What we shipped

The total deal count appears next to "Deals" in the spreadsheet view heading. It sits next to "Deals" rather than next to the pipeline name — a deliberate design decision, given that the pipeline name serves a different dropdown function in the new layout.

The Smart Start Guide checklist no longer shows a permanent full-height scrollbar. The list is capped at a maximum height and uses a styled thin rounded scrollbar. A progress count and full-width progress bar appear above the scrollable area.

Known limit: This improvement is visible on partner and reseller accounts only.

Once the widget is on your dashboard, you can see at a glance how each goal is performing by lifetime conversions.

Known limits: Conversion counts are lifetime totals only — date-range filtering isn't available for goals data. Progress toward a target isn't shown because goal targets are stored as free text rather than numbers. Time-scoped questions like "how are my goals tracking this month" aren't supported; the widget won't place a lifetime number under a monthly label.

Tool Tips on stat cards and section headings in the dashboard are triggered by hovering the info icon rather than appearing on the element itself.

When viewing a contact with many tags, tag badges in the contact detail panel render all at once after all name lookups have settled, rather than appearing one at a time and causing the row to repaint repeatedly.

The collapsed view shows one page of tags with a "Show N more" button rather than fetching every tag on open. Expanding a contact with many tags issues one lookup per remaining visible tag ID.

Previously, Active Intelligence grouped trend data in UTC while the standard Open/Read Trends report applied your account's time zone — so the figures differed, and the same contact could be counted multiple times across different days of the same weekday. All time grouping now respects your account's local time zone, so the numbers you see in chat match the numbers in your report.

Previously, the voice profile was built from your account's most recent campaigns in recency order, so a sale email would be styled after whatever went out last, even if that was a newsletter or announcement.

No action is required. If Active Intelligence can't identify the type of email from your brief, or if your account hasn't sent that type before, it falls back to recency order as before.

Previously, permission accordions opened expanded regardless of what permissions were already granted, requiring you to scroll past expanded panels before making changes. Saving a group without expanding an accordion continues to work as before. The tab row alignment now matches the layout used by other settings pages.

The sync drawer includes a multiselect option so you can select more than one item at a time, and scope drawer radio buttons display correctly in all themes.

When you're in the Deals CRM pipeline view (pipeline or tasks view), a Search all pipelines action appears in the search field while it's focused or holds a term. Activating it takes you directly to the all-pipelines spreadsheet view in one step, carrying your typed search term and your account-wide filters — owner, status, and tag — with you.

Stage filters are dropped when you cross pipeline boundaries, since they're specific to a single pipeline. The back button returns you to your previous view with your pipeline restored.

Known limit: If you type a term and activate the action within 300ms, the search term may not be carried back when you press the back button. This is expected behavior.

When adding recipients to a scheduled task preview, the recipient search field returns all users on the account, not only admins. Previously, non-admin team members had to be added another way.

A new Ideas portal lets you submit ideas or flag issues without leaving the product. Open the help menu in the global navigation to reach the modal, where you can describe your idea, search for similar ones already submitted, and see which page you're on when you submit.

Once submitted, your ideas are available at Settings › Ideas, where you can see status updates, customer notes, and voting.

The portal is off by default. It must be enabled for your account before it appears in the navigation.

The goals table loads correctly and displays all rows without errors. Checkboxes are vertically centered against row text, and the checkbox appears to the left of the chevron arrow.

Errors that appeared when an account's billing currency was not supported are resolved.

In edit mode, you can select notifications in the Active Intelligence notifications center and delete them. The delete button shows how many items you've selected, confirms successful deletion with a count, and reports clearly if something goes wrong.

The filter bar on the Deal Sales Performance report has consistent vertical padding, matching the spacing used by the Deal Task Overview and Deal Forecast reports.

Previously, the timeframe appeared under the title and the freshness stamp appeared at the far right of the card header. They're now consolidated into one line, making it easier to scan at a glance. Timestamps differ per card because each widget refreshes independently. If one label is absent, the dot separator is omitted.

The activity stream and pending task list on the contact detail page load correctly and stay loaded.

A mismatch between how the page was requesting data caused the entire feed to go blank for a number of accounts, retrying every 30–60 seconds without success. A separate issue caused deleted records to be re-fetched repeatedly on every feed refresh. Both are resolved.

A related fix prevents the add-contact association panel from replacing the entire account page when a search selection is cleared.

No action is required.

Charts in Active Intelligence, Campaigns, Contacts, Automations, and the Partner Portal have updated visuals. Changes include a six-series color palette, area fill gradients, hover tooltips, and bar value labels that automatically use a contrast-safe color. Each chart assigns colors from its series labels, with red reserved until the other five colors are used.

No action is required.

The loading spinner on the campaigns page is correctly positioned with consistent spacing, matching the in-page loading indicator.

No action is required.

Those controls only work during the preparation phase; once emails are going out they have no effect. Pause and Stop remain available during preparation, and Stop remains available for paused campaigns and campaigns awaiting review. Accounts without the sending-progress experience feature see no change.

Bulk edit and archive modals on the Contacts page open, apply changes, and close as expected.

In Automations, messages that had never been sent were incorrectly labeled as drafts. The label correctly reflects that these messages have simply not been sent yet.

Active Intelligence reporting widgets show the time window that was actually queried, not the window you requested. If you asked for "last 35 days" but the data resolved to a 30-day query, the widget subtitle reads "Last 30 days" rather than "Last 35 days."

When a window is both shorter than requested and ends materially before today, the subtitle switches to absolute dates rather than a relative label, so the description is always accurate. Calendar-month and trailing-month windows are unaffected — partial months are expected there.

