RoverQA

The autonomous QA agent for modern engineering teams.

Comment

Inspiration

During work, one of our machine learning engineers deployed a feature that caused a crucial flow to fail, causing downstream cascade failures that led to a critical incident in the company.

What it does

RoverQA is an agentic AI QA testing platform. Give it a staging URL and it will create a visual agent that crawls the webapp and extracts the flow. It will then generate and execute QA test cases and compile its findings into a verifiable report.

How we built it

Claude's computer use model, put into a sandbox environment through a virtual machine; React for front-end, Python back-end. Firecrawl API

Challenges we ran into

  • Ensuring the computer use model could access and make actions in the browser, and then through a virtual machine sandbox environment.
  • Have it run through the tests

Accomplishments that we're proud of

  • Setting up an AI agent! Since we've never done this before.
  • Completing this tonight since we pivoted at 7pm xD

What we learned

  • AI agents can be used to execute repeatable tasks with reasoning
  • AI agents can identify semantic context (beyond simply machine code)

What's next for RoverQA

  • Refinement of RoverQA test agent
  • Refinement of flow graph outputs
  • QA for Mobile Apps
  • Enterprise pilots

Built With

Share this project:

Updates