays the animation rather than showing a frozen first frame. Crop mode continues to use a still frame, which is expected behavior. Editing a GIF — crop, resize, or flip — still exports a single frame.

When you ask Active Intelligence about something time-specific — "this week," "last month" — the search returns pages from that period. Active Intelligence also no longer states a publish date for a source unless the source itself shows one; if the date isn't confirmed, it says so. Searches that don't ask about a specific time period are unchanged.

The thinking animation in Active Intelligence keeps playing smoothly while it works through multiple steps to answer your question. Previously, the animation restarted from the beginning each time Active Intelligence moved between its own reasoning and a tool lookup, making the experience feel choppy.

The animation runs continuously until your answer appears. If a step encounters a problem, its own error message appears without interrupting the rest of the response.

Unread task-run dots in the Active Intelligence sidebar previously used a different color from notification dots, making the visual language inconsistent across themes. They now match.

The search field sits below Task runs and filters chats only. Typing a query hides Task runs so results are uncluttered; clearing the search brings Task runs back.

A broad set of visual and reliability issues in the Active Intelligence sidebar and agent surfaces have been resolved:

• Insight cards on the Campaigns, Contacts, and Automations pages keep their full content height when the layout stacks, rather than collapsing to a strip that paints over the list below.
• The agent landing shows two columns from a 1001px window as before.
• Agent polls survive a transient failure again.
• Modals opened over a full-screen takeover correctly dim the page behind them.
• Accounts without the agent feature no longer see a blank drawer column or lose nav corners.
• A reload mid-run keeps the agent's busy state and the open campaign editor.
• Email previews whose frame loads early no longer stay invisible.
• Campaigns pages no longer show an error card for a closed prompt takeover when its code fails to load.
• Agent markdown code blocks display their background color, and the Brand Kit card's social icons draw in their brand colors.
• The loading animation in the agent chat updates to themed light and dark versions.
• The in-progress Active Intelligence thinker card is unframed while active, so text changes during processing feel less jarring; the error state remains framed.

A rendering loop was causing every page that mounts the Active Intelligence tune-insights modal to fire roughly 1,000 re-renders per second while idle, which blocked the global navigation from loading — the top navigation bar stayed on its loading skeleton indefinitely.

The loop is eliminated. The global navigation loads normally and the page is idle between interactions.

When you turn Auto on, the "Auto is on" banner appears above the composer in the Active Intelligence sidebar. In narrow layouts, the banner text wraps correctly and the Auto label and toggle remain visible on the right.

When you open a notification insight, Active Intelligence starts a fresh chat and explains why it would create something before it creates it. The prompt includes the insight title, summary, and original ask, giving you a chance to review the reasoning and decide whether to proceed.

Opening an insight from outside the agent home page sends the same explain-first prompt in the Active Intelligence sidebar. Notification conversations opened with a direct link replay from the top — the request appears, the sending state shows, then the reply.

The bulk-edit dropdown in the Contacts list was rendering behind other page elements, making it inaccessible. The dropdown appears on top as expected.

Ask Active Intelligence what conditions a segment uses and it gives you a clear, readable answer: every condition in plain English, grouped with its boolean operators. You can ask by segment name or ID; Active Intelligence looks up the contact by email first if needed. Conditions that reference other segments show only that segment's overall result, not a nested breakdown.

During campaign variant composition, Active Intelligence shows progress at each phase of generation rather than displaying a generic "Still working" banner after 20 seconds of silence. You can see where generation is in the process throughout.

When you ask for a single version, Active Intelligence skips the options picker and generates one campaign directly — you're not walked through a picker you didn't ask for.

When you ask Active Intelligence for a recurring report, it checks your existing Scheduled Tasks before making an offer. If the same report is already active, no offer is made. If it's in draft, paused, or done, Active Intelligence offers to bring it back. If the same report runs on a different weekday, a second task is offered.

New tasks are always created as drafts; making them live is a separate step.

All attached filenames appear in the chat, in the order you attached them, both while the response is generating and when you reload the conversation. Two files with the same name on one message are kept distinct automatically. Markdown files (.md) are read as text, up to 8,000 characters, with a note if the file was cut.

If you attach more than five files or the total exceeds 40 MB, the upload is refused with a clear message naming the limit. Files over their individual format size limit are also refused at upload, with the limit named.

One thing to be aware of: an import still reads only the email and ignores other attached material.

When creating a new account inline from the add-deal modal, an Account Owner picker appears, pre-filled with the signed-in user. You can choose any user from the full user list. If the typed name matches an existing account, that account is reused and its owner is unchanged.

Clicking an unsubscribe link in a Campaign preview or test email redirects to the list unsubscribe form. Real sends are unchanged.

Unsubscribe links that were corrupted by copy-paste in the email editor are repaired before your email goes out, and at the point of paste in the Classic Designer. Third-party preference-centre links are also preserved — they were previously being replaced with a bare ActiveCampaign unsubscribe URL at send time.

This affects Campaigns and Automations built in the Classic Designer and Stripo-based editors. You don't need to rebuild existing emails; the repair happens on the send path.

Emails already delivered with corrupted links are not retroactively corrected, and stored email HTML that was previously affected is not backfilled. Going forward, links in new sends will be correct.

A Translate checkbox and a Review button appear on each Automation email in the campaign editor.

