With two decades of experience in business and operations, Teresa Lechuga doesn’t have a background in engineering, coding, or technology.
Despite that, she’s built an entire website and brand exploring how AI agents, automation, and human workflows can evolve together. On Cultivated Cure, Teresa publicly documents her personal experiences, hypotheses, and experiments testing AI systems to solve problems and create tools and systems to make her life easier.
Why? Because she’s curious.
“I was always curious in how things worked, and AI was no different,” she shares. “I wanted to know what I could build, what AI was capable of and where human judgment still mattered.”
Approaching AI with an experimental mindset
What’s the best way to get started with AI? Experimentation, says Teresa.
While Teresa had been using work-sanctioned AI systems in her day job, she wanted to take the technology a step further, and had some real-life problems she hoped to solve using AI.
“I came into it the way most of the interesting turns in my career have happened,” she writes on her website. “Something amazed me, and I immediately went to go break it on a real problem.”
She created the AI Lab (no relation) section on her site where she conducts experiments and outlines what she built, the exact tools she uses, the professional build cost versus the cost using AI tools and lessons learned.
Before Teresa begins, there are three questions she asks to determine if a problem is a good candidate for AI experimentation:
- Is there friction?
- Is this a process that can be repeatable?
- Can it be measured?
“Whenever I had a situation that I was thinking about AI for, if it produced a faster result and outcome, and the throughput met the demand, then that's when I would say, ‘okay, we have a case for AI here,’” she explains.
For AI newbies, Teresa recommends picking a platform like ChatGPT or Claude and complaining to it. (Venting encouraged here!)
“That’s the easiest way because it’ll naturally want to make it better,” she says.
This is exactly where Teresa starts each new experiment: brainstorming — literally having a conversation with AI — about problems she’s hoping to solve.
These conversations are often long and in-depth, and when Teresa doesn’t get the result she’s looking for, she presses AI by continuously asking more and better questions. In these exchanges, Teresa also often pushes back on the LLM she’s using, reminding them, “Hey, I’m not a developer. Start from square one. Don’t assume I even know where to click.”
By doing this, the platform really gets to know Teresa; her abilities, her weaknesses, opportunities to learn and more. This continuous conversation has been a big part of why Teresa has been successful in completing many of her experiments with no technical background.
A peek inside Teresa’s AI experiments
Here’s an example of a practical experiment Teresa has conducted and documented.
Creating a Gmail organization agent with Claude to handle email overload
The background: Like many of us, Teresa has had a Gmail account for years and years, leading to an overwhelming number of emails to manage.
“My emails were in the thousands,” she says. “It was tons of spam and ads, but I didn’t want to get rid of it because it’s the hub of where a lot of my emails go.”
Teresa set out to create a personal productivity agent for email triage, categorization and automated response workflows.
The pain point: Sifting through thousands of irrelevant emails to find the messages that actually mattered.
The workflow: Teresa was venting to Claude about her overflowing inbox, and began to work toward a solution. The logic Claude recommended was reviewing Teresa’s inbox every day at 5 p.m. and sorting the messages into set categories and folders.
Together, they also developed guardrails to help Claude learn what an important email is versus one that is not. Or to help determine when to delete emails after an appropriate amount of time like how long after traveling to delete plane tickets or how long after coupons expire before getting rid of them.
The initial setup took a few hours, and together, Teresa and Claude honed the process over the span of a week.
The human layer: AI is not perfect, and Teresa regularly trains and re-trains Claude when she finds an email in the wrong place.
“The corrections feed back into how future email gets handled,” she says.
And, when it comes to generating draft replies, nothing is sent without a human reading it first.
The end result: A streamlined, tidy inbox. “I don’t have to go through and delete thousands of emails,” she says.
Learn more about this experiment.
In addition to the email inbox experiment, Teresa has also documented projects, including:
- An AI-assisted coordination system for a ranch house restoration.
- A trip planner, creating a personalized brochure for a family getaway.
- An operations workflow architecture to assess if a process is on track.
- Creating an autonomous agent to help shape a strategy and trade crypto on her behalf.
How experimentation fits into the bigger picture: building systems to solve problems
Teresa’s overarching philosophy: we're not replacing humans, we're augmenting them.
This ethos guides the way Teresa conducts experiments. She says it’s more impactful to use AI to build systems to solve problems, rather than simply asking AI to do something and trusting it will accomplish exactly what you ask.
In other words, it should be used as a tool to amplify human ability
“At first when I was implementing AI, I was telling it, ‘Okay, I want you to do this for me,’” she explains. “AI can be problematic with prompting and making sure it's producing the same type of output.”
By using AI to build systems, rather than simply to complete tasks, this adds an additional lens — a critical human layer — to the process.
“The most powerful use of AI isn't asking it to think for you. It's learning how to think with it,” she says.
Plus, creating guardrails and governance is incredibly important to Teresa when using AI; it’s something she regularly chats with AI systems about before beginning any experiment.
You have to be very thoughtful of what [AI] is telling you to do because it can be wrong. You have to be very much still in control, and there has to be that human judgment level that’s still very much involved when it’s directing you to do something.
The future isn’t AI or human; it’s both
In the ongoing (tired) debate of “is AI going to replace humans?” Teresa has a firm answer: no.
For Teresa, AI amplifies human knowledge and allows us to accomplish tasks much more quickly and efficiently.
“Working side by side with AI makes us faster and better,” she says. “It augments us in that we're able to obtain that knowledge faster and those resources faster, so that we can produce more.”
That's the message Teresa wants to get across: AI is not replacing humans; it's simply augmenting and amplifying our abilities.
