Sepurux is an AI agent debugger.
Paste any trace — JSON, logs, or trace ID — and instantly see:
- the exact step that failed
- what broke next
- how to fix it
No setup. No config. No eval pipelines.
Just drop in a trace and get a clear diagnosis.
From there you can:
- replay the failure
- run checks across mutations
- turn fixes into test runs
Built for teams working with LangChain, LangGraph, OpenAI Agents, and more.
No reviews yetBe the first to leave a review for Sepurux
Hunter
📌
Hey Product Hunt! 👋 Felix here, founder of Sepurux.
I'm a backend engineer with years in fintech and I built this out of a real frustration.
AI agents were getting deployed into production at companies I worked with, but nobody had tooling to answer the basic question: *what happens when this agent gets unexpected input, a slow tool call, or a bad context window?*
The answer was usually "we find out from a customer."
That's what Sepurux fixes.
We're in early access and I'd genuinely love feedback from this community. If you're deploying agents in any workflow (ops, fintech, SaaS, internal tooling), I want to hear about the failure modes that keep you up at night.
Drop a comment or reach me at felix.sarpong@sepurux.com. Happy to give Product Hunt hunters early access priority. 🚀
Would love your upvotes and questions, every one helps us reach the right engineers today.