TestPilot App

Traditional E2E testing assumes stable DOM structure. AI-generated apps don't have one — a single prompt can rewrite your entire frontend and invalidate every selector you wrote. TestPilot navigates semantically instead. No scripts to maintain, no setup. What it does beyond happy-path clicking: Multi-role workflows — test flows that span several users (dispatcher assigns, contractor bids, admin approves) with isolated concurrent sessions Security checks — IDOR, unauthenticated endpoints, API replay, tenant data leakage Interactive mode — direct a live browser session by chat One-line embed — a floating QA button on staging so clients and PMs can run tests without a dashboard login Bring your own Anthropic API key. You pay raw inference cost (~€0.10–0.50/run) directly to Anthropic and set your own cap. No token markup, no run limits. Founder note: I built a multi-role field-services app on Base44 — dispatchers, contractors, admins, permissions, billing. Then I went looking for a way to test it and every option meant writing Playwright. So I built the thing I wanted.