An MVP is not a small version of the product. It is the smallest thing that tells you whether the idea is worth building. Often a clickable front end is enough. People can try the flow, you can watch where they hesitate, and nothing expensive has been built behind it.
Look first
Almost every idea sits close to products that already exist. A quick benchmark shows you which patterns people already understand and where competitors get it wrong. It also saves you from redrawing something that was settled years ago.
Week one
- I pull real screens and flows from proven products with Mobbin MCP and note what each does well.
- I pick the one flow that proves the idea. Everything else waits.
- I draw the user flows: the happy path, and the most important thing that can go wrong.
- I set up a small design system: tokens, a few components and a few patterns. Not a full library, but enough rules that the screens match.
- I design the screens in Figma using that system. Claude Code reads the real components through the Figma MCP, so the design and the build stay aligned.
- I show half-finished screens to a real person. A short conversation now saves a rebuild later.
Week two
- I build the front end from the Figma designs with Claude Code.
- I make the core flow clickable from start to finish, with believable content.
- I test it with a few people and write down what I learn.
What you get
A competitor benchmark, the user flows, a starter design system, the Figma screens and a live coded prototype.
What it is not
It is not a finished product, and it can look more real than it is, since there is no backend behind it. That is the point: you can test the idea cheaply. If it works, the same design system and code become the start of the real build.
All writing