Designgineer

Over the past year, I've had the opportunity to design and deploy several initiatives with AI as part of my workflow, and one of the biggest things I've learned is that the real power of AI isn't just speed. It's how much further I can take an idea before I ever need to hand it off.

1. 1

   Concept with AI

   Explore ideas with AI, most often in Claude Design. Sometimes from a few screens, a rough flow, or sketches on my iPad.

2. 2

   Pick the path

   Complex feature

   Build it out in Figma, with more intention and control.

   Simpler, system covered

   Keep iterating directly with AI.

3. 3

   Refine together

   AI asks questions, I answer them, and we keep refining until the design feels right.

4. 4

   Branch in GitHub

   Give AI context on where the work needs to live.

5. 5

   First pass of the code

   AI reads the existing stack and patterns, and drafts the implementation.

6. 6

   Audit the system

   Make sure it uses the real design system components, not lookalikes. An MCP makes this easier.

7. 7

   Test and refine

   Breakpoints, interactions, and edge cases. An experiment can be set up in GrowthBook in the same pass.

8. 8

   Partner with an engineer

   Before the work crosses the finish line: endpoints, architecture, and better ways to build it.

My process usually starts by partnering with AI to conceptualize ideas I may not have thought of on my own. Most of the time, I'm doing this in Claude Design. Sometimes I'm starting with a few screens I've already created, a rough flow, or sketches from my iPad. From there, I riff with AI to explore different directions, concepts, and angles that can push the idea further.

The design usually takes one of two paths depending on the complexity of what I'm working on. If it's a more complex feature or project, I'll take those ideas into Figma and build out the experience with more intention and control. If it's something simpler and I already know the components, patterns, and design system are accounted for, I'll usually keep iterating directly with AI.

I've found the better the context I give AI, the better the outcome.

Once I'm happy with the direction, that's where the workflow starts to feel very different from how I used to work.

Instead of stopping at the handoff, I create a new branch in GitHub and give AI context on where the work needs to live. At that point, it can review the existing tech stack, understand patterns already in use, and start producing a first pass of the actual implementation.

That part is incredibly powerful, but it's also where I've learned to slow down and pay attention.

One of the biggest mistakes I've seen AI make is recreating something that looks like a design system component instead of actually using the real component. Visually, it can look almost perfect while being completely disconnected underneath. So I audit the code, challenge the implementation, and ask it to verify that the components being used are actually tied back to the design system. When an MCP is available, this gets much easier because AI has access to the real atoms and components instead of trying to mimic them.

From there, it becomes another round of refinement. I review the experience across breakpoints, test the interactions, look for edge cases, and keep iterating until the implementation matches the expectation I had for the design. If I'm working on an experiment, I can also point the work toward something like GrowthBook and configure the test as part of the same process.

This is also where partnering with an engineer becomes critical.

Even when AI can get the implementation surprisingly far, I don't see that as a reason to remove engineering from the process. Quite the opposite. Bringing an engineer into the work gives me another level of context and understanding that AI alone can't replace. They can challenge assumptions, help me understand why certain patterns exist, identify technical risks I may not see, and teach me better ways to approach the implementation.

That partnership has become one of the most valuable parts of the process for me.

I may be able to take the idea much further than I could before, but I still want an engineer involved before the work goes across the finish line. There may be endpoints to connect, architectural decisions to make, implementation details to clean up, or simply better ways to build something than the approach AI produced.

The goal isn't to bypass engineering. It's to show up to that partnership with more of the work already explored and a much deeper understanding of how the idea actually becomes a product.

For me, that's what has made AI such a powerful partner in design. It hasn't removed the need for design judgment, collaboration, or technical expertise. It has given me the ability to carry my judgment further into the actual product and learn more along the way.

And I don't think designers should wait for their company to officially make space for AI before they start exploring what's possible. Take it upon yourself. Start with something real you want to build. Don't worry about knowing every step. Ask questions, experiment, and learn where AI can help you move faster and where your own judgment still needs to lead. The biggest thing I've learned is that you don't understand the power of AI by watching someone else use it. You experience it by doing.

That's how you find your aha moment.

Don't wait for your company to make space for AI. Take it upon yourself. Become the ambassador of AI by doing the work.

Somewhere along the way, I have been transformed into a Designgineer. A title coined by my EVP of Experience, Seth Taylor.

Start with the thing you wish you knew how to build. Give AI the context. Ask questions. Keep going. Somewhere in that process, you'll find your aha moment, the point where you realize you're capable of taking an idea further than you thought possible.

See it in the work

• Leadership storyPassword protected

  Pay It Forward

  The DAVID page I built twice, and Community Impact, which I designed and developed.

• Design case

  Photon

  The design system I started at Angel, across web, mobile, and TV.

• Leadership story

  Communication is king

  Using AI in real product design work is one of the four things I hire for.