Last week, I had the opportunity to handcraft a mobile app prototype and I decided to actually enjoy every part of it. Months of management pressure to adopt AI and automate every possible piece of my work with AI had deprived me from the honest pleasure of actually designing something on Figma, brick by brick, frame by frame. Another team had been working on an MVP and faced some UX roadblocks: some critical flows and user journeys had been completely skipped based on a hyper-focus on delivery speed, and the gaps only became evident after the project was ‘almost’ done. So here I was, pulled into an ongoing project, close to the delivery date and with the goal of completing a series of unclear/undefined user flows.

The team had Claude Code, Claude Design, Cursor, Gemini, Google Stitch and GPT at their disposal, and they had been using these tools over and over, but none of the AI-generated ideas really nailed it.

So, again, here I was, happy as a kid opening a birthday gift. Why? because the environment had been primed for proper design. They had tried AI, and it didn’t satisfy anyone, so for this specific project, people were open for hand-made design.

And I loved it.

I can’t share much about the project just yet because the app has not been launched, but I’ll breakdown the process, mappings, decisions and how unclear ideas became proper tech requirements (plus I’ll add some pictures).