Active Intelligence is more transparent when something goes wrong during an approved action.

Partial and failed writes: Active Intelligence distinguishes between fully applied, partially applied, rejected, and unknown write outcomes. When only some steps of an approved action land, the summary says which steps applied and explicitly states that the rest did not, rather than claiming the whole request is done.

Approved calls that return a plan instead of executing: When Active Intelligence is told to execute but the backend returns a plan to approve instead, it retracts any "executing now" text it already showed and displays an "awaiting approval" state rather than "nothing changed." This persists across page reloads.

Known limits: The new failure path does not yet increment the sub-agent error counter.

When you open a Deals CRM pipeline view via a shared link and some filters can't be applied — for example, because a referenced field no longer exists — a warning is shown instead of those filters being silently dropped.

Three areas regain correct spacing:

• The Finish and Next buttons in the campaign flow header have their right margin back.
• The Navigation Color swatch on the Settings Appearance page has correct spacing next to its label.
• The lock icon and "Account Security Measures" heading on the Settings Security page have correct spacing, as does the level label below it.

Three screens regain correct spacing:

• The Finish and Next buttons in the campaign flow header have their right margin back.
• The Dark/Light label in Settings Appearance has its left margin restored.
• The "n/4 completed" chip in Settings Security has its gap from the card title restored, and the card content layout is correct again.

The segment footer has 32px horizontal padding on both sides, matching the page header, so the save button aligns with the close button above it. In edit mode, the loading spinner is overlaid on the button at the correct size, keeping the button's height and width fixed while saving.

The segment builder in the score editor stretches to the full width of the rule card instead of collapsing to content width. As a side effect, error messages appear below the builder rather than squeezed against the right edge.

No action is required.

SMS reply rows in the contact activity feed were showing "Replied to SMS message [campaign] with" and then nothing. The reply text is displayed correctly.

No action is required.

The gap between the last alert and the subdomain banner on SMS pages is now 32px as intended, not 64px.

Filter dropdowns across seven reports used to load every record on your account before the picker became usable — up to ~200 sequential requests on mount, then blocking the browser while rendering thousands of options. On a 10,000-row account this measured as 42 seconds with the picker disabled and a 38-second frozen browser.

Filters load one page of results on open, then search your account as you type. Your selected value is pinned so it survives a search that doesn't include it in the first page.

The seven reports updated:

• Tag Trends — Tag filter
• Form Submission — Form filter; a zero-result search no longer shows the "create your first form" empty state
• All Automations — Folder and automation filters
• Deal Task Overview — Six unbounded filters (deals, task owners, contact tasks, deal tasks, any tasks, pipelines) replaced with server search; four bounded filters (task type, user group, task outcome, pipeline) keep a single request with client-side filtering
• Deals Overview and Deal Task Metrics — Owner filters
• All One-to-One Emails — Automation, email, and label filters
• Deal Forecast — Owner filter

No action is required.

The Active Contacts, Contact Trend, and other KPI widgets on the legacy overview dashboard display correctly without an unwanted scrollbar.

Three layout issues are resolved across the automations, campaigns, forms, and SMS index pages.

• Filter dropdowns are sized correctly. Each control sits at its natural width, the search input scales between a minimum and maximum width, and dropdown menus have a consistent minimum width.
• The search icon in dropdowns is corrected. The magnifying glass and placeholder text are the right size and weight.
• Folder cards display at equal widths. A folder with a long unbroken name no longer forces its column wider than the others. Folder names wrap to two lines rather than cutting off with an ellipsis.

Toggle buttons on the automations index page have also moved to the left side, and caret sizes in filter controls are smaller.

Campaign, form, SMS, and automation folder views paginate their contents at 12 items per page. A numbered pager appears when a folder has more than one page. Out-of-range page requests clamp back to a valid page.

Known limits: Page-position announcement copy, an in-flight page-turn indicator, and search-truncation surfacing for forms and SMS are not yet available.

Error toast notifications across the platform display with the correct red styling, including background, border, and icon. Previously they rendered as unstyled boxes.

Large numbers on engagement chart axes are displayed in compact notation — 10K instead of 10,000 — keeping charts readable when contact or engagement counts are high.

In folder view on any asset index — campaigns, forms, SMS, and automations — folders that contain none of that page's items are grouped behind a "Show N empty folders" toggle, collapsed by default. Opening it lists them below your populated folders; the label flips to "Hide N empty folders."

A few things worth knowing: the toggle resets each session, it doesn't appear when every folder has content, and it disappears while you're searching (since empty folders are already excluded from search results). "Empty" is evaluated per page, so a folder full of forms counts as empty on the Campaigns page.

On the contact record, email activity rows previously showed only the subject line. The full message body renders for one-to-one emails, campaign replies, campaign sends, and link-data opens — matching what the previous version of the contact record showed.

The orders widget on the contact record refreshes automatically when new order data arrives, matching the behavior of the customers and conversions widgets. This applies when the updated contact record is enabled for your account.

On the updated contact and account detail pages, a rendering problem in one section — such as custom objects, notes, tasks, the activity stream, deals, or entity actions — previously replaced the entire page with an error screen. Each section is now isolated, so a problem in one area removes only that section while the rest of the page continues to work normally.

Report filter inputs no longer show a visible text label above the field. The accessible label for screen readers is unchanged, and the filter bar's layout and alignment are unaffected.

Campaign cards in folder view show a status chip, so you can see at a glance whether a campaign is draft, scheduled, sent, or needs attention — without opening it. The chip matches the status shown in the campaigns table, so the two views always agree.

