Designing and building with AI: from idea to product in hours
2026
Project overview
In 2025 and 2026, the line between design and engineering is getting thinner. AI tools have made it possible for designers to go from idea to fully functional product without writing a single line of code. These three side projects explore that space: each one was conceived, designed, and built entirely with AI assistance, in a matter of hours.
This is not about replacing engineering. It's about understanding the full product lifecycle well enough to move fast, prototype with real products instead of static mockups, and close the gap between design intent and technical output.
Tools & workflow
Every project followed the same process. Ideas and PRDs were defined with Claude, designs were created in Figma and explored with Google Stitch, and code was generated using Claude Code, Cursor, GitHub Copilot, and Antigravity. No code was written manually.
This workflow allows me to go from a design concept to a working product in a single session, test real interactions instead of prototypes, and iterate based on how the product actually behaves.
A daily UX tip tool designed for both students learning the fundamentals and professionals who want to stay sharp. Every day a new tip appears, with a quiz section to test understanding and an archive of the last 7 days for anything you missed.
The idea came from a simple observation: UX theory is vast but rarely revisited after you learn it. Short, daily doses make it more digestible and easier to retain over time.
A documentation example showing how a design system could be presented and explored interactively. Components are displayed with their variants and states, making it easy to understand how to use them and how they behave.
Having worked on design systems at both Luzia and Zurich Insurance, I wanted to explore how the documentation side could be made more engaging and interactive for the teams consuming it.
A fully functional digital version of the classic dice game, built as an experiment in a category I've always enjoyed but never worked in professionally: gaming.
The current version supports one player and follows the original rules. Future iterations could include multiplayer, score history, and UI improvements.
Learnings
Speed changes how you think. When a product takes hours instead of weeks, you explore more ideas and kill bad ones faster.
Real products beat prototypes. Testing an actual working product reveals interaction problems that Figma never would.
AI doesn't replace design judgment. Knowing what to build, how to structure it, and what problems it solves still comes entirely from the designer.
This workflow has limits. Complex products with real users, data, and edge cases require engineering expertise that AI tools can't replace yet. But for side projects, MVPs, and rapid validation, it changes everything.