Prentis is trying to turn a difficult part of office work into an AI business: getting software to use computers the way employees do. The new AI research lab, co-founded by Ritankar Das, Reid Hoffman, and Mark Pincus, is in talks to raise $100 million at a $1 billion valuation, according to two people familiar with the discussions.
The company launched in April and is building computer-use models meant to observe and learn how office workers move through documents, systems, and routine workflows. Its broader goal is to create AI agents that can control computers and automate the repeated tasks that often sit between teams, files, and back-office software.
A bet on office automation
Prentis is not pitching AI as a writing assistant or a coding helper. Its focus is the practical layer of office work where employees switch between systems, look for documents, verify information, and complete process-heavy tasks.
The company’s models are being trained to understand how workers navigate those workflows across documents and systems. If that works, the resulting agents could perform tasks that currently require people to click through software, gather paperwork, and resolve exceptions.
Examples cited for Prentis include handling insurance claims and automating customs duty refund exceptions. In those cases, the agent would be built around customer needs rather than offered as a general-purpose chatbot. The intended value is straightforward: reduce the amount of manual searching, checking, and handoff work required for recurring operational processes.
That positioning explains why the startup is emphasizing computer use. An agent that can operate across real software environments could be more useful in messy office workflows than one limited to text generation. The challenge is that such systems must do more than answer questions; they need to identify what is on screen, choose the right controls, and complete tasks end to end.
Early contracts and performance-based projections
Prentis has already signed contracts worth up to $50 million with several customers, the two people familiar with the discussions told TechCrunch. Those customers include healthcare management service organization, a manufacturer, and goods and clothing manufacturers.
Investor materials obtained by TechCrunch point to an estimated $75 million annualized run rate by the third quarter of this year. The pitch deck also says those figures are based on estimated annualized value tied to a contracted fee equal to 20% of savings realized.
That distinction matters. The deck notes that the figures are not recognized revenue. It also describes them as “performance-dependent and subject to final execution.” In plain terms, the projected business value depends on whether the automations are completed and whether the promised savings are actually achieved.
For a young AI company, that kind of structure can be both attractive and demanding. It connects pricing to measurable customer outcomes, but it also puts pressure on the technology to deliver in real operating environments. If Prentis is charging against savings, the business case has to survive contact with real workflows, not just demonstrations.
The model claim at the center
Prentis says its Hive-32B model outperforms rivals on two computer-use benchmarks. The rivals named in its materials include OpenAI’s GPT-5.4 and Anthropic’s Claude Opus 4.6.
The two benchmarks are WindowsAgentArena and ScreenSpot-v2. WindowsAgentArena measures end-to-end task completion on real Windows applications. ScreenSpot-v2 tests whether a model can locate the correct on-screen control.
According to the company’s pitch deck, Prentis believes its advantage comes from using a much smaller and cheaper model. The deck claims roughly 10 times lower cost per task than frontier APIs, which would make the model more economical to deploy across everyday workflows.
That cost argument is central to the company’s story. Office automation often involves repeated tasks, so the price of each completed action can become a major constraint. A model that is cheaper per task could be easier to apply across many routine processes, provided it is reliable enough.
There is an important caveat: TechCrunch hasn’t independently verified the company’s benchmark results. The claims remain part of Prentis’s own pitch, not a confirmed outside assessment.
A crowded race for computer-use agents
Prentis is betting that everyday office automation will soon become a larger AI use case than coding. It is entering a market where major AI companies and new labs are already active.
Anthropic, OpenAI, and Mira Murati’s Thinking Machines Lab are also working on AI agents for computer use, one of the sources said. Anthropic has also moved directly in the category by acquiring talent: it bought the Seattle computer-use startup Vercept earlier this year, brought in its founders, and shut down its product.
The competitive pressure is obvious. If computer-use agents become a major AI market, companies with large models, deep research teams, and existing customer relationships will all want a role. Prentis is trying to differentiate through specialization, benchmark performance, and lower deployment cost.
The founders behind Prentis
Prentis brings together founders with different backgrounds in AI, startups, and investing. Ritankar Das is CEO of Prentis and also the founder of Titan, a holding company that builds and operates AI companies.
Das, now 31, was UC Berkeley’s youngest University Medalist in more than a century. He graduated at 18 with a double major in bioengineering and chemical biology, then earned a master’s in biomedical engineering at Oxford.
He founded Titan in 2014 after leaving an AI PhD program at Cambridge, where he had been a Gates Cambridge Scholar. Titan’s other businesses include AI-powered virtual care provider Tala Health, Forta Health, and disease prediction company Dascena, which was acquired by CirrusDx in 2022.
For Reid Hoffman and Mark Pincus, Prentis is described as a side project of sorts. Hoffman, the LinkedIn co-founder and Greylock partner, said last month that he was stepping down from Microsoft’s board after nearly a decade to go “founder mode” on Manas AI. He was also an early OpenAI investor and co-founded Inflection AI with Mustafa Suleyman before Microsoft absorbed most of that team in 2024.
Pincus, the Zynga founder, now runs the investment firm Reinvent Capital with Hoffman as a senior adviser. He also published a memoir, “Life at the Speed of Play,” last month.
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 didn’t respond to TechCrunch’s request for comment.