Clicking Review asks for confirmation before translations are turned on. If there are unsaved edits, the confirmation calls that out, because translations are built from the last saved version.

Once translations are running, Review opens a translations screen inside the campaign panel. The screen lists the original and every translated language, searchable by name or code. Selecting a language previews its subject, preheader, and the full email. A language still generating is marked as translating; one with no copy says it is not translated.

When an Active Intelligence edit finishes on an Automation email, translations are automatically resynced so translated copies stay current with the latest saved content.

Unchecking Translate on an email that already has translations asks for confirmation before deleting them. If the delete fails, an error is shown and the checkbox stays checked. Starting translations on an account with no language field mapped shows a setup prompt rather than a silent failure.

The toll-free registration card supports five states: unregistered (Register button), pending (chip plus 3–5-day copy), approved (number under the title, green chip), rejected with edit allowed (danger banner with reason, field, deadline, and Update button), and rejected without edit allowed (no resubmit, support guidance). The "Needs update" callout shows localized field labels rather than raw field names — for example, "Business website" instead of a machine identifier.

When an account's primary admin record can't be loaded, the platform returns a plain, actionable message — "This account is temporarily unavailable. Please try again later." — along with a reference ID you can use when reporting the issue.

The settings page is accessible from the integrations page as it was before.

When you pick an approved US or Canadian toll-free sender for an SMS Automation or Campaign block, that choice is saved and used when messages send, so messages go out on the toll-free number instead of a 10DLC number.

The selection is only accepted when the toll-free registration is approved. When it is not approved, normal 10DLC traffic continues unchanged.

If you query SMS send totals via the v1 API, the success rate field now returns an accurate fraction between 0 and 1 — for example, 0.8 for 10 sends with 2 failures. The admin report screen was not affected by this.

Retrying an Automation build with the same session returns the original Automation rather than creating a duplicate.

Reseller child accounts whose partner has "Hide product links" enabled receive the unbranded domain-verification email. Direct customer accounts are unchanged.

When a Form is paused, visitors see the form's card with a message and optional link rather than a bare sentence or empty page. The form's header and image are preserved; input fields are replaced by the message, and the submit button is hidden.

The default message is: "This form isn't accepting submissions right now. Please check back later." You can set your own message, button text, and button link in Form Settings (message up to 500 characters, button text up to 60, URL up to 2,048 characters; http and https addresses only; both button text and link must be provided together).

Inline form layouts show the full card. Preference centers show the plain message. Pop-up layouts stay hidden while paused. Embedded forms show the card only when you've set a custom message or button; otherwise they show nothing.

The platform serves the Meta data-deletion status page correctly and looks up the confirmation code accurately. An unknown confirmation code shows that the request was not found. The page never reports a request as pending — a receipt is only written once the request has been fully processed.

Tag names containing underscores wrap inside the Link Actions section of the campaign editor instead of overflowing their container. Names like WEB-DolorMenstrual_2026_Interesados_Newsletter_SOY previously grew past the bordered container because browsers treat underscore-joined text as a single unbreakable word. Hyphenated names already wrapped correctly; underscore-separated names behave the same way. This applies to tag, list, and Automation link-action chips.

The list contact trends report is available through the public API. You can retrieve it using GET /api/3/ai/report_trends_activity, filtering by list_id; passing graph=unsubscribes returns the windowed unsubscribe count in unsubscribe_total.

This endpoint was previously marked internal and excluded from the API catalog, which meant connected tools had no way to call it. It's now available to any tool connected through the API once the updated spec is published. No changes to your existing queries or setup are required.

Two new email-blocking options are available in the Forms builder.

Block personal email providers. A toggle in form settings blocks approximately 6,100 personal email domains — including Gmail, Yahoo, Outlook, Proton Mail, Hey, Tuta, and others. Employer-operated domains are excluded from the block list.

Wildcard domain blocking. Entering *.example.com blocks all subdomains of example.com but not the root domain itself.

Builder UI. The blocked domains setting shows a title, help text, an Add field with paste support (commas, semicolons, spaces, or one domain per line), and blocked domains as removable chips. An inline note appears when a domain is already blocked, and the field explains the 100-domain limit when reached.

Field-level error. When a submission is blocked, the message "[domain] addresses can't be used here. Please use a different email address." appears under the email field and clears as soon as the submitter edits it.

The public URL uses your account's custom domain when one is set, otherwise your account domain. The embed code is a script tag ready to paste into your site; inline-form and preference-center layouts also include a preceding div placeholder.

Both fields are computed on request. Saving a form from the builder ignores these fields — they won't overwrite anything.

When you create an ecommerce customer via the API and a duplicate is matched by phone number, the duplicate error correctly identifies the phone field. This applies when phone-only contacts are enabled.

The previous limit of ten could be reached in a single heavy afternoon of drafting — each Campaign with one or two regenerations uses about three sets. Thirty covers ten Campaigns each regenerated twice, which exceeds any realistic drafting day.

When you use the Custom Objects records API with the same filter key repeated in a request, every value is applied. Previously, only the last value took effect — for example, filtering not_eq for both A and B would only exclude B.

The fix applies to plain repeated keys and array-style parameters. Requests that don't repeat any filter key are forwarded exactly as before.

