Discovery Loop

Extracted Attributes

• Founders

  Jeff Dean, Sanjay Ghemawat, Quoc Le, Oriol Vinyals

• Headquarters

  Palo Alto, United States

• Inception Date

  2026-01-01

• Legal Structure

  Public Benefit Corporation

• Co-lead Investor

  Radical Ventures

• Primary Objective

  Automating multi-step scientific and engineering experimental loops

• Founding Investor & Partner

  Google

Timeline

• Discovery Loop was formally established in Wikidata records. (Source: Wikidata)

  2026-01-01

• Jeff Dean left Google after nearly 27 years to publicly announce the founding of Discovery Loop alongside Sanjay Ghemawat, Quoc Le, and Oriol Vinyals, backed by Google and Radical Ventures. (Source: Web Search (Jeff Dean Leaves Google to Automate the Scientific Method With Discovery Loop))

  2026-08-05

Oriol Vinyals

Oriol Vinyals (born 1983) is a Spanish machine learning researcher. He has developed most of his career at DeepMind. He is a co-founder of Discovery Loop, and was technical lead on Gemini until August 2026, along with Jeff Dean.

Web Search Results

• WSU Discovery Trail

  The Weber State University Discovery Loop trails are self-guided interpretive science trails in the foothills above the WSU campus. They feature point-of-interest signs and learning stations to educate students of all ages about the unique geology, biology, and wildlife of the Wasatch Front. The trails were planned with three audiences in mind: the general Ogden community, tourists, and students. Just a few minutes from the Discovery Trailhead at a series of steps, the two Discovery Loops split. The North Loop heads uphill where it reaches the Bonneville Shoreline Trail in less than a half mile. At this point, you may decide to return the way you came or hike north to Strong’s Canyon. After a short, steep descent along a badly eroded section of trail, you will come to Strong’s Creek. If

• Jeff Dean and Sanjay Ghemawat Depart Google to Co-Found Discovery Loop

  AI-for-science refers to the application of machine learning systems to automate or accelerate the experimental cycle of scientific discovery — hypothesis generation, experiment design, execution, data analysis, and model refinement. AlphaFold, which predicted the three-dimensional structures of more than 200 million proteins and earned its creators the 2024 Nobel Prize in Chemistry, is the sector's most prominent example. Discovery Loop's specific approach — fully automating multi-step experiments — places it alongside companies like Periodic Labs (AI for materials science), Isomorphic Labs (drug discovery), and Project Prometheus (physical AI for aerospace and manufacturing). What distinguishes Discovery Loop is its founders' systems-engineering background: the hardest part of automated [...] This background matters for understanding what Discovery Loop actually proposes to do. The company's stated mission is to automate multi-step scientific and engineering experiments — first building an internal technology stack to automate machine learning research itself, then expanding into hardware design, drug discovery, and clean energy. The engineering challenge that mission implies is not primarily a modeling problem. It is a distributed systems problem: run millions of experimental configurations in parallel, close the loop from experiment results back into model retraining, maintain fault tolerance across long-running automated research pipelines, and do it at a scale no human team could orchestrate manually. That is the problem Dean and Ghemawat built their careers solving. [...] ### How Is Discovery Loop Structured? Discovery Loop is incorporated as a public benefit corporation — the same legal structure used by OpenAI and Anthropic. Under Delaware law, a public benefit corporation must pursue a stated public benefit alongside profit, and directors are explicitly authorized to balance shareholder interests against that public benefit — insulating them from purely profit-maximizing shareholder lawsuits that could otherwise constrain a mission-driven organization.

