Hey Product Hunt! 👋 I'm Luca, a solution architect who built this as a side project out of a frustration I kept running into: getting a second opinion on AI-generated code meant copy-pasting between browser tabs, losing context and manually comparing results. It was slowing everything down. So I built AI Peer Review to let the models talk to each other or, at least, work on the same problem in a structured way. A few things I'm particularly happy with: Challenge Mode ended up being my favourite to use watching ChatGPT and Claude compete on the same prompt side by side is genuinely informative. They make very different choices. No backend was a deliberate decision. I wanted something anyone could clone and run in 2 minutes without setting up a server or managing secrets. Companion Mode came from realising that "architect + developer" is actually a better split than "coder + reviewer" for complex problems. The architect model reasons freely, the developer model just executes. The project is fully open source under AGPL v3. Fork it, break it, improve it. The only rule is don't push directly to main 😄 Happy to answer any questions about the architecture, the model choices, or where I'm planning to take this next. What mode would you find most useful in your workflow?

AI Peer Review

One AI writes the code. Two others tear it apart.

One AI writes the code. Two others tear it apart.