36 Labs is an applied AI research firm in London. It works on a problem most models are still bad at: making creative decisions, and explaining why they were made. The team wants models to reason creatively, more like a person would.
Today the product is a harness that sits over any frontier model, video generation models in particular. It handles ideation, generation and distribution end to end, and the content goes out to real audiences. The results feed a data loop that keeps improving the harness. The first use case is music marketing, and the clients are the major labels, including Sony, Warner and Universal.
The company has been profitable since it started operating. It is closing its first seed round within the next couple of weeks, and the round was oversubscribed. The team is about ten people, with four engineers who report to the CTO. He is still very hands-on in the code.
You would join at a big moment. V3 of the harness is being released, which is a huge scope of work. A migration to Cloudflare is planned, and it falls mostly to the backend engineer. The stack is Python, TypeScript, Postgres, LLM and generative model APIs, and open-weight models.