• Our Investment in Discovery Loop

  Discovery Loop is focused on automating this process. Instead of humans running experiments in series, the Discovery Loop team is building AI systems that run thousands of experiments in parallel. The AI proposes the experiment, conducts the run, learns from the result, and iterates recursively. Jeff, Sanjay, Quoc and Oriol are starting by automating machine learning research, acting as their own first customer. From there, they will direct these systems toward the physical world, targeting grand challenges in engineering, medicine, materials design, and clean energy. [...] Radical Ventures # Radical Blog ## Our Investment in Discovery Loop By Jordan Jacobs, Co-Founder & Managing Partner; Aaron Rosenberg, Partner August 05, 2026 The Discovery Loop team: Oriol Vinyals, Sanjay Ghemawat, Jeff Dean and Quoc Le Today we are announcing our co-lead investment in Discovery Loop, a public benefit company founded by Jeff Dean, Sanjay Ghemawat, Quoc Le, and Oriol Vinyals, legends in the field of artificial intelligence. Radical Ventures was founded on the thesis that AI will both replace traditional software and unlock progress in science and engineering. Such transformative progress will improve lives and mark an extraordinary moment in human history that will be remembered for centuries. [...] Discovery Loop is building an in-person team in Palo Alto. If you want to join the team and help build the engine for scientific discovery, please visit www.discoveryloop.com. Radical Reads is edited by Ebin Tomy (Analyst, Radical Ventures) 1765918712773 Radical Blog Geoffrey Hinton and Jeff Dean in Conversation image 2 Radical Blog A Functional Taxonomy of World Models Radical Blog Conversation on The Rise of NeoLabs Screenshot 2026 07 16 152852 Radical Blog Putting Frontier Drug Design in Every Researcher’s Hands

• Jeff Dean Leaves Google to Automate the Scientific Method With Discovery Loop – Unite.AI

  ## What happens next The near-term observables are concrete. Discovery Loop is recruiting a lean, in-person founding team through its open roles page, having started with no hires and no office. Its first executable milestone is the automated machine learning loop it plans to turn on its own stack, and its first year of operation runs on Google-provided compute under the Cloud partnership. The stated expansion path, from ML research into chip design, biology, drug discovery, and materials, will be measurable against the company’s own criterion: learning loops with outcomes that can be scored. Whether the approach produces a result the field recognizes as a discovery, rather than an optimization, is the question the company’s structure is built to answer. mm [...] Unite.AI Unite.AI ## Unite.AI #### Jeff Dean Leaves Google to Automate the Scientific Method With Discovery Loop ### AI Models & Platforms # Jeff Dean Leaves Google to Automate the Scientific Method With Discovery Loop mm By Jeff Dean, Google’s chief scientist and one of the central figures in modern AI infrastructure, is leaving the company after nearly 27 years to co-found Discovery Loop, a startup whose stated goal is to automate the experimental loops of scientific research itself. The departure was announced on August 5, 2026 as part of a broader reshuffle of Google’s AI leadership, with the company confirming it will back the new venture as a founding investor and Cloud partner. [...] ## What Discovery Loop actually plans to build The mechanism, per the company’s own description, has three stages. First, it will focus on automating machine learning research and engineering, using frontier AI models and large-scale compute to propose, run, and learn from evaluations. Second, it will act as its own first customer, using those automated capabilities to optimize its own technology stack. Third, it intends to generalize to any learning loop with measurable outcomes in science and engineering, citing the National Academy of Engineering’s Grand Challenges as its eventual target class, including engineering better medicines, advancing health informatics, and making solar energy economical.

• Medium

  Building and releasing often is critical, to be sure. But I also believe that you can get critical information to guide your path to product-market fit even before you build any version of your product, let alone bring it to market. I use a tool I call the Product Discovery Loop to break down product-market fit into components you can more easily identify and test along the way. What Is the Product Discovery Loop? The product discovery loop includes market segment, customer pains, desired outcomes, and solutions. Going through discovery around these four components gives you a better understanding of who your business is trying to target, what they want, what gets in the way of that, and how you can help them. [...] Each successful pass through the Product Discovery Loop increases customer love for your product, growing the addressable market by covering additional customers, pains, or use cases. Next, I’m going to share more detail about each component and show how focusing on one pass through the Product Discovery Loop with each build phase sets your team up for high-growth product development. 1. Market Segment The first component of the Product Discovery Loop involves knowing which segment of the market you’re building for. Typically, a founder will have a vision for the business they’re going to build and the product that goes along with it. They’ll have an idea of what market it’s for, but many times they’ll describe a market in terms that are overly general for a new product.

• Country

  United States

• Founder

  Jeff Dean

• Located In

  Palo Alto

• Instance Of

  Q129238

• Headquarters

  Palo Alto

• Inception Date

  1/1/2026