Designing with AI: Inside One Team’s Workflow
Interview with Taylor Palmer, Founding Designer at Stable
Everyone’s talking about AI. And while it is important, a lot of early-stage teams are getting distracted spinning their wheels chasing the latest tools instead of focusing on what actually matters for their business.
Here at Startup Soup, Jen and I are big believers in the “good enough” approach: integrate new tools when they’re useful, but don’t let them hijack your strategy.
That’s why I wanted to talk to someone who’s doing this well, not as an AI thought leader, but as a practical operator.
My friend and former colleague Taylor Palmer is leading design at Stable, and in integrating AI into their design practice, he’s not just experimenting with new tools — he’s rethinking workflows, team dynamics, and how designers grow in a world where AI can handle the basics faster than ever.
So I asked if he’d answer a few questions for Startup Soup readers, and he graciously accepted. In this interview, Taylor shares:
How his team is using AI for discovery, synthesis, and prototyping
How his team members are using AI as a coach, not just an implementation tool
What’s working (and what’s not) as they experiment
How he talks about AI in hiring, and what kind of designer thrives on his team
If you’re curious what actually using AI in a design org looks like, this one’s for you.
TLDR; AI is transforming design teams but not replacing human designers, so Taylor’s hiring. If you’re an experienced product design generalist who wants to work on a thoughtful, fast-moving remote team that’s integrating AI in meaningful ways, check out the open role at Stable at the end of the post.
Now let’s dive into my conversation with Taylor!
Prototyping & Coaching with AI
Can you share a recent design problem where AI helped you move faster or see things differently?
In most circumstances, AI is making our team more efficient and faster rather than seeing things differently. For example, I’m working with a small group of customers in a closed beta environment, and after each of those calls, I synthesize what I learned and roll it up into a summary for the team. That would normally take me 10–15 minutes, but with our AI call recorder Fathom, I can automate it in just a few seconds, and then refine based on what I heard personally on the call.
Separately on the design side, I’m working on a complex automation system to allow our customers to create logical rules. Typically, when beginning to explore designs, I would reach for pen and paper or for Figma to start to explore interfaces. Instead, just using words and ideas, I can have ChatGPT (or whatever tool) spin up a quick app for me to start clicking through. Almost immediately, I can punch holes in my old ideas and get something much faster into the hands of my test customers or my engineers to provide feedback.
You mentioned using AI as a coach to me earlier. How are you structuring those interactions? What context do you give it, and what kinds of questions do you ask?
This is something I’m still exploring myself, but I’ve seen other teammates have success with this approach that looks like something like this:
Pass it documentation (like a PRD or similar) to give context about what you’re working on
Give the model a prompt to “interview” you to gather more questions in order to achieve a specific outcome
Using what it learned, help you construct that outcome (like a plan, a prototype, or similar)
Once it has this much context, you can also ask questions like, “what could I be missing?”, “how could I improve this plan?”, or “in what ways could we move faster?”
We’ve also started exploring the idea of how to represent the ideas of stakeholders and leaders within the “coach” context as well. For example, what would it look like if we had the AI interview each stakeholder about the concerns, goals, and context for the project before a designer even started investing in it?
What’s something you tried that didn’t work well with AI?
Plenty of things. To name one, I recently spent some time seeing if I could make some design updates in code with some of the early ideas I had. I pulled down our codebase locally and started having Cursor help me work through it. I pretty quickly got bogged down in the right way to architect a new component I was proposing. I realized that things were already becoming very inefficient. Even though I could have gotten it to the finish line, it wasn’t the best use of my time where I could more uniquely contribute to other parts of the project.
How the Team Uses AI Today
What are some concrete ways you and your team are using AI tools in your design workflow right now?
We’re trying to regularly and consistently push our approach and comfort level with AI in everything we do to try and make ourselves more efficient. A few of the more discrete and concrete ways would be:
Discovery plans: Using AI as a coach to formulate ideas about what methods to deploy and when.
Research synthesis: Once we’ve identified and gathered data, using AI to look for patterns, summarize, and help us pull out insights.
Prototyping: Generating quick functional mock ups of early ideas to align team members or test with customers.
What does your current stack of AI tools look like? What tool do you recommend someone start with?
As with any conversation about tools, I think it’s important to focus more on the need than the tool itself. What pain or inefficiency or problem is our team experiencing and how could the tools currently available help us solve it?
