AI Tester

An autonomous end-to-end testing service. You describe a test in plain words — “log in”, “check that the home page loaded” — and an AI agent runs it in a real browser: it understands the interface visually and doesn't break when the markup or selectors change.

Tests are assembled in an editor from three kinds of steps: act, assert, and screenshot. A run snapshots the steps, then a worker executes it in the browser, recording video plus a verdict and a reason for every step — you can see exactly where and why it failed.

Structured like a grown-up tool: workspaces with members, projects with a base URL, run history across environments.

Stack

Next.js, tRPC, Drizzle, and Postgres; execution — Playwright driven by a multimodal model; auth — Better Auth.