Forget coding assistants for a moment. A newly minted AI research lab called Prentis is wagering that the biggest prize in artificial intelligence isn’t writing software — it’s the endless grind of office busywork, from insurance claims to customs duty refunds. And it’s raising money at a fever pitch to prove it.
Prentis, launched in April, is reportedly in talks to raise $100 million at a $1 billion valuation, according to people familiar with the discussions. Not bad for a startup barely a few months old. The lab was co-founded by serial entrepreneur Ritankar Das alongside two familiar names in tech circles: LinkedIn co-founder Reid Hoffman and Zynga founder Mark Pincus.
The core idea is deceptively simple. Prentis trains models to watch how office workers navigate documents and systems, then builds AI agents that can drive a computer to do those same jobs — no human needed to dig through paperwork. Think automated handling of insurance claims or customs duty refund exceptions.
Where things get spicy is on the benchmark front. By its own account, Prentis says its Hive-32B model outperforms rivals — including OpenAI’s GPT-5.4 and Anthropic’s Claude Opus 4.6 — on two computer-use tests:
- WindowsAgentArena, which measures end-to-end task completion on real Windows applications
- ScreenSpot-v2, which tests a model’s knack for locating the right on-screen control
The clever pitch is efficiency. Rather than throwing a giant frontier model at every task, Prentis runs a much smaller, cheaper one — claiming roughly 10 times lower cost per task than frontier APIs. That, the company argues, makes it economical enough to deploy across mundane, everyday workflows where a pricey model would never pencil out. (Worth noting: these benchmark results haven’t been independently verified.)
The business side already has traction. Prentis has signed contracts worth up to $50 million with several customers, including a healthcare management organization, a manufacturer, and goods and clothing makers. Investor materials project an estimated $75 million annualized run rate by the third quarter of this year — though the company is careful to note those figures reflect estimated annualized value based on a fee equal to 20% of savings realized, not recognized revenue, and are “performance-dependent.”
It’s a crowded arena. Anthropic, OpenAI, and Mira Murati’s Thinking Machines Lab are all chasing computer-use agents. Anthropic has even bought talent outright, snapping up Seattle startup Vercept earlier this year and shuttering its product.
The team behind Prentis has pedigree to spare. CEO Das, now 31, was UC Berkeley’s youngest University Medalist in more than a century, graduating at 18. Hoffman and Pincus treat the lab as more of a side project — Hoffman recently went “founder mode” on drug-discovery startup Manas AI, while Pincus runs Reinvent Capital. Prentis has already hired more than 25 people, including alumni from OpenAI, Google DeepMind, Meta, Tencent and Alibaba.
Whether Prentis can actually out-cheap and out-perform the giants remains to be seen — but the pitch is loud, and the money is listening.
