How to use AI in product design: Lessons from building VibeLab at Miro

Mauricio Wolff at Miro built VibeLab, an internal platform where the whole design team turns ideas into working prototypes with Claude Code in minutes rather than days.

Miro makes an online whiteboard for teams. Mauricio Wolff, a staff product designer there, built VibeLab, an internal platform where designers use Claude Code to build prototypes that look and behave like the real product. Since its launch in March 2026 it has reached more than 400 users and nearly 1,000 prototypes.

• A single copied command installs about 25 tools and the Miro credentials a designer needs, so designers without a development background can start prototyping.
• Early prototypes drifted from Miro's design system, so the team added templates built on it. Designers now branch from a template and send improvements back to it.
• Prototypes are shared only on Miro's internal network, to keep ideas off the public internet. A shared key lets a prototype call AI models without each team requesting its own.
• Prototypes do not go straight to production at Miro's scale. Engineers and their agents still translate them into production code.

A design team that starts every prototype from a shared template built on its design system can keep fast prototypes looking like its product, as Miro found. The article is a Miro designer's account of an internal tool and ends by promoting Miro's own product.

Derived by Working Surface from the article; more in the Playbook. Source line: You can delegate everything to a model, but you'll get a very average experience.

21 October 2026: as AI tools multiply code, documents and prototypes, the work that counts moves to judging, reviewing and keeping track of what they produce.