Composal

Software teams need to verify that critical user flows work before shipping, but manual testing before every release is slow and labor intensive. Developers currently review code changes directly but lack an easy way to systematically explore their product for edge cases and failures across real user journeys. This manual process slows down release cycles and risks letting bugs reach users.

Composal connects to a team's product URL and sources of product intent such as GitHub, Sentry, PostHog, and Notion to map out the product. It then deploys a swarm of browser agents that explore the product, test key flows, and uncover edge cases in parallel, simulating real scenarios like booking flows and payments. For each test, Composal produces a video, logs, and reproduction steps showing what worked and what broke. These reproduction steps can then be passed directly to a coding agent (via a CLI) so it can fix the issue and learn from it.