Three request shapes return a 400 error to prevent ambiguous behavior: more than one value for relationship filters (relationships.primary-contact, relationships.deal, relationships.account); a filter nested deeper than filters[key][operator][n]; and a non-numeric list index. One value for a relationship filter, plain or in a one-element array, continues to work as before.

Custom HTML templates that have one or more categories assigned can be opened and edited in the template editor. Previously the editor stayed blank and showed an error.

Your Contact CSV exports include a new column showing each contact's open-tracking opt-out status. The column appears after the WhatsApp username field and uses 1 for opted out and 0 for opted in. The numeric format keeps the file compatible with re-imports across accounts in any language.

Click to Open Rate is calculated as unique clicks divided by unique opens, rounded to 3 decimal places (for example, 20.053) — consistent with what the Reports page shows. Exports run before this fix will show different Click to Open Rate values than exports run after it.

Each card carries a role label — current, other kit, from your emails, calmer, bolder, or industry — shown as a pill. Real looks always appear before generated ones, and your own brand before industry defaults. Keeping the saved kit card leaves your brand unchanged and triggers no regeneration. Cards that would preview identically to one already shown are skipped.

When there's no brand to anchor from, the picker shows two industry looks instead. Until both sides of this change are deployed together, the picker falls back to the previous lineup.

Browser default link colors were previously being read as your brand's primary color, pulling all proposals in the Brand Kit rebrand picker toward the same hue. Those colors are now excluded from brand color extraction, so proposals are based on your actual brand palette and are more visually distinct from one another.

Brand Kits can no longer be saved with a malformed website URL — submitting an invalid domain or URL returns an error, so only valid addresses are stored. Kits with a valid URL or no URL are unaffected.

For accounts with no business name on file, proposals in the Brand Kit rebrand picker include a short, plausible brand name drawn from your domain or industry — not a description of how the name was derived. This means every proposal looks like a real brand option rather than a placeholder.

Brand Kit logo upload accepts all valid image URLs

Adding or changing a Brand Kit logo on accounts with a custom domain or multi-region hosting previously returned an error even when the image was valid. Logo uploads save correctly regardless of which valid base address the link uses — the Content API base URL, the account's custom domain, or the account's own domain on multi-region accounts.

Template previews in the Brand Kit rebrand picker reflect your brand's accent color as a colored band above the logo. This means the proposals you see in the picker look as they would in a real email, giving you a more accurate read on each option before you commit.

When you pause an Automation, a confirmation modal lets you choose a return-to-active datetime and set deduplication rules. The Automation restarts on its own when that time arrives.

A one-click pause still works as before: it pauses immediately with no scheduled reactivation. If you activate or deactivate an Automation manually, any stored reactivation datetime and deduplication rules are cleared.

The Automation schema generation and build routes are documented in the public API, making them accessible through the API reference and the MCP catalog. Developers can find full documentation for generating an Automation schema and building an Automation — including the request body and response fields — without needing internal access. Runtime behavior is unchanged; only the documentation was previously missing.

Each build call creates a new Automation. Repeat-call idempotency on build is not yet supported.

When pausing an Automation, you can set a reactivation date and time. The Automation returns to active automatically when that time arrives, releasing the queue with the stored deduplication rules. A one-click pause with no reactivation time set continues to behave as before.

This capability is currently off by default and must be enabled for your account before it appears.

Translation controls are built directly into the campaign editor for Automation emails. A Translate checkbox and a Review button appear on each Automation email.

Clicking Review asks for confirmation before translations are turned on. If you have unsaved edits, the confirmation flags that, because translations are built from the last saved version.

Once translations are running, Review opens a translations screen inside the campaign panel. It lists the original and every translated language, searchable by name or code. Selecting a language previews its subject, preheader, and the full email. A language still generating is marked as translating; one with no copy says it is not translated.

Translation status updates every few seconds while a run is in progress. When Active Intelligence finishes an edit on an Automation email, translations are automatically refreshed so translated copies stay current with the latest saved content.

A few edge cases are handled explicitly: unchecking Translate on an email that already has translations asks for confirmation before deleting them — cancelling leaves translations intact. If the delete fails, an error is shown and the checkbox stays checked. Starting translations on an account with no language field mapped shows a setup prompt. When translations are unavailable, the control is hidden. A failed translation run shows a banner with a retry option.

When a send failure occurred at the start of an Automation series, a contact's run could appear as Queued in Manage Contacts even though the Automation itself had never paused. The status is now accurate: a failed series-start send preserves the original batch and does not incorrectly mark the run as paused. Only runs that were genuinely released from a pause are restored to paused on a send failure.

The status badge on the Automation Canvas (Active, Paused, Inactive) updates immediately whenever the automation's status changes — no page reload needed.

All status-change paths are covered: a single status save, a pause expiry, a scheduled reactivation, an entitlements-triggered deactivation, and a bulk status change via Active Intelligence. Writing the same status a second time does not send a redundant update.

The status badge on the Automation Canvas (Active, Paused, Inactive) updates immediately whenever the automation's status changes. Previously, the real-time update only fired after a bulk status edit.

A single status save, a pause expiry, a scheduled reactivation, and an entitlements-triggered deactivation all update the badge immediately. A bulk change still updates the canvas, and writing the same status a second time does not send a redundant update.