Admittedly though, we are in a bit of a backward state right now where there are more tools coming out than any single person can keep track of. So, following along with the community and checking in on how others are approaching their work is an important way to make sure our team is continuing to evolve.
Some tools we’re currently finding valuable:
Cursor: While engineers on our team are already significantly more productive as a result of Cursor, we’ve also onboarded designers to help them understand different parts of the code base and to also create quick prototypes or suggestions about ways to improve the front end.
Fathom: Our entire company records almost all our calls through this tool, which automatically transcribes and summarizes them. We’re also able to group similar calls together and start to ask the AI questions about patterns and insights across the calls.
ChatGPT/Claude: these are catch-alls to help with writing, synthesis, brainstorming, outlines, and anything else.
Collaboration + Communication
How has integrating AI changed how your team communicates both internally and across functions?
I think there’s a lot more for us to explore here. One problem currently on my mind is how we centralize and plug in our team members’ daily or weekly updates into one centralized system that can help us watch out for collisions or opportunities to collaborate more closely. For example, you could imagine getting a ping that says, “Hey, so-and-so is working on something similar to you. Want to connect with them?”
What new bottlenecks have emerged?
In many ways we’re all able to produce faster than we can consume. I’m still experimenting with my approach to feel that I can rely on tools to help me summarize and analyze heavier amounts of context without missing anything.
Leadership + Hiring
How do you talk about your team’s use of AI when hiring? What kinds of designers are excited by that?
We talk about this openly and frequently. I would hope that all kinds of designers are excited by this because we’re at a turning point in the industry where the future of product design is very unclear. While I feel confident the practice will remain, the shape and the specialties of it will look very different over time. I have pitched current openings on my team as a way to come shelter from that storm because we're actively working on figuring that out. We’re also always pushing on how we shift up team members’ responsibilities.
What signals should candidates look for if they want to join a team that’s serious about integrating AI?
I occasionally hear from candidates that their companies won’t let them use even the most basic conversational models, so that’s a pretty obvious red flag that those teams are going to have difficulty integrating AI even further.
What’s something surprising you’ve learned from talking to designers about AI during hiring conversations?
As AI gets better and better at the basic fundamentals of UI or layout design, the specialization of designers shifts to focus more on taste and expertise. Occasionally designers will show me their experimentations with AI that result in something higher quality than what’s in the rest of their portfolio, inadvertently revealing that they have subtly automated themselves out of their workflow.
In this case I have encouraged these designers to double down on showcasing how they differentiate themselves and highlight their specialties, where producing a barebones utilitarian UI is becoming too easy.
The Future of Design Work
What does a “design team of the future” look like to you? What roles shift or emerge?
I don’t have a crystal ball, but I do have a guess. I fully expect roles to flatten, as traditional Product Design and Product Management roles already have in many smaller teams.
Even further down the line I could see a general “technologist” role emerging of someone who has a solid grasp on business, strong user-centered principles, and good understanding of how to build software.
How do you think earlier-career designers should approach learning in this new landscape?
Make sure to learn the underlying foundations alongside the high-output AI tools. I anticipate many people and teams making the mistake of skipping past these foundations both in hiring and education, resulting in too many people building too many things without understanding why they actually work (or maybe don’t).
What do you hope remains unchanged about design work, even as AI becomes more central?
I hope we don’t lose the person-to-person, human centered aspect of what we do. There are many tools promising to automate user research entirely. So if you can eventually run an entire process end-to-end without ever talking to a single human, we’ve removed one of the most gratifying parts of the building process.
AI isn’t replacing designers anytime soon — but it is reshaping what the job looks like. As Taylor put it, “producing a barebones utilitarian UI is becoming too easy.” The challenge, and the opportunity, is in defining what makes your contribution uniquely human.
Whether you’re leading a team or growing as a designer, Taylor’s perspective is a reminder that staying at the forefront isn’t about using the most tools. It’s about staying curious, having solid fundamentals, and shaping your workflows to make the most of what these tools can offer.
👉 Taylor’s team at Stable is hiring a product design generalist.
I can personally vouch that Taylor is an incredible teammate, thoughtful leader, and all-around great human. Take a look if you’re looking to join an early-stage remote team and build alongside folks who are thinking deeply about the evolution of design.
Learn more and apply here
Thanks for chatting, Jean! Would love to hear ideas or questions from other teams too—we’re definitely all still figuring this out.