Campaigns pending approval show "Attention needed." Series campaigns show "Automation" once scheduled or sent.

When you answer an Active Intelligence permission card — choosing which steps to allow or skip — your answers are saved and restored if you reload the conversation. Previously, reloading showed the card as unanswered even though you'd already responded.

The transcript also shows the answered permission card itself rather than the plain confirmation sentence the card sent. If a conversation has no saved record (older sessions), the original confirmation is preserved so no decision history is lost. All four answer states are covered: all-Allow, executed subset, all-Skip, and partial answer blocked by a skipped dependency.

Several issues have been resolved:

• Audience no longer dropped on single-list accounts. If your account had exactly one list, the review-and-send panel could show "Choose a list" with a blocker badge even though Active Intelligence had already proposed an audience. Your chosen audience carries through correctly.
• Campaigns send only to the audience you picked. A campaign could previously send to an audience nobody selected when attaching a list with no segment. The send configuration clears any previously held segment when no segment is chosen, so sends go only where you intend.
• Recipient lists chosen during generation are preserved. Active Intelligence reads your campaign's current lists first and merges its suggestion onto them. The applied plan also survives page reloads.
• Spinners no longer run indefinitely. Audience lookups that previously left loading indicators spinning resolve correctly.

Known limit: Active Intelligence reads your campaign's current lists and then writes the merged set as two separate steps. A manual save made between those two steps can still be overwritten. Eliminating that gap requires a backend change that isn't yet in place.

When you ask Active Intelligence to generate a campaign, it reads and applies the audiences you've already built in your contacts, and confirms the outcome in your conversation's language.

Previously, generated campaigns ended up with no audience at all, and the audiences you'd built weren't offered as options. Four separate problems caused this:

1. The audience wasn't being saved to the campaign. The request to attach an audience was going to the wrong address, so the platform rejected it before processing. After that was corrected, a second problem appeared: the platform reads campaign data from a replica that hadn't yet caught up with the write, so the system incorrectly reported failure even though the audience had been set successfully.

2. Your audiences weren't appearing as options. Audiences you create in your contacts are stored separately from classic segments. Active Intelligence was only reading the classic catalog, so named audiences like "Nurture Engaged Subscribers" or "Welcome New Subscribers" never appeared, and the system composed a new segment instead of using one you'd already built. Both catalogs are read and merged.

3. Selecting an existing audience could silently overwrite it. When an existing audience was chosen, the process could re-link it in a way that overwrote its name and deleted its conditions. The audience's resolved ID is carried forward instead, so no re-linking occurs.

4. Audience outcomes are reported in your language. When an audience couldn't be set, the reply used hardcoded English regardless of the language you were using. The outcome travels as structured data, and Active Intelligence describes it in its own words, in whatever language you're using.

This capability is available on accounts where the campaign send configuration feature is enabled.

When Active Intelligence runs read-only lookups before asking for your approval, the approval card tells you what it already found — not just what it's about to do.

Previously, every step appeared as pending, even lookups that had already run and returned results. The card now presents completed lookups in plain language with their resolved values, and frames only the actions still waiting for your go-ahead as pending. For example, instead of listing "Look up the TikTok list details" as a future step, the card reads "I found your TikTok Leads list and can create the new Snapchat list with the following settings copied from it."

Your account's own font families are available in two places.

Email editor: The redesigned email editor's font picker includes your brand's custom fonts, matching what the previous editor offered. If the font fetch fails, the editor falls back to its built-in list and still opens normally.

Brand kit font picker: The brand kit font picker expands from 39 fixed options to approximately 1,967 searchable open-source font families. A default set loads on open; typing searches the full catalog. Fonts are grouped as "Displays everywhere" (families that need no download) and "Web fonts" (stylesheet-required), with no-download fonts listed first.

You can send a specific trigger or action block directly from the automation builder as context to Active Intelligence. It reads that context and uses it when responding, so you get answers that are grounded in the exact part of your automation you're working on.

The split campaign report header shows one of three states: pending (the test is still running), selected (a winner has been flagged), or no automatic winner (an even split). The supporting text states the configured rule — for example, "The winner will be chosen by open rate 7 days after the test variants finish sending" — with no countdown shown.

The chip uses a neutral style for pending and no-automatic-winner states, and a success style when a winner is selected. All wait-period units are covered (hour, day, week, month, with plurals). English copy is the fallback for any locale where translations aren't yet available.

The spacing between the preview label and value in Settings, Advanced is correct again.

The Users and Groups tabs in Settings display with the correct horizontal padding.

On the campaigns list, the disabled "View details" link in the Sending Status column shows a Tool Tip when detailed sending progress isn't available. The Tool Tip reads "Detailed sending progress is only available for 30 days after sending," so you know exactly why the link is inactive.

Previously, lists beyond the first 20 were hidden with no indication more existed. The pickers page through your full list, so no list is out of reach when building segments.

Your active filters and query parameters stay in place after you save a segment. Previously, saving redirected you to a URL built from the page path alone, so parameters were dropped every time — including the one that controls the Save Segment button itself, which meant the button disappeared after every save and had to be restored by hand. Parameters controlling read-only state, truncation, import mode, and collapse state were lost the same way.

All parameters carry through the redirect after saving. The exception is the sample JSON parameter, which is intentionally excluded: keeping it would reload the sample data instead of the segment you just saved.

Decorative Active Intelligence markers in condition-type dropdowns are hidden from assistive technology, since they sit next to their visible labels and don't need to be announced separately.

In the segment builder's campaign, 1:1 email, and SMS selection dropdowns, truncated option names appear in a readable Tool Tip with white text on a dark background.