The status badge on the Automation Canvas (Active, Paused, Inactive) updates immediately whenever the Automation's status changes — no page reload needed.

All of the following now update the badge immediately: - Scheduled status changes: when a pause expires and the Automation moves to inactive, or when a scheduled reactivation moves it to active. - Active Intelligence bulk status changes: when Active Intelligence pauses or activates an Automation, the badge in an open builder session updates immediately. - Single status saves, pause expiries, scheduled reactivations, and entitlements-triggered deactivations.

A bulk change still updates the canvas, and writing the same status a second time does not send a redundant update.

Clicking the count in the Goals column takes you directly to the Goals report for that Automation, so you can move from a high-level view to the detail you need in one step.

The Goals column is not sortable and is not included when you export the report to CSV.

When Active Intelligence describes an Automation, it resolves and names the conditions on If/Else, Wait, Jump To, Split, and start trigger blocks — the same readable text the Automation Canvas shows. Previously, these appeared as raw identifiers that were difficult to interpret.

If a condition can't be resolved, it's omitted from the description rather than shown as an unreadable string.

Active Intelligence brand monitoring includes pages on your www host and subdomains — for example, news.example.com — when those pages appear in your site map. Previously, only pages on the exact brand host were included, so www and subdomain pages were silently skipped even when your site map listed them.

A page is treated as on-brand when its host matches your brand host, or when it appears in your site map and its host is either the www equivalent of your brand host or a subdomain of it. Pages on parent domains and look-alike hosts remain excluded.

Your brand's own homepage is also no longer listed as one of its important pages — it's already scraped as the main page, so including it again was redundant.

API v3 field details endpoints documented

GET /fieldDetails and PUT /fieldDetails/{id} are now documented in the v3 API spec.

What each endpoint does:

• GET /fieldDetails lists standard contact, deal, and account fields. Each record includes the field's id, name, type, required status, and default value. An optional type filter limits results to one record type.
• PUT /fieldDetails/{id} updates the required flag and default value for a field. Validation rules apply: marking a field required needs a non-empty default value (except for contact email); contact email cannot be made optional; and at least one contact name, email, or phone field must stay required at all times.

Custom fields are not covered by these endpoints — they continue to use /fields, /dealCustomFieldMeta, and /accountCustomFieldMeta as before.

The GET /api/3/fields and GET /api/3/fieldOptions API endpoints handle non-numeric or negative pagination parameters correctly, matching the behavior of every other list endpoint. Large imports that previously stopped partway due to a malformed request complete as expected.

The API documentation for the group definitions endpoint correctly shows the required JSON request body — a `groupDefinition` object with required `groupType` and `label` fields, and an optional `ordernum`. Following the previous documentation produced a 422 error; the API endpoint itself is unchanged.

Automations that use a custom object trigger created by Active Intelligence trigger correctly when a qualifying event occurs. A missing flag at creation time was preventing these Active Intelligence-generated triggers from firing; it's set correctly when the trigger is created.

Automations using a "Wait until [custom date field]" block were getting permanently stuck. Two separate code paths both produced an invalid value being sent to the database for date comparisons — including one triggered specifically by blank-check operators ("is empty" / "is not empty"), which was responsible for roughly 4,332 errors per 24 hours. Both paths are fixed. Contacts in affected Automations advance normally through wait blocks.

When Active Intelligence generates a new WhatsApp marketing template, it draws on your Brand Kit and account marketing context. The Brand Kit (voice profile and brand guidelines) is applied at generation time. Active Intelligence also reads up to five active Business Goals and up to five recently completed Campaigns to inform the template's tone, naming, and content.

This context is used only for new marketing templates. It is skipped for edits, submissions, channel moves, listings, and utility templates. If fetching context fails, generation continues with a warning rather than stopping.

The automation email sent webhook event has a display label in the webhook configuration UI. The label appears correctly whether "All Lists" is selected or a specific email list is chosen, and the index table tooltip shows the event name rather than undefined.

When you edit a custom connector's URL or description after cancelling the consent view and resubmitting under the same display name, a warning banner appears explaining that the display name must be changed before the edit can be applied. This prevents the edit from failing without any explanation.

Thumbnails on the Automations list and Campaigns list use 8px rounded corners and a softer border color. The hover state, loading placeholder, and error frame all match, so corners stay consistent regardless of what the thumbnail is doing. The large Campaign preview modal is unchanged.

When you choose to resend a split Campaign, a modal asks which version to reuse. The winning version is pre-selected if it's still attached to the Campaign; if no winner is available, nothing is pre-selected and Continue is disabled until you make a selection.

Continuing creates a draft copy of the Campaign using the selected version and opens it in the editor. The modal notes that the copy will no longer be a split test. Nothing is sent from this modal — you choose recipients in the editor.

A "This page is taking longer than usual" message and a reload button appear when a page's loading state has been visible for 15 seconds, so you're never left with an indefinitely spinning placeholder. The page can still finish loading on its own.

When viewing a segment's Contacts tab, the segment name appears directly above the KPI card. The agent segment preview does not yet show a name on this surface.

The conditions accordion in the segment builder toggles only when you click the chevron or label text, so accidental clicks elsewhere in the header row no longer collapse or expand conditions you're working on.

