Jeff Dean, one of Google's most influential executives and its 30th employee, is stepping down to launch his own AI startup alongside several of the company's leading researchers, according to a report published by TechCrunch on Wednesday. The new venture, called Discovery Loop, is structured as a public benefit corporation designed to deploy artificial intelligence to accelerate scientific research. Dean, who has been at Google since 1999, will reportedly assume the role of chief executive.
Joining Dean as co-founders are Sanjay Ghemawat, a senior fellow and top engineer at Google; Quoc Le, a principal AI researcher and founding member of Google Brain; and Oriol Vinyals, a senior research scientist working within Google DeepMind. The startup aims to run thousands of experiments at the same time using advanced algorithms, with the objective of partially automating how research is performed and dramatically expanding the scale at which experiments can be carried out. Discovery Loop also intends to explore using AI to build even more capable AI systems—a concept called recursive self-improvement—which would eliminate the need for human involvement in iteration loops entirely. The company secured initial funding through a round co-led by Radical Ventures and Khosla Ventures, with participation from Kleiner Perkins, Lightspeed, and Doerr Capital, and received financial backing from Alphabet, Google's parent company.
In a joint statement, the founding team said that "the next great frontier for AI is to go beyond answering questions and to begin making discoveries." The company described its mission in a press release, noting that "while science and engineering have tremendously advanced society over past centuries, progress has traditionally relied on slow, sequential human iterations, creating a significant bottleneck." Dean told the New York Times that the opportunity lies in more fully automating what has historically been an intensely human-driven experimental loop, predicting that the result will be "both a higher quantity and a higher quality of experiments."
The use of AI to speed up scientific discovery has attracted sustained interest from both scientific and technology communities over many years, but the field has remained largely experimental with narrow commercial uptake until recently, according to the report. Discovery Loop's approach revolves around leveraging massive computational scale to transform the speed and efficiency of innovation by automating complete experimental loops, the company explained. During his tenure at Google, Dean contributed to foundational elements of the company's search infrastructure—including its crawling, indexing, and query-serving systems—and played a pivotal role in the company's early AI research and the development of Google Gemini's multimodal models.
By automating the research cycle, Discovery Loop hopes to deliver the benefits of transformative technologies to the world far sooner, the founding team stated. The startup's vision is to fundamentally accelerate how both engineering and scientific discovery are performed, with the ultimate goal of achieving breakthroughs and advances that would otherwise take far longer under traditional methods. The departure of such senior figures from Google to pursue this independent venture signals growing confidence in AI's potential to reshape the pace of innovation itself. For organizations weighing strategic investments in research infrastructure, the choice between building internal automation capabilities and partnering with specialized platforms may become increasingly consequential as these tools mature.