Hi HN! We are building an AI agent that can use a computer for software tasks. You can have it review and manually test all your pull requests. Here is a demo video of how it works: https://www.loom.com/share/71135bb2ef784ff593940fbe6fb42c4c?...

We realized the current technology is not yet capable of automating non-trivial software tickets reliably but is pretty good at manual testing code changes for bugs. On the other hand, we knew tools like CodeRabbit got interest from developers yet could only point out obvious bugs or style nits as they are limited by ‘reading’ the code.

Fume can use its own computer. If you’ve made UI changes, it can spin up a browser and click through edge cases. If you’ve made more functional changes, it can write and run test scripts to ensure there are no bugs.

We are language/framework agnostic. The setup is pretty easy. You just install a Github app and set up Fume’s computer (once) like you would set up your local environment.

We’re thinking of adding integrations like PostHog to give Fume more context around using your complex app. We’d love to hear your feature requests too!

We have a 30-day free trial (no CC): https://fumedev.com - please do tell us what you think!