LUML is an open-source MLOps and LLMOps platform where engineers and AI agents work side by side across the entire AI lifecycle — from first experiment to production.
At the center is Prisma, LUML’s agent module: AutoResearch-style AI agents that autonomously design, iterate on, and build machine learning solutions. Instead of just assisting with code, Prisma agents run real ML workflows — proposing approaches, executing experiments, and refining models — while every run stays tracked, versioned, and reviewable in the platform.
Around the agents, LUML covers the full AI lifecycle:
-
Flow – real-time experiment tracking for classical ML runs alongside LLM traces and evaluations
-
Core — a model and artifact registry with full lineage, one-click deployments to any infrastructure, and monitoring of the deployed models
The platform is fully open source and free to start with at app.luml.ai.