Previously, the Tool Tip rendered white text on a white background, making truncated names invisible — though still selectable. The Tool Tip appearance now matches the other truncation Tool Tips already in the segment builder.

Both halves of the AND/OR toggle are real buttons that assistive technology can read, grouped so they're announced as one control. Tab, Enter, and Space all work as expected. Mouse behavior is unchanged.

The search bar in the Segment Library panel stays pinned at the top as you scroll through segment cards. Previously, the search input scrolled out of view with the list, requiring you to scroll back to the top to refine a search.

The improvements come from eliminating redundant recalculations that were happening on every interaction. Option lists, data mappings, and condition controls are now recomputed only when something actually changes. The result is a builder that keeps up with you instead of lagging behind.

When reviewing an Active Intelligence-generated campaign, the recipients popover previously showed "Saved filter (details unavailable)" instead of the audience name. The popover looks up the audience name directly when it isn't immediately available, matching the resolution the campaign page already used correctly.

You can upload SVG files in the content manager. SVG files appear in the file picker, pass upload validation, show an SVG type label in the detail pane, and display as thumbnails in both grid and list views.

SVG files can't be opened in the image editor, which is designed for raster formats — the same restriction that applies to PDFs and GIFs.

On the SMS reports page, entering an unsupported page size or misaligned offset in the URL snaps to a valid page instead of showing a broken view.

Previously, the opt-in step in the SMS product tour gave no visible feedback while waiting for keyword confirmation. The status message now appears during that wait, so the tour doesn't appear stalled.

In Settings → Users and Groups → Edit Group → Permissions, a new SMS section lets you grant or restrict access at a granular level. Individual permission rows cover registration, reporting, sending options, and more across SMS Campaigns, SMS Inbox, SMS Keywords, SMS Management Access, and SMS Sending Options.

When an SMS is scheduled outside your quiet hours but a delivery delay would push it into them, the platform withholds the message rather than delivering it at the wrong time. These sends are recorded with the status: "Not sent. Delivery was delayed and would have fallen inside your quiet hours."

The status reads "Not sent" rather than "failed" because nothing went wrong — the send was withheld intentionally to respect your quiet hours settings, and no credit is consumed. The status appears in both the broadcast report and the inbox conversation view.

If you see this status on a send, it means your quiet hours settings worked as intended.

SMS double opt-in activity entries in contact history are translated into your account's configured language. Previously they appeared in English only, regardless of your account's language setting.

Revenue sorting in the revenue report view works correctly. Translations for the revenue Tool Tip are also in place across supported languages.

When you remove contacts from automations, inactive automations are included alongside active ones.

Previously, only active automations were considered, which could leave contacts enrolled in inactive automations without any way to clear them through the standard removal flow.

Contacts waiting to enter an automation open in a full-screen view instead of a side panel, giving you more room to review and act on them. The banner and modals stay in sync with the list as you work.

Clicking a contact or account link from a deal detail page reliably opens the correct record.

Previously, with the updated contacts or accounts experience enabled, clicking the primary contact on a deal detail page would blank the page and redirect to the home screen. The same problem affected account links. This affected 27 entry points into contact records, including contacts list rows, deal list links, task links, and the deal activity drawer. Direct loads and opening links in a new tab were unaffected.

When you click a wait or goal queue step in an automation, you can open the Manage Contacts drawer right there. The contacts list is already filtered to that step, so you see exactly who's waiting and can act on them immediately.

From the drawer you can sort the contacts list by column, run bulk actions including Manage Duplicates and End Current Run(s), and jump straight to a contact's path view — pre-filtered to the step you came from. A loading indicator shows while results are fetching.

Together, these tools give you a complete picture of who's at any point in an automation and the ability to act on them without leaving the builder.

The forms list page could show a blank screen when any form had a thumbnail that couldn't be loaded. The page handles missing or unparseable thumbnail URLs gracefully, showing a placeholder icon instead of going blank.

Previously the filter bar appeared on the bare white page background. The controls themselves are unchanged; only the visual presentation is updated to match the rest of the platform.

Previously, the forms filter bar appeared on the bare white page background rather than inside the shared toolbar. The change is visual only and applies automatically. No action required.

Previously, some strings in the Active Intelligence form tools appeared in their raw, untranslated form. All text in the form list retrieval, form statistics, and aggregated form statistics tools is fully translated and displays correctly.

You can browse your automations, campaigns, forms, and SMS assets in a Folders view. Toggle between the familiar list and a grid that shows every folder with the assets it holds. Drill into any folder; search, status filters, and filter chips all work the same way in both views.

Each index table gains a dedicated Folders column so you can see which folder an item belongs to at a glance. The terminology across all four surfaces has also been updated: "label" is now "folder" everywhere.

The Folders view is rolling out gradually. When it reaches your account, you'll see the toggle — until then, your pages look exactly as they do today.

Known limit: Folder drill-down currently shows the first 20 assets per folder. Support for browsing beyond 20 is built and on its way.

On the Apps tab in the Add-an-action modal, a category dropdown lets you narrow the app grid to a specific category instead of scrolling through all apps. Category names match the labels used on the apps list page, so the vocabulary is the same in both places.

To use this: The category dropdown is rolling out gradually. Contact your account team if you'd like it enabled.

Known limits: Two category labels may display as raw text keys until translations are finalized.

The order-state and store filter dropdowns on the contact record open fully instead of being clipped by the panel edge. Both menus open fully, and the Products button label and arrow have correct spacing.

When deleting a user fails, the error messages displayed are correctly translated instead of showing untranslated or missing strings.