When adding recipients to a scheduled task notification email, the disclaimer shown for non-admin recipients reads: "Recipients who are not admins or users will be informed that they must contact the task creator to stop receiving notifications. By adding them, you confirm that you have their consent to send these notifications and understand that you are responsible for managing opt-out requests."

The disclaimer is also restyled to match how other disclaimers appear in the product — smaller supportive-colored text that blends into the form, replacing the previous banner format. The disclaimer appears when at least one non-admin recipient has been added.

The four Scheduled Tasks status chips — Active, Paused, Done, and Draft — use a consistent size and style. Active previously rendered at a larger text size than the others. Draft changed from an outlined pill to a grey chip with a pencil icon. These changes apply to Scheduled Tasks pages only.

When scheduling an Automation to return to Active after a pause, the time picker offers only :00, :15, :30, and :45 increments. This matches the 15-minute schedule that actually reactivates paused Automations, so the time you set is the time your Automation comes back — no more waiting up to 15 minutes longer than expected.

The default suggested time is one hour from now, rounded up to the nearest 15-minute mark. Typing a time that doesn't land on a 15-minute boundary shows a step error and keeps the "Pause automation" button disabled.

Hovering the labeled Schedule task button on Active Intelligence Home shows a two-line Tool Tip. The first line displays the slash command /schedule-task and the second line reads "Run a prompt on a repeating schedule." The Tool Tip is left-aligned with the button edge. The icon-only split view retains its existing single-line tooltip.

Reports shown during the migration onboarding experience display a "SAMPLE REPORT" label, making clear that the data shown is illustrative.

Three new SMS registration paths are available.

France: French businesses can register for SMS. France is added to the brand registration business address Country dropdown, and the State/Region field accepts free text for countries other than the US.

Denmark: Businesses with Danish mobile numbers can register for SMS. The registration form collects business name, Danish address, and CVR (Danish business registration number). Validation enforces an 8-digit CVR and a 4-digit postcode; Danish characters (Æ/Ø/Å) are accepted. Estimated processing time is 10 business days. MMS is not supported for Denmark.

US/Canada toll-free numbers: Accounts can start, save, and submit US/Canada toll-free SMS number registration. A toll-free registration card appears in the SMS registration hub for Admin users when the toll-free feature is enabled. The card supports five states: unregistered, pending (3–5-day processing), approved, rejected with edit allowed, and rejected without edit allowed. The "Needs update" callout shows readable field labels rather than raw system names. Processing takes 3–5 business days.

Business ID formats are handled automatically: a Canadian Business Number or EIN with extra characters has them stripped before validation, so either standard format is accepted.

From the SMS Automation block details page, the failures and replies footers link directly to a recipients page filtered by deliverability or engagement. You can drill from a summary stat straight to the list of affected contacts.

The Contacts tab in a segment includes a Phone column. Contacts who have only a phone number — no name or email — appear in the table alongside other contacts. When a contact has no name, email, or phone on file, a dash is shown, matching the main Contacts list. Name, email, and phone all link to the contact detail page when data is present.

Pause and resume controls appear in the Forms list and in the builder, including a paused banner with an inline Resume button.

While paused, visitors see the form's card with a message — by default: "This form isn't accepting submissions right now. Please check back later." You can set your own message, button text, and button link in Form Settings. Builder previews and screenshots continue to work while a Form is paused, and double opt-in confirmation links already sent to contacts remain active. Duplicating a paused Form creates an active copy; the original stays paused.

Preference centers do not get pause/resume.

The Pages AI prompt banner on the Pages builder reads "Generate a page in seconds" — sentence case, consistent with the Forms banner.

MCP clients can send, schedule, or unschedule a Campaign through five tools: prepare_campaign_send, confirm_campaign_send, prepare_campaign_schedule, confirm_campaign_schedule, and unschedule_campaign. All send and schedule actions require a two-step prepare-then-confirm flow so you see a server-written summary before anything is committed.

One-time and A/B (split) Campaigns are supported. RSS, recurring, date-based, autoresponder, and Automation Campaigns are not.

The confirmation token is single-use, scoped to the account and user that prepared it, and expires after 10 minutes. If the Campaign's content, lists, sender, subject, or schedule changes between prepare and confirm, the operation is cancelled automatically.

Sending or unscheduling requires campaign-send permission; scheduling requires campaign-edit permission.

The integrations page is sharper in several ways: integration names highlight correctly on hover, integrations without a name sort last in both ascending and descending order, and Tool Tip text displays at regular weight.

Mark unread. A new option in the "…" menu on open conversations lets you mark the conversation unread. The option appears only when the thread is already read and is disabled for viewers and while a write is in progress.

Bulk mark read. Selecting multiple threads and marking them read sends a single request for all selected threads, up to 500 at a time.

Auto read-receipt behavior. The read receipt is tied to the thread's newest message per visit. Marking an open thread unread sticks while you remain on it; a new message arriving on an open thread marks it read; returning to the thread later marks it read again.

The impersonation banner in the partner portal displays correctly in both light and dark themes.

The Forms builder header and sidebar field tiles display with correct padding and spacing on all builds.

The Flows builder has been rebuilt around a docked rail that sits beside the canvas. All step editing happens in the rail. On the canvas: drag a step onto a "Move here" target to reposition it; the Add step "+" button fades in rather than shifting the flow; wait steps are named by their duration (for example, "Wait 1 day"); comment triggers display their keywords.

