A rehearsal stage built from a theatre's own real scenery flats on rolling casters, with many sets standing ready and one chosen set being wheeled through a narrow inspection gate toward the main stage, the figures pushing it seen from behind, the whole scene drawn with wide empty margins on every side.

How I AI, Lenny's Newsletter

Why Reddit built its own AI prototyping tool | Blair Metcalf

Claire Vo with Blair Metcalf · 26 October 2026

Company story · prototyping · handoff · After the prototype

Takeaway

Reddit built an internal prototyping tool on its real components and account data, and sends code meant for production through a generated requirements document and engineers' review.

Summary

Reddit, the online community site, wanted designers and product managers to test ideas without learning Git, the system engineers use to manage code. Blair Metcalf, principal UX engineer at Reddit, showed the podcast host Claire Vo how his team built Reddit Proto for that purpose. The tool now serves more than 150 people in design and product and holds about 700 prototypes.

Key points

Reddit Proto hides Git and starts each prototype from Reddit's design system, so builders choose features, devices and user states instead of setting up code.
Reviewers can sign in with their own Reddit accounts, so a prototype shows real feeds and search results instead of placeholder content.
The team mapped every design token and component to production code, then refined that map from engineers' feedback on pull requests that passed or failed review.
For production, a Figma frame becomes a generated requirements document with acceptance criteria, then a draft pull request in a separate repository that engineers review.

Implication

A product or design team can keep fast prototyping and production apart: prototypes run on real components, and only reviewed code moves toward production. That reading is an inference and is not Metcalf's claim.

Plays

Design

Build prototypes from the production design system and let reviewers load their own account data, so that reviews discuss the idea rather than placeholder content.

Engineering

Keep a file that maps each design token and component to its production code, and update it after engineers pass or reject each pull request an AI agent writes.

Derived by Working Surface from the article; more in the Playbook. Source line: So basically you start with a map and then you try, fail, iterate, and you refine the MCP through feedback.

Source issue

27 October 2026: AI produces reliable work when a team writes down how its own system works and checks the output against something the AI did not write, from frozen rules to audits.