Open-source observability for LLM apps
Hi HN, Hugh and Vince here. LLMonitor helps you record, trace & search your LLM queries and chatbot conversations. You can also capture user feedback on your frontend and correlate it with backend LLM queries then use that to fine-tune your own models. The project started has an internal tool in our previous (failed) AI startup. We’re aware the LLM observability space is very crowded. Apart from being open-source, we differentiate with: - Model-agnostic and minimal lock-in (no MITM of requests). - High focus on DX and dashboard clarity. - Support for complex scenarios: e.g. a chatbot that…
In plain words
LLMonitor is an open-source observability platform that records, traces, and searches LLM queries and chatbot conversations. Users can capture customer feedback and correlate it with backend queries to fine-tune models. The tool works model-agnostic without intercepting requests, emphasizes dashboard clarity, and handles complex scenarios with multiple agents and tools. It's designed for early-stage startups and indie developers building LLM applications.
written from the facts on this page · September 2026
From the sources
In the maker’s words, at launch
Hi HN, Hugh and Vince here. LLMonitor helps you record, trace & search your LLM queries and chatbot conversations. You can also capture user feedback on your frontend and correlate it with backend LLM queries then use that to fine-tune your own models. The project started has an internal tool in our previous (failed) AI startup. We’re aware the LLM observability space is very crowded. Apart from being open-source, we differentiate with: - Model-agnostic and minimal lock-in (no MITM of requests). - High focus on DX and dashboard clarity. - Support for complex scenarios: e.g. a chatbot that invokes multiple agents & tools before giving an answer. - Targeting early-stage startups + indie developers. Two of our big focuses in the near future are: - Tests (evaluations) written in either code or English (with GPT-4) to know when queries go wrong. - OpenTelemetry support to expand traces to the rest of the stack. (Both of those have turned out to be much more complicated than anticipated.) One decision we made is to stick with Postgres for everything instead of going OLAP with something like Clickhouse. As we are writing & updating many rows concurrently for the traces, Postgres seems to be the best tool for that. It has been sufficient for our analytics and search needs at the moment. We hope that also makes it easier for people to self-host and contribute. The entire codebase is licensed under Apache 2.0 and we’re fully bootstrapped. Being open-source has already begun to foster a community, e.g. with people building integrations for tooling we wouldn't have been able to support otherwise. To be honest we're still quite embarrassed with the state of the product, but I guess we’d be doing something wrong if that wasn’t the case. We’d love to hear your feedback - thank you! Landing page: https://llmonitor.com Docs: https://llmonitor.com/docs Repo: https://github.com/llmonitor/llmonitor
More ai this month
the category →
I trained a 125M-parameter transformer to autocomplete piano performances in real time (~108 notes/sec on an iPhone 15). The idea is basically GitHub Copilot or Tabnine, except instead of prompting it with code, you prompt it by playing a few notes on a MIDI piano. The model then continues what you played, entirely on-device. The app is free if anyone wants to try it. Happy to answer questions about the model, training, Core ML, or the many things that didn't work.
AI · 16d ago · simedw.com
Astute▲585Automate your B2B brand going viral, with new media creators
AI · 18d ago · company-app.joinastute.com
Hey HN, Henry from Cactus here! We previously released Cactus Needle, a 14MB agentic LLM for tool call, device use, and structured extraction for phones, wearables, smart homes, small robots and microcontrollers. We got really great feedback here, and have now incorporated the suggestions to release Needle 2. The whole model is a single 14MB binary that runs a full session in 28MB of RAM; 45m parameters at 2bit compression. Needle hits 500 tokens/sec decode speed on a Raspberry Pi 5, sits between 400-1,500 tokens/sec on VR devices like Meta Quest 3S and Apple Vision Pro, and ranges…
AI · 26d ago · cactuscompute.com
Launched alongside, October 2023
the whole month →
Nudge 2.0▲1,051In-app experiences to activate, retain, & understand users
Growth · 2023 · nudgenow.com
Unlock AI magic for elevated customer engagement, fast
AI · 2023 · tiledesk.com
- OD
Effortlessly discover API behaviour with a Chrome extension that automatically generates OpenAPI specifications in real time for any app or website.
Dev tools · 2023 · github.com