Run history is a table — individual runs open in the rail rather than on a separate page. "Move here" drop targets appear only where a drop would actually change a step's path. A flow whose stray step is re-homed on load no longer opens showing "Unsaved changes."

The inline banner reads: "Contacts are still on a step you removed. Keep that step until they move past it, then save." The save is held until the step is restored or contacts have moved past it, and your unsaved edits remain in the Automation Canvas throughout.

Any other save conflict still shows the existing message.

Previously, expanding the Run rules section at the bottom of the sidebar produced a second scrollbar, and using it shifted the entire builder up, leaving a blank band at the bottom of the screen. The Re-engage sequence editor had the same issue. All three affected views are corrected.

A cluster of visual and interaction improvements shipped for the Engage flow builder.

Wait steps are drawn as a named capsule on the connector line — for example, "Wait 1 day" — rather than a full step card. Who is waiting appears beside the duration.

Comment trigger cards name their keywords in the card title, such as "Someone comments 'LINK' or 'PRICE'", and show whether the match rule is "any" or "every". Past three keywords, two are named and the rest counted.

Drag-and-drop shows a dashed "Move here" pill in every valid drop gap while a step is held, making it clear where the step will land. Invalid gaps show nothing.

Connector colors stay grey at all times. A previous change had drawn the path of the most recent run in green, making one contact's history look like part of the flow design.

Step cards are laid out on a grid so the underline sits just below the step name. Cards no longer show a contact-reached count (that figure belongs to the rail); contacts parked on a wait still show.

Zoom controls stay fixed on the canvas regardless of which panel is open in the rail.

Panning works from the gaps between steps; only the "+" button holds the board still.

Activity range in the agent view is a dropdown menu (Last 24 hours, Last 7 days, Last 30 days, Custom range…) beside the Filter control.

Weekly reply and handoff counts in the agent header display as "200+" when all 200 loaded turns fall inside the week, making clear the figure is a floor rather than a complete count.

Builder rail dropdowns use a consistent style: 38px height, strong border, 4px corners, blue focus ring, and compact rows with a check on the current selection.

Engage Settings is rebuilt as a section-based navigation with a dedicated page for each connected account, making it easier to manage multiple channels and spot accounts that need attention.

The nav shows all channels, each connected account with a status indicator (amber or red when attention is needed), Messaging rules, Members, Audit log, Webhooks, and General. Every section and account has its own URL; old tab URLs redirect automatically so existing bookmarks keep working.

Each account page shows status notices with a Reconnect option, run rules (re-entry and stop-on-message, saved as you change them), ice breakers where the channel supports them, and the provider's webhook status. The flow builder reflects the channel's real rules and shows "Differs by account" only when accounts genuinely differ.

Members is read-only, showing real names and roles from your ActiveCampaign groups. Password, organization, billing, API keys, usage, and Delete workspace have been removed from Engage Settings, with links to the relevant ActiveCampaign pages in their place. The delete control for inbound endpoints sits beside Rotate on the Webhooks page.

The delete confirmation dialog for Engage Links displays the number of clicks recorded on the link being deleted, with correct singular and plural wording ("1 click" vs. "N clicks").

The range control offers 7 days, 30 days, 90 days, all time, or a custom date range. The channel filter and range control drive the summary, figure strip, chart, Channels, Delivery, and CRM figures together. Custom date ranges read at most 90 days back, ending today.

The Flows, Links, and Posts tables read the same data as those pages, so numbers match across the product. Delivery shows reads on top of delivered, with each outcome as its own share of sent.

When a channel filter is applied, only message counters have values — everything else shows "Not by channel." Lists are lifetime-only and carry an "All time" tag unless the page is already set to All time. The channel filter counts sends over the selected date range, consistent with the Channels card.

On the email subscriptions page, the empty-state message — shown when there are no subscriptions, no search results, or no lists — is centered and capped at 500px wide. Previously the text stretched the full page width and sat left-aligned beneath a centered illustration.

Two additions to the Engage settings panel:

Delete inbound endpoints: Each inbound endpoint row in the Developers tab has a Delete action alongside Rotate. Deletion is confirmed through a dialog. If a delete is refused, the confirmation stays open with a failure notification so you can retry. If the one-time token card was showing the deleted endpoint's token, the card clears.

Webhook subscription status: The channel drawer in the Channels tab shows a Webhook row in the media and cost table. It reads either "Receiving messages, comments" or "Not receiving events", with a timestamp showing when the status was last checked. Connections with no recorded subscription show no row.

How each field type saves: - Text and currency fields save 800 ms after typing stops, or on Enter or click outside. - Dropdowns (status, stage, owner, source) save on selection. - Date and datetime fields save when you pick a day. - Multiselect fields save 800 ms after the last toggle or click outside. - Pipeline saves when you pick a stage.

Pressing Escape closes the editor and discards unsaved changes. Required fields never save empty — clicking outside reverts to the last saved value. Navigating away mid-edit saves any pending value. Each autosave shows a success confirmation.

1:1 emails appear in the Activity tab. 1:1 emails sent from a deal, along with their tracked opens and link clicks, appear in both the Recent Activity card and the Activity tab with labels "1:1 Email Sent", "1:1 Email Opened", and "1:1 Email Link Clicked".

