Tracium - Open-source, OpenTelemetry observability for LLMs

Hey Product Hunt 👋 I’m Antonij, the maker of Tracium. If you’re shipping AI features, you’ve probably had this moment: the bill comes in higher than expected, an agent starts failing, and nobody can say exactly why. Which model? Which workflow? Which customer? Tracium answers that in minutes. It shows you what every AI call costs, how long it took, and where it broke, all in one clean dashboard. What teams use it for: 💸 Knowing where the money goes: spend by model and workflow, day by day 👥 Seeing which customers drive your AI cost: usage and spend per client 🐞 Fixing failures faster: open any failed run and jump straight to the span that broke it 🚨 Catching problems early: automatic flags when cost, errors or volume spike above normal Setup is quick. Tracium is built on OpenTelemetry, the open standard many teams already use. If you’re already sending OpenTelemetry data, connecting takes a couple of minutes, with no code changes. If you aren’t, our docs get you there quickly. And today’s big news: Tracium is now open source 🎉 You can run it on your own servers for free, so your data never leaves your hands. Or, if you’d rather not manage anything, our cloud plans start with a 14-day free trial. I’d love for you to give it a spin. I’ll be here all day answering questions, so tell me what you’d want to see next! 🙏