Overmind

What makes it different

Overmind is the open-source platform for AI agent training, founded by Tyler Edwards — co-founder and CEO, formerly of British intelligence, where he worked at MI5 and MI6 on using AI to automate intelligence analysis — alongside Sam Brunt, a serial startup operator who helped scale Funding Circle, Pipe and Vertice. Its premise: most AI products run on the same few frontier models, while a frontier lab works differently, collecting data, building evals, training, testing, shipping, and repeating. Overmind exists to give every product its own AI lab, so any team building an agent can run that same loop on its own data, without hiring an ML team or building infrastructure.

Getting started is one MCP connection and one prompt pasted into Cursor, Claude Code, Codex or OpenCode. What you get is the full loop: tracing, with OpenTelemetry traces scored as they arrive and matched to the part of your agent that produced them; datasets, with production traces turned into versioned eval and training sets; evals, where you define what good means for each capability and Overmind measures it on live traffic; optimisers, with prompt, tool and code experiments run in your repo and the winner returned as a git diff; fine-tuning, with small specialised models trained on your data, benchmarked against the model you run today, and downloadable with weights you can export; and inference, with your own models and frontier models behind one OpenAI-compatible API.

Overmind open-sourced its training and deployment platform on 1 October 2026, positioning it as one workflow from agent observability and data preparation through training, evaluation, fine-tuning, and deployment. The bet behind it is that creating and improving models should not stay concentrated in a few teams: every application, business, device and individual has different needs, and meeting those needs will require far more people to be able to create, test and run specialised models.