Phospho is an open-source text analytics platform for LLM apps in production. Companies like Quivr or Sudowrite use phospho to monitor interactions with their LLM app and take action. We want to be the Datadog the LLM OS. Text will be the main mean of communication between humans and computers, and also computers to computers. A tool is needed to analyze these flows in real time.
✨ Star us on Github and follow us on Twitter
TL;DR: Phospho is building open-source text analytics tools for LLM Apps. We help companies turn their LLM prototype into a product with testing, evaluation, monitoring, and guardrails at the semantic level.
Hey everyone, we’re Paul-Louis and Pierre-Louis. phospho is on a mission to help you monitor, test, and improve your LLM app at the semantic level.
🚨Problem
Building LLM apps has never been easier. There are TONS of tools. Yet, companies that ship to production are scarce. And lots of AI tools that have made it to production have a HIGH churn rate and LOW usage rate. Why?
Unfortunately, many AI builders are trapped at ground zero:
No wonder they feel stuck. Their best chance at improving is either guesswork or reading through thousands of logs.
There is no secret. Here is what the best companies shipping LLM products do that others don’t:
🧪 This is why we are building phospho. phospho is an open-source text analytics tool for LLM apps.
We are gathering all the tools that enable your team to go from prototype to product at record pace: testing, evaluation, monitoring, and guardrails at the semantic level. Let’s deep dive.
Python and Javascript SDKs to easily integrate into your LLM stack
Phospho can be self-hosted or used in our managed cloud version