Prentis AI lab in talks to raise $100M at $1B valuation

Prentis launched in April as a new AI research lab focused on computer-use models. The startup is co-founded by Ritankar Das, who is also CEO and founder of Titan (a holding company that builds and operates AI companies), alongside prominent tech investors Reid Hoffman and Mark Pincus. Das, now 31, was UC Berkeley's youngest University Medalist in more than a century and earned a master's in biomedical engineering at Oxford before founding Titan in 2014 after dropping out of an AI PhD program at Cambridge where he was a Gates Cambridge Scholar. Hoffman, LinkedIn's co-founder and a Greylock partner, recently stepped down from Microsoft's board to focus on "founder mode" at Manas AI, an AI drug-discovery startup he is backing. Pincus, the Zynga founder, now runs the investment firm Reinvent Capital.

Prentis is training models to learn how office workers navigate routine workflows across documents and systems, with the goal of building AI agents that can control computers to automate routine tasks. The company has identified specific use cases—handling insurance claims, automating customs duty refund exceptions—where it will develop agents tailored to customer needs without requiring humans to hunt down paperwork. The startup claims its Hive-32B model outperforms rivals including OpenAI's GPT-5.4 and Anthropic's Claude Opus 4.6 on two computer-use benchmarks: WindowsAgentArena, which measures end-to-end task completion on real Windows applications, and ScreenSpot-v2, which tests a model's ability to locate the right on-screen control. According to Prentis's pitch deck, the company's edge comes from running a much smaller, cheaper model—it claims roughly 10 times lower cost per task than frontier APIs, making it more economical to deploy across everyday workflows. (TechCrunch has not independently verified these benchmark results.)

Prentis has already signed contracts worth up to $50 million（約80億円） with several customers including a healthcare management service organization, a manufacturer, and goods and clothing manufacturers. Investor materials obtained by TechCrunch predict an estimated $75 million（約120億円） annualized run rate by the third quarter of this year, though the company notes these figures reflect estimated annualized value based on a contracted fee equal to 20% of savings realized, not recognized revenue, and are performance-dependent and subject to final execution. The company is now in talks to raise $100 million（約160億円） at a $1 billion（約1600億円） valuation.

The market for AI agents in computer use is crowded. Anthropic, OpenAI, and Mira Murati's Thinking Machines Lab are all developing AI agents for computer use. Anthropic has moved to acquire talent directly—it bought the Seattle computer-use startup Vercept earlier this year, folding in its founders and shutting down its product. Prentis has hired more than 25 employees, including researchers who previously worked at OpenAI, Google DeepMind, Meta, Tencent, and Alibaba, according to its website.

Prentis enters a market where several AI incumbents are already investing. Anthropic has moved aggressively into computer-use AI agents, including acquiring the Seattle startup Vercept earlier this year. OpenAI and Mira Murati's Thinking Machines Lab are also developing in this space. What distinguishes Prentis, according to its own materials, is both model performance and cost efficiency—the company claims its Hive-32B model, which is substantially smaller than competitors' offerings, delivers comparable or better results on computer-use benchmarks (WindowsAgentArena and ScreenSpot-v2) while costing roughly 10 times less per task.

The $50 million（約80億円） in signed contracts suggests that enterprise customers are willing to pay for automation of high-friction office tasks. Prentis positions itself as addressing a gap between off-the-shelf AI APIs and the specific workflows real companies use—insurance claims, refund processing, document retrieval. Whether these early wins translate into a defensible business model will depend on execution and on the pace at which larger players like Anthropic and OpenAI add similar capabilities to their own platforms.

No comments yet. Be the first to share your thoughts!