You can open an automation and land directly on a specific canvas action by sharing or bookmarking a URL that includes that action's unique step reference. The action opens and receives focus automatically, so there's no need to scroll or search for it.

This is useful when collaborating with teammates or when support or documentation points you to a specific step in a complex automation. The link works for any automation canvas action.

When the Deals CRM advanced filter can't load its available filter fields, you see a properly translated error message with a count of unavailable filters, plus Retry and Clear all filters actions. Previously, the error placeholder appeared with missing or untranslated text.

When you open a shared or bookmarked URL that carries filter conditions that can't be applied — because they reference a field or operator that no longer exists — you see a "Some filters could not be applied" warning.

Opening the date picker while using the add-contacts drawer keeps the drawer open so you can continue without interruption.

When building conditional content blocks in the message variables editor, your account's custom contact fields appear in the field picker dropdown. Previously the picker used a static list that never included custom fields, even though those same fields were already available in the "insert personalization tag" feature. The picker now pulls from the same data source used elsewhere in the editor.

This applies to the conditional content picker at the message variables page only.

The attribution section of the contact record includes a conversion touchpoint timeline. Each conversion shows one marker per source touchpoint, spaced by elapsed time, ending at the conversion. You can scroll through touchpoints using a scrollbar, drag a scrubber, click the bar, or use arrow buttons.

Selecting a touchpoint shows a detail table with: Time, URL, Source, Medium, Campaign, Term, Content, and Referrer. Clicking a value in a conversion activity's detail panel expands that conversion, collapses the others, and scrolls it into view. Clicking a deep link from the activity panel while the Unconverted Sources tab is active correctly switches to the Conversions tab before expanding the row.

Ecommerce filter menus (order-state and store dropdowns) no longer get clipped by the panel.

This is part of the ongoing contact record update and is available when the updated contact record is enabled for your account.

When the Contacts page can't load its first-time experience, it shows a notification so you know something didn't load correctly. Previously the page recovered silently, giving you no signal that anything had gone wrong.

Previously, these dropdowns floated outside a bordered control rather than matching the inline style used elsewhere. The change is visual only and applies automatically. No action required.

The subheading shown on the Contacts empty state screen reflects the current design.

The count appears for basic search terms and advanced search or segment views. Filter-only views (Tags, List, Status) are unchanged. Large totals are formatted for your account's locale — for example, "8,450" in US English. The count doesn't appear while a search is in flight, so it always describes the results you're looking at.

In the "Add to list" and "Add to automation" modals on the contact record, checkboxes respond to clicks correctly.

The Sending Status column shows a relative send time — such as "2 days ago" — beneath each campaign's status chip.

For campaigns sent before the sending-progress service existed, that relative time is now a tooltip trigger: hover it to see the audience size and send time. That context was previously unavailable for those older campaigns. Campaigns with full progress data are unchanged.

Draft campaigns show a "Last saved" label with a relative time in the same column.

On the Campaigns list, the status shown under the campaign name and the status shown in the Sending status column now match for three cases that previously disagreed:

• Disabled campaigns show a grey "Disabled" badge (previously showed "Draft"). The cell is not clickable and has no "View details" drawer.
• Automation campaigns show an "Automation" badge instead of "Sent" or "Scheduled." Send date, recipient counts, progress bar, and details drawer continue to work as before.
• Campaigns under review that require action show a yellow "Attention needed" badge (previously showed "Pending review"). Opening the details works the same as for "Pending review."

Note: a campaign showing "Sending" can still appear as "Delayed" in the Sending status column when a send is running late — this is intentional.

The send progress drawer displays a "Waiting for approval" step when a campaign is held in review before it's allowed to send. You get a clear status instead of a blank or ambiguous state while your campaign awaits approval.

When you change a CTA link URL, a CTA button label, or the from-name or from-address in one email, and other emails in the same campaign still use the old value, the editor surfaces a suggestion chip prompting you to review and propagate the change.

Suggestion strings are English-only at launch; other languages follow through the standard localization process.

After a campaign is generated, a row of edit suggestions appears above the Active Intelligence composer. The row shows a reviewing state while suggestions are being worked out, then surfaces up to 3 suggestions. Each suggestion notes whether it came from your brief, your email, or your account data, and a click sends it through just as if you'd typed it yourself. If there's nothing worth adding, a single neutral line appears. If analysis doesn't complete, the generation experience is unchanged.

This feature is available to accounts where it has been enabled.

The "Match my brand" action has been removed from the action bar. Brand import is reached from the pill above the email, shown only to accounts that don't have a brand kit yet. Repainting an email onto a brand is available by asking for it in chat.

No action required.

Three improvements to the review-and-send panel in Campaign Agent:

Physical address picker. The "Add an address" action stays visible and disabled when it can't be used, with a tooltip explaining why: admins-only accounts see "Only admins can add a sending address," read-only campaigns show "This campaign is read-only," and scheduled campaigns show "Unschedule this campaign to add a sending address." Previously the action disappeared with no explanation.

Recipients button. When all lists are selected, the button shows "All lists" instead of listing every list name.

Send timing note. The predictive send timing description reads "Each contact receives the campaign at the time of day they are most likely to engage."

The bulk automations list sorts by newest first, so your most recently created automations appear at the top.

When removing automations, each one shows its status so you can see its current state before confirming removal.

In the automations overview, the overflow chip popover closes correctly when you move the mouse off it.

When you open the manage-contacts view for an automation that hasn't had any contacts enter yet, you see an illustrated empty state instead of a blank list. The paused queue banner has also been restyled for visual consistency.

