Jeff Dean's new venture, Discovery Loop (aimed at automating the scientific method)

Jeff Dean, one of the most influential engineers in Google’s history, left after 27 years to start Discovery Loop. The company wants to automate the scientific method itself. Google is a founding investor, along with Khosla Ventures, Radical Ventures, Lightspeed, Kleiner Perkins, and Doerr Capital. Funding size hasn’t been confirmed in the reporting.

This matters because it’s not just another AI lab. Dean built foundational systems at Google, the kind that run at planetary scale. He’s now betting that AI can generate hypotheses, design experiments, interpret results, and iterate without waiting for human researchers to do each step manually. If that works, it compresses the timeline from question to answer in fields like materials science, drug discovery, or climate modeling.

Google’s involvement as an investor signals this isn’t a talent loss they’re trying to contain. It’s more of a hedge. They keep ties to whatever Dean builds while letting him operate outside their structure. The investor list is notable for backing deep tech, not consumer apps. That suggests Discovery Loop is aiming at problems that take years to solve and require serious compute.

For business owners, this is a signal about where frontier AI is heading. The next wave isn’t chatbots or image generators. It’s systems that can run entire research loops, the kind of capability we’re starting to integrate into tools like the Omni Command Centre for companies that need AI to handle complex, multi-step workflows. If automating science becomes real, automating internal R&D, competitive analysis, or product testing isn’t far behind. Watch how Discovery Loop structures its work. The methods they develop will trickle into commercial tools faster than you expect.