Activity entries open a detail drawer. Clicking any entry in the Activity tab opens a drawer showing the full title, date/time, and who made the change. Task entries show type, assignee, due date, and description. Note entries show the full note text. Deleted records show a "(Task Deleted)" or "(Note Deleted)" label.

Completed tasks show full details. Completed tasks on the Tasks tab display the same detail as open rows — title, due date and time, assignee, and up to 100 characters of the task note.

Deal header shows forecasted close date and age. The Forecasted Close Date and the Created/Updated dates with relative age (e.g., "Created 09/28/2026 (2 days ago)") are restored to the deal detail header.

Tab labels show item counts. Notes, Files, and Inbox tab labels display item counts (e.g., "Notes (2)", "Files (0)").

Filter by activity type. A dropdown lets you view All activities, Campaign Emails, Deal Updates, Emails, Notes, or Tasks. "All activities" merges both the campaign-email feed and the deal-activity feed without gaps; "Load more" fetches from whichever list is furthest behind.

Named activity entries. Each entry identifies the specific task, note, contact, role, pipeline, or custom field it refers to. Custom-field values are formatted to your account's date and datetime format; multiselect and checkbox values are joined with commas. Notes are trimmed to 30 characters at a word boundary. If a record can't be found, the entry keeps generic text; a deleted contact shows as "Deleted Contact (ID: n)."

Campaign Emails. A new data source shows the primary contact's campaign sends, with a link to the campaign report. A deal with no primary contact shows "No Campaign Emails activities found."

Deal Updates excludes notes, tasks, and emails; Emails includes 1:1 opens and clicks; site visits and tracked events are not included in All.

A "Custom" free-form crop choice is available in the Content Manager image editor, sitting alongside the existing fixed-ratio crop options. This shipped as part of the Content Manager page migration, which brings a redesigned library and image detail view — including system folders, custom folders, tile and list views, bulk selection, drag-and-drop upload, and image editing with crop, rotate, flip, text, undo/redo, and save-as-copy.

Previously, a custom connector sharing a name with a built-in integration would incorrectly inherit that integration's curated tools and instructions. Custom connectors are now routed to the correct generic toolset regardless of their name.

Tiles, list rows, the detail thumbnail strip, and modal thumbnails display as expected. Copy URL and View full image produce working links.

The Content Manager modal in the inbox composer has been rebuilt with a full set of file management capabilities. You can browse files in grid or list view, search, sort, use bulk actions, upload by button or drag-and-drop, and preview files. Folders are fully supported, including search and sort within folders.

The image editor offers Crop, Resize, Flip, and Text tools. Save overwrites the existing file; Save as copy appends "-copy" to the name. An AI image generation dialog is also available from the Content Manager page.

The Content Manager page shows the correct confirmation text when deleting or removing multiple files. The bulk-delete confirmation no longer displays a raw translation key, and removing files from a folder shows the correct copy with the actual file name.

Previously, hovering or selecting a row caused the scroll shadow to disappear, making it harder to tell there was more content to the right. The shadow remains visible regardless of row state.

Two fixes to the Contacts bulk-edit modal (Edit all → Update a field):

Selecting a field in one row no longer clears the value dropdown for fields already selected in other rows. Previously, each row's field selection replaced the shared options data rather than merging into it, causing other rows' options to disappear.

Dropdown options for custom fields appear in a consistent order across all selection types — dropdown, checkbox, and property. Options are sorted by their defined order; options with an unusable order ID sort last.

The upgrade icon in the contact limit banner displays in the correct teal color on the Contacts, Campaigns, Automations, and Active Intelligence sidebar pages.

The contact limit banner behaves differently for accounts that cannot self-serve an upgrade. Accounts that require assisted upgrades see a message directing them to reach out, with a single Contact us button instead of the self-serve upgrade buttons. Starter plans at their limit also see the Contact us button. The banner is hidden entirely for cancelled or expired accounts. Non-admin users on assisted-upgrade accounts see a message to contact their account admin.

Supported languages include Spanish, French, Italian, German, Portuguese (BR and PT), Dutch, Hungarian, Russian, Czech, Polish, Bulgarian, Slovenian, and Indonesian. Values are trimmed and case-insensitive. A blank cell leaves the contact's existing value unchanged. A row with any other value is rejected. Re-importing the same value does not move the compliance timestamp.

The import error message is updated to reflect the full set of accepted values. This field is available when the open-tracking opt-out feature is enabled on your account.

Contact archiving messages updated for clarity

The contact archiving flow shows clearer messages at each stage.

When a cleanup run stops because your contact deletion allowance for the billing period is exhausted, you'll see: "We archived {archived} of {total} contacts. You've used your contact deletion limit for this billing period." This message does not suggest retrying, because the allowance doesn't refill until your next billing period.

Other messages in the archiving flow are also updated: the confirmation modal makes clear that archived contacts can be restored later, without naming a specific timeframe, since the restore waiting period varies by account. The error message for a run that archived nothing is reworded for clarity.

The Disconnect and Dismiss buttons in the Connectors modal display their correct labels instead of raw internal key strings.

The @ context selector in the Active Intelligence sidebar includes a "Connectors" group listing your connected apps — such as Calendly and Stripe — alongside Campaigns, Automations, Campaign Templates, and Triggers & Actions. You can reference your integrations directly when working with Active Intelligence.