If you try to activate an automation while Active Intelligence is still generating one of its emails, you see a plain-language message telling you the email is still being prepared and to try again shortly. Previously, a raw key identifier appeared on screen instead of a readable sentence.

You can ask Active Intelligence how your WhatsApp credits were used and get a direct answer in the chat. Type your question in the Active Intelligence chat and the breakdown appears inline.

In If/Else conditions and other segment filters, list pickers previously showed only the first 20 lists. Lists beyond that were hidden with no indication more existed. The pickers page through all your lists, so every list is reachable regardless of how many you have.

The to-do and completed states of the onboarding checklist are translated into additional languages.

When Active Intelligence can't fulfill a widget request, the message names Active Intelligence directly and points you toward supported alternatives rather than listing broad topic areas.

One-click recovery prompts have a localized heading. Raw technical decline details no longer appear in the interface.

When you attach a campaign or template using the @ menu or "Add context" button, Active Intelligence reads the subject line, preheader, and body copy inside it and writes from that content. Previously, attaching a campaign told Active Intelligence which one you meant but never showed it the words, so it answered from the name alone.

Several related improvements shipped together:

• Attached copy reaches the model. For a campaign, the subject, preheader, and body copy are read. For a template, the words inside its markup. Both are capped at 4,000 characters, and up to three assets per message are read at once.
• Attaching one campaign and one template at the same time works correctly. Previously, the template was silently discarded. The template governs turns where you haven't stated a specific operation; an explicit copy, split, or edit still lets the campaign take precedence. If both can't be honored, Active Intelligence names the one it couldn't use.
• Attached documents survive when you reopen a conversation. A document chip previously lost its connection when a conversation was reopened — it still appeared, but Active Intelligence no longer had the file. The connection is preserved through save and reload.
• Asset names no longer run into your message text. Names were being inserted as plain text alongside structured data, causing them to run together with preceding words. On the campaigns index, a hardcoded English phrase also appeared for non-English users. Asset names no longer appear in the message text.
• The @ search tells you when results couldn't load or were cut short. If the search fails, you see "Couldn't load your items. Try again." If results are partial, you see "Some items couldn't be loaded, so this list is incomplete." If results are capped, you see how many are shown and how many exist, with a prompt to keep typing to narrow the list.

These improvements apply automatically when you attach a campaign or template.

Known limit: Up to three assets are read per message, each capped at 4,000 characters.

When you open a system alert from the Active Intelligence notifications center, Active Intelligence receives a troubleshooting prompt so it can provide relevant guidance in context. You get actionable next steps directly in the conversation instead of a bare alert with no direction.

Previously, results were cut off much earlier than expected — dropping to around 5 records once your account had 20 or more items of that type.

Complete results are visible up to approximately 95 records before any compaction occurs. If your account has fewer than 95 records of the type you're asking about, you'll see all of them.

No action required.

Per-variant results include each variant's subject line, sender, send share, winner status, and engagement rates (open rate, click rate, bounces, unsubscribes). The winning variant is flagged, and the split settings are included.

This works for all accounts. If the split data can't be retrieved, the rest of the campaign report is still returned — only the split section is omitted. A single-message split returns an empty variant list.

Several improvements apply automatically:

• Series names in Dutch, Portuguese, Spanish, French, German, Italian, Swedish, and Danish are recognized as welcome or onboarding journeys, so the duplicate-welcome warning fires correctly for those accounts.
• Numbered replies ("2", "3") and localized option labels (such as "Optie 1") are accepted as valid "create anyway" responses in confirmation prompts.
• Lead-nurture series are no longer incorrectly flagged as duplicate welcome journeys.
• Welcome intent is matched from your most recent message, so tool paraphrases don't get misread as your instruction.

No action required — these improvements apply automatically.

When you asked Active Intelligence to send to "everyone" or "your whole list," it could pick a list based on its name rather than its size. A list called "Master Contact List" is very often not the one that reaches anyone — across accounts with that kind of list name, the median size of the name-matched list was 1 contact, against 475 for the largest list on the same account. On thousands of accounts, the name-matched list reached zero contacts.

Active Intelligence now judges audience breadth by actual subscriber counts. It also reads all pages of your lists, not just the first 100, so accounts with many lists get the full picture. For segments, it reads up to 1,000, covering 92.7% of accounts with more than 100 selectable segments.

If a narrowed audience doesn't save correctly after a campaign is written, Active Intelligence tells you what happened instead of reporting success. It distinguishes between two outcomes: a segment that came back empty (meaning the campaign reaches the full list) and a segment that came back as a different audience entirely.

When Active Intelligence proposes a plan with multiple changes, the redesigned card presents them as a visual stack: the current change is in front, with the next two visible behind it. As you answer each step, a receipt settles above the stack confirming your decision. Auto mode moves into the card header as a labeled toggle.

If an approved step depends on a step you skipped, the entire plan is refused and the blocking step is named so you know exactly what to address.

Known limit: A card created before plan tracking was introduced may stay pending if execution stops before any step runs. All cards created today include plan tracking, so this is recovered automatically.

Previously, starting from a past campaign preserved the layout but replaced the copy with freshly generated text.

Active Intelligence now carries both design and wording by default. If you ask to keep the wording but haven't pasted any copy in, Active Intelligence will ask for it. If you didn't mention wording and haven't pasted any, the design is reused and the copy is refreshed. The reply always tells you what actually carried over.

Starting from a template is unchanged — templates always generate fresh copy. No action required.

Previously, approving a per-item task — such as tagging contacts — that covered more than 100 records caused the entire step to be refused. Your approval was consumed, nothing ran, and re-approving led to the same result.

