Composal
Our mission
Give teams the confidence to ship and keep their products working as they evolve.
Coding agents have fundamentally changed how software is produced. Teams can build more and faster than ever. Confidence in that output becomes the bottleneck.
A change can pass tests and deploy successfully, yet still break a user’s journey. Someone receives an invitation before a release and accepts it afterward. Can they still join the correct workspace with the right permissions?
We’re building Continuous Agent Validation. Agents test important user journeys after each product change. They check that new features and existing flows still work as intended.
Our approach
Composal reads specs, product documents and team discussions. Agents use this context to learn how a team’s product should work.
Teams give agents a goal. The agents explore the product autonomously. They try different paths, roles and permissions. They probe edge cases and look for unexpected ways to break the product.
Agents learn as they test. They record how the product works and where journeys fail. Useful findings become knowledge for the next run. Agents retain that knowledge and use it to guide their checks as the product changes.
Agents capture failures with recordings and reproduction steps. Teams and their coding agents use this evidence to investigate and fix problems.
Meet the founders
Backed by 
Combinator