I had a design process diagram. You've seen it (or one like it). Three phases. Nine stages. A clean, sequential march from Discover to Design to Deliver. Kick-off meetings. Brainstorming sessions. Empathy maps. Wireframes. Prototypes that were really just clickable slideshows. User testing with those clickable slideshows. Polish. Hand-off. QA. Ship.
Each stage had its own set of tools, its own rituals, and its own weekly meetings. A full cycle could take months. And honestly? It worked. It was thorough. It forced you to think before you built, to validate before you polished, to document before you handed off.
I even gave it a catchy name (D³) and created a lot of detailed artifacts to help onboard new designers and aligning the team. The basic process should look familiar...
This process served me well as I was growing the Design and Research team at Mynd, and creating the Product & Design function at Friendly. Just as it had served tens of thousands of high-functioning design teams worldwide.
Except that this year I found that I just wasn't following it anymore. Something fundamentally changed.
I got my hands on Claude Code.
Just take a look at all the tools and activities that were part of the process, helping generate ideas and refine priorities and divining which feature would move which KPI needle the most.
So many meetings, research sessions and tech consultant speak: How might we (HMW), Kano model, RICE, Benchmarking, Impact v. Effort matrix, UX Copywriting Jams, Bug bashes
This was weeks and months of, essentially, educated guesswork that would only be validated once we got an actual working implementation before real users in the final Measure phase.
I noticed how for a lot of our design initiatives, the process changed from:
Discover ➡️ Design ➡️ Build ➡️ Test
To something that looks more like:
Discover ➡️ Build ➡️ Test ➡️ Design & Build Properly
In the old process, you designed before you built. Obviously. You researched, wireframed, prototyped in Figma, tested with clickable mockups, polished the visuals, and then handed a spec to engineering. Design came first. Code came second. That was the whole point of the process.
Now? You skip Figma entirely in the ideation phase. You describe what you want to a tool like Claude Code, and it builds a working application. Not a mockup with "lorem ipsum" and fake buttons. Actual software that does actual things. With real interactions, real data, and real state management.
That means you can put a functional prototype in front of users (or better yet, into their daily workflow) within days of having the idea. No wireframe stage. No visual design stage. No "imagine this button would take you to..." disclaimer during testing. People use the thing. You watch what happens.
And only then, when you know for sure that this is something users need, they enjoy using it and it makes significant contributions to your KPIs, would you start design and development.
The design team whirls into action turning required screens and components into something that matches the brand and Design System. At the same time the backend & data teams take the working, verified prototype and start implementing it into the code base in an optimized, secure, and enterprise-ready manner.
And once the design team finishes, the front-end team will make required updates to the components. Voila!
PS This is partially why I'm starting to believe that Figma missed the boat and is more or less cooked. But more on that later.