Active Intelligence now executes the first 100 records in order. The outcome card tells you how many were attempted, how many remain, and the total — for example: "over the first 100 of 250 records; 150 more remain beyond this run." You can continue from there without losing your place.

No action required. If a task covers more than 100 records, Active Intelligence processes the first 100 and reports the remainder automatically.

The in-chat approval card includes an Auto mode toggle. When Auto is on for a chat, Active Intelligence can act without asking for approval during that session only. Starting a new chat returns to your standard Customize permissions.

The toggle label reads "Auto." The Tool Tip reads: "Lets Active Intelligence act without asking in this chat only. A new chat uses your Customize permissions."

Note: Non-English locales currently display these controls in English. Translations are coming in a follow-up.

Active Intelligence send settings, scheduling status, and layout-selection prompts are fully localized. Non-English users were previously seeing English text in several places: audience-change and design-edit warning dialogs, the address picker empty state, send-time status labels, and the message the canvas sends when you open the Layout panel.

The layout-selection fix also corrects an issue that affected every locale, including English: without the translation key registered, the raw key string was being sent to Active Intelligence as your message instead of the actual request for layout options, causing incorrect responses.

No action is required.

When Active Intelligence finishes analyzing your campaign and has nothing to add, it shows a clear message: "Your campaign looks good — give it a look." The message encourages you to review your campaign, then disappears after 10 seconds. Previously, the analysis could finish with no visible feedback at all.

You can pick a Dropbox file directly from the personalization tags modal in the email editor, under the APP group. A Connect button loads the Dropbox file picker; selecting a file inserts its link into the editor. Images keep their preview link; other files are rewritten to the direct-download address. The link is inserted as plain text.

The email editor toolbar includes a desktop/mobile view toggle. Switch between viewports while editing to see how your email looks on each. The toggle defaults to desktop; clicking the active view again doesn't deselect it.

When you exported deals from the "All pipelines" view, the export ignored your active filters and exported every deal instead. The export applies the same filters you see on screen, so what you export matches what you're looking at.

When you refresh a dashboard widget, its subtitle correctly reflects the data window you're looking at. Previously, a subtitle correction you made would silently revert the next time you refreshed — the refresh path wasn't applying the same validation as widget creation. That gap is closed, so your subtitle stays accurate.

This was observed across all accounts in live testing.

Known limits: Rebuilding data sources on the refresh path is separate work not yet shipped, so some refresh behavior remains unchanged beyond the subtitle fix.

Three improvements to contact search shipped together.

Saved searches are in the search input. When you focus the main contacts search input, a dropdown opens showing your saved searches under the heading "Saved Searches." Click a row to load that search. A delete option appears on hover or keyboard focus and opens the existing delete confirmation. The split button beside Advanced Search is gone.

A result count appears after a search runs. Once results load, a count appears beside the table actions — for example, "Edit all — 101 results." This shows for basic search terms and advanced search or segment views. Filter-only views (Tags, List, Status) are unchanged. Large totals are formatted for your account's locale (for example, "8,450" in US English). The count doesn't appear while a search is in flight, so it always describes the results you're looking at.

Basic search is available again in advanced search results. When a segment is active and the advanced search panel is open, the basic search input is available as a plain input. A previous issue where two search inputs could appear at once has also been resolved.

The activity feed on a contact record correctly labels and displays deal-related emails and several other activity types.

Deal emails appear as "Email Received," "Email Sent," "1:1 Email Sent," "1:1 Email Opened," or "1:1 Email Link Clicked," with the email subject, attachment count, and a 90-character excerpt. The activity badge uses the right color and icon for each type.

Eight activity types that previously showed as generic entries now display real titles and detail: automation blocks, contact restores, ecommerce orders ("Made Purchase" / "Abandoned Cart"), contact enrichment, Facebook audience events, WhatsApp conversations, WhatsApp flows, and omni-channel link data.

The activity card caret is restored to its previous appearance. The Reply button has been removed from deal activity detail panels — replies to deal emails start from the emails tab, which is unchanged.

The Initializing step in the campaign sending progress bar reflects what the sending system actually reports.

• In progress — shown only when the system confirms initialization has started.
• Not started — shown when nothing has been reported yet, including scheduled campaigns, campaigns awaiting review, and campaigns without progress data.
• Done — shown once initialization finishes or the send moves to a later stage.

"View details" availability is also smarter by campaign type. One-time campaigns and A/B splits keep their existing behavior. Campaigns that send on their own cadence — RSS, auto-responders, date-based, recurring, text, series, and special campaigns — no longer offer "View details" while scheduled; the link becomes available as soon as the sending service has progress data.

When "View details" isn't available, the sending-progress chip shows the same hover tooltip as the status badge (for example, "Sent to 1,000 contacts 2 days ago"). When "View details" is available, no tooltip is shown on the chip.

Several issues with the campaign preview action bar in Active Intelligence are fixed:

• The start-source picker opens as a full popover above the email rather than clipping into a narrow band of cards.
• The picker's title names the tab you're viewing. The preview action is called "Reuse a design."
• Cards scale to the available width instead of a fixed column count, and card height matches other template pickers in the product.
• Card edges meet the 3:1 contrast threshold required for UI boundaries.
• Two action bar icons that rendered as blank space are replaced: the brand action uses the Active Intelligence mark, and the reuse action uses the email-template icon.
• The "Change layout" entry is removed from the bar. You can still ask for a different layout in conversation.
• Actions taken from a campaign preview stay in the conversation they were started from and apply to the correct campaign.
• The list of past campaigns and templates loads more as you scroll, replacing a "Load more" button.
• Only one panel covers the email at a time: opening the picker closes any open send-settings pill, and opening a pill closes the picker.

