Hi Product Hunt! I built ReasonLens around a problem I kept coming back to: when an AI application produces a different answer, how do you understand what changed? Looking at the final response isn't enough. The retrieval results may have changed, a tool may have failed, evidence may have disappeared, or the execution may simply have become slower and more expensive. ReasonLens lets you trace AI execution, inspect evidence and claim relationships, compare runs, and investigate potential quality regressions. It supports chat, RAG, and agent workflows, with comparisons designed to distinguish observable quality signals from structural differences. I'm sharing it to get feedback from developers building real AI applications. **I'd especially like to know:** What's the hardest part of debugging your LLM or RAG application today? Try it: https://reasonlens.vercel.app/ GitHub: https://github.com/talusannni/Re... Feedback and feature requests are welcome!

ReasonLens

Understand, compare, and debug AI application traces

Understand, compare, and debug AI application traces