CarloFrinolli.it - Logo
Home Page
About Me
Journal
CASE STUDIES
CONTACT

FAQ

**Is the prototype code I can keep?** Sometimes, but it is built to learn. We decide together whether to grow it or rewrite it.

Formats

5 hours
Sprint
To test one specific hypothesis.
2 days
Workshop
For a team that wants to learn the method on a real project.
Over time
Ongoing support
For teams that want to build this way of working into their process.

The problem

Design sprints work when the problem is already defined, and lean works when you can measure. Many projects start on moving ground instead: the development team is at capacity, agencies cost too much, and the idea stays stuck in a Figma file no user will ever see.

Proof it works

Stakko, a streaming search engine for Italy that covers around 80 platforms, I built with Claude Code without going through Figma. Findry, a visual inventory for objects and boxes, started the same way.

When it is the wrong tool

If the people deciding want a demo to approve rather than an experiment that can say no, the sprint turns into theatre. The same goes for when the problem is already clear and needs production engineering: that calls for developers, and a quick prototype risks becoming the code everything else gets built on.

Got an idea stuck in a file?

How I work

I use OPUS, a four-stage cycle. We question our certainties, separate what really matters, build deliberately rough prototypes that teach us something, and keep what we learned. AI tools speed up every stage: Claude or ChatGPT to challenge assumptions, v0, Bolt or Lovable for working prototypes, Claude Code or Cursor when a real MVP is needed, Ekitai for landing pages in a few hours.