The Automation Canvas includes a configurable webhook action form. You can set up webhook actions directly within an Automation through the form UI, without leaving the canvas.

In Campaigns, the conditional subject line modal accepts up to four rules plus the default subject line — five subject lines in total. Once four rules exist, the option to add another is hidden; it returns when a rule is removed.

Campaigns that already have more than four conditional subject rules continue to load and work in full — the limit only applies to adding new rules going forward.

Campaign subject lines can be built with conditional logic. Each rule combines one or more conditions and maps to its own subject line, with a required default subject line as the fallback.

For example: if a contact has the tag bargain_hunter and %PRODUCT% is T-shirt, send one subject line; if they have a different tag combination, send another; otherwise use the default.

A new icon button inside the subject field opens a modal editor. Supported conditions include tag exists/doesn't exist, contains/doesn't contain, value exists/doesn't exist, and comparison operators. GEO fields are supported. The personalization button works inside each subject line field in the modal.

When a conditional subject is set, the subject field shows a read-only summary ("Conditional · N possible subject lines" / "Otherwise: [default]") instead of the raw template.

Previews and test sends render conditional subjects correctly.

The conditional-logic icon on the email subject line displays the correct Condition icon. The icon is also hidden for split Campaigns, where conditional subject logic isn't supported.

In the Campaign builder, the conditional subject line indicator and its icon label are reworded to be easier to understand at a glance. Each conditional subject rule is numbered, so you can track and reference individual rules without losing your place.

When connecting a previous contact database whose account has IP-based security restrictions, the import flow previously showed a generic "invalid key" error. The error message tells you specifically that the connection was blocked by your source account's IP security settings, and instructs you to check your email for a blocked-IP notification from that provider, authorize the IP, then try again.

When a Campaign's recipients are a saved segment and you choose "Add conditions from a saved segment," the Campaign's own segment loads its conditions correctly so you can preview and reuse them without the preview spinning indefinitely.

The button in the header of an AI-written email that previously read "Change look" is labeled "Rebrand." The label reflects what the action does: selecting a brand repaints the open Campaign in place rather than generating a new one.

Forms can refuse submissions from specific email domains — up to 100 of them. Visitors who submit a blocked address see an inline message naming the domain and asking them to use a different one. This keeps low-quality or competitor addresses out of your contact list without any manual review.

To set this up, open the form builder and find the "Block email domains" setting. Enter the domains you want to refuse. A toggle also lets you block personal email providers (Gmail, Yahoo, and similar) in one step.

A few things worth knowing:

• Wildcard patterns like *.example.com block subdomains only. Add example.com separately if you also want to block the root domain.
• The input field disables once you reach the 100-domain limit.
• The setting saves with the rest of your form — no separate save step.

The Automations first-time-use empty state button reads "Start from scratch" again. The button opens the recipe takeover, the same as the "Start from scratch" link in the regular Automations overview header.

The All Automations CSV export modal displays a title, a failure message, a progress note, and a timeout note so you always know the status of your export.

The 15-minute steps match the schedule that actually reactivates paused Automations. The default suggested time is one hour from now, rounded up to the nearest 15-minute mark. If you type a time that doesn't land on a 15-minute boundary, a step error appears and the "Pause automation" button stays disabled until you correct it.

In the Automation Canvas path view, the run history header subtitle includes the end reason — covering manual cancellation, cancellation by another Automation, pause expiry, deactivation, duplicate detection, and other causes.

When a link to another Automation appears in the end reason, it's shown as the Automation's name. The subtitle truncates with an ellipsis in the fixed-height header and shows the full text on hover. When a run has no steps, the checkmark hover hint is omitted.

Emails linked to an Automation by series ID display the visual indicator that the email belongs to an Automation.

The Automation Canvas path view accounts for changes made to an Automation after a contact entered it. Steps a contact moved through after the path view's revision was drawn appear as normal blocks. Steps that were deleted before the contact reached them appear as ghost blocks — greyed out — surfacing the discrepancy directly in the canvas rather than leaving it unexplained.

The path view also reports when the live Automation has a newer revision than the version the contact entered, and shows the date that newer version was saved. This lets you understand why a contact's path may look different from the current Automation design.

The Automation Canvas preview in the Active Intelligence chat panel renders correctly. The Automation Canvas builder at its own page was never affected.

When scheduling a pause reactivation for an Automation, the date-time picker lists only five-minute intervals — for example, 10:05, 10:20, 10:55 — consistent with the rest of the platform. The browser's native picker, which showed all minute options, is no longer used. Typing a time that isn't a multiple of five keeps the "Pause automation" button disabled.

Clicking an Automation name in the path view opens the Automation Canvas directly; clicking a Contact name opens that Contact's record. Cmd/Ctrl/Shift clicks still open the target in a new tab.

After Active Intelligence edits an email inside an Automation, the "Edit email" and "Edit with AI" hover actions on the canvas step work correctly and no longer show stale state.

The hour and minute offset fields in the Automation Canvas "Wait until" date step accept only values available in the dropdown.

Active Intelligence web search results include the publish date of each source when one is available. Each result shows the date — for example, "published 3 days ago" — or states explicitly that the publish date is unknown. Web search hits that carry no date in the underlying data will show "publish date unknown"; this is expected.