These fixes apply automatically.

Previously, clicking an automation link in the activity feed opened the automation builder without showing the contact's path through it. The run wasn't highlighted, leaving you to find it manually.

This applies to both the activity feed and the activity detail panel. No action required.

This applies from the contact record page. Clicking the checkbox or its label text enables the Archive button and lets you complete the action.

When Active Intelligence detects an existing welcome series and warns you, you can confirm "create anyway" in the same conversation and the automation is created. Previously, two separate issues blocked this path — the warning wasn't retained across conversation saves, and the confirmation signal was fragile — so "create anyway" often failed silently.

Both are resolved. If you start a fresh conversation with no prior warning, Active Intelligence still warns you before creating a duplicate. Typing "create anyway" without a prior warning in the same conversation also still triggers the warning, as expected.

Previously, if a lookup returned several matches and your target wasn't first in the list, Active Intelligence could answer with data for the wrong record. For example, asking about a tag called "CM - Fund Manager" could return information about a similarly named tag that appeared first in the results.

Active Intelligence now checks that the record it selects actually matches your request. If the first result isn't the right one, it finds the correct match. It also refuses to act on a record it can't confirm — for example, if it would have to guess or fabricate an ID rather than look one up.

These checks run at three points: when planning, before you approve, and before any action is taken.

You can connect your SurveyMonkey account directly from the personalization tags modal in the campaign editor, under APPS → SurveyMonkey. After connecting via OAuth, you can browse your surveys and insert a personalization tag for any survey. A disconnect option and a refresh button to reload your survey list are available alongside the connection.

SurveyMonkey tags appear as green chips with a survey icon in the editor canvas. If a campaign already contains a SurveyMonkey tag when you open it, the tag is recognized automatically.

Previously, timestamps were shifted by the difference between your local time zone and US Central time — affecting individual message timestamps, day groupings, and Today/Yesterday labels in the conversation list. All timestamps now use your account time zone, with fallback to your local time zone.

Previously, the summary cards used deliveries as the denominator, but the individual broadcast report and table rows used sends — so the same metric could show different numbers depending on where you looked. All three views now use deliveries, matching SMS industry convention.

Delivery rate and fail rate remain calculated on sends. This change applies to the Broadcasts report only.

Searching by list name in the message variables page's list dropdown works correctly. On some accounts, lists with an empty name or missing identifier caused the dropdown to stop filtering entirely.

Previously, the filter only showed folders that already had at least one form in them, so empty folders were invisible and couldn't be selected. Empty folders appear with a "(0)" count. Within the Folders group, order is initiatives first, then alphabetical by name.

The same fix applies to the Campaigns folder filter, so selecting an empty folder by URL works correctly there too.

The gear menu on custom campaign template cards opens next to the gear icon where you'd expect it, and you can reach it with a keyboard.

Button labels are centered across the platform, including on template cards, the personalization popover, address blocks, paginated dropdowns, the Deals CRM pipeline view selector, forms, reports, campaigns, goals, SMS, and settings.

When Active Intelligence builds a WhatsApp flow and the platform returns a fixable validation error, it retries the correction internally instead of surfacing the error to you. Recoverable issues — including empty contact fields, missing named steps, bad template bindings, and structural errors — are handled silently in the background.

On the last retry attempt, if the issue still can't be resolved, Active Intelligence asks you for guidance rather than burning the final attempt on a guess. If all retries are exhausted, you receive a plain summary with no internal technical details.

In the 30 days before this change, 14 of 24 created flows went through at least one visible rejection — up to 5 in a row. Those self-corrections are now silent.

When you asked Active Intelligence to create a segment and included a description, the description wasn't being saved — Active Intelligence reported success, but the description was dropped.

Descriptions are accepted and saved alongside the segment. They're validated against a 1,000-character limit (matching the segment builder in the product) and trimmed of whitespace before saving.

Known limit: Editing the description of an already-saved segment is not included in this change.

When you asked Active Intelligence to set a campaign audience and the audience ID was handled as text rather than a number, the audience was silently dropped — your campaign was created with no audience attached and no explanation.

Active Intelligence correctly retains the audience you specified, regardless of how the ID is formatted internally. Validation is unchanged: audiences that don't belong to your account are still rejected with a note explaining why.

This applies when the campaign send configuration feature is enabled for your account.

Key Findings were previously appearing as unlabeled continuation bullets under the Recommendations section, making it hard to tell the two apart.

Key Findings appear under their own "Key Findings" header, placed before the Recommendations section. The structure matches what Active Intelligence intends to show you.

Two specific situations now get their own clear messages:

• Data not supported: Active Intelligence explains that the requested data isn't available and offers three alternative widget ideas. Selecting one fills in the prompt and submits it immediately, so you can get to a working widget without starting over.
• Data not ready yet: Active Intelligence lets you know the data exists but isn't available yet.

Other situations — including connection issues or unrecognized errors — continue to show a general message. Previously, Active Intelligence would retry until it timed out, leaving you waiting through a long delay that ended in an error.

When Active Intelligence proposed a plan and you replied with a correction, it was re-presenting the original value instead of your corrected one. For example, correcting "Enrolled Course Category" to "Enrolled Courses" caused the next turn to show "Enrolled Course Category" again.

Your corrections are treated as authoritative: the latest message wins, and the plan is updated accordingly.

Previously, each language card stretched to fill the full panel height, meaning only one card was visible at a time and you had to scroll a full panel-height to reach the next language.

The 1:1 email composer on a contact's page sends email as expected. The Send button enables once domain verification completes, so you can send directly from a contact record without interruption.