Introduction

Lumen is a workspace where several AI models work as one crew. Instead of one assistant in one chat, you run a team: a manager plans, specialists execute, and they hand work to each other in threads you can read.

A single model hits a ceiling. It loses context on long tasks, it can't check its own work, and it's only great at some things. Lumen lets each model do what it's best at and connects them, so Claude can plan, DeepSeek can code, GPT can test and Gemini can dig through history, all on the same project at the same time.

What you get

A crew, not a chatbotA manager and specialists, each with its own role, mission and folders.

Real modelsClaude, GPT, Gemini, DeepSeek, Qwen and Mistral today. Grok and Llama soon.

Visible handoffsEvery time work moves between models, you see who sent what to whom.

You stay in controlRisky actions wait for your approval. Everything else runs on its own.

Who it's for

Developers and small teams who already use more than one AI model and are tired of copying context between tabs. If you run real projects across several repositories and want models to split the work like a team would, Lumen is for you.

In one line

Lumen turns separate AI models into one coordinated crew that divides the work and delivers.

Where to go next

New here? Read How Lumen works for the mental model.
Ready to try it? Jump to the Quickstart.
Want to understand the arrows? See Handoffs & connections or open the crew map.