Hey Product Hunt 👋 I kept seeing the same thing at every engineering team I worked with: ship velocity went up 5× in two years, and the QA stack stayed stuck in 2018. Brittle Selenium scripts. Playwright fixtures nobody owns. A CI suite everyone learned to ignore. Coverage shrank silently as the codebase grew. We built Probe because QA shouldn't be something you maintain — it should be something that runs itself. Probe is one autonomous agent that covers your web, mobile, and APIs end-to-end: it discovers surfaces on its own, generates tests in plain English, runs them on every commit, and triages regressions while you ship. No selectors. No fixtures. No DSL. Verified bug reports — with screenshots, traces, and reproduction steps — land straight in GitHub, Linear, or Slack. The hardest part wasn't building the agent. It was building one whose bug reports you'd actually trust. We threw out our first three architectures before the answer clicked: less prompting, more deterministic tooling, tighter feedback loops, and a planner that explains exactly what it's testing and why. Two things that are still surprising us: 1. Most regressions Probe catches are flows nobody had a test for in the first place. 2. Teams that adopt it stop writing tests within a week — they just describe surfaces. If your team ships faster than your tests can keep up with, please try a run on your own product and tell us what broke — or, hopefully, didn't. We're reading every comment today Hritvik & the Probe team

Probe

Agentic testing for teams shipping AI native products faster

Agentic testing for teams shipping AI native products faster