Prentis Eyes $100M Raise as AI Agents Move Into Office Workflows

Lead

The next major AI agent race may not be about writing more code. It may be about getting software to do the ordinary, repetitive work that office employees handle across documents, forms and enterprise systems. According to TechCrunch, Prentis, a new AI research lab launched in April, is in talks to raise $100 million at a valuation of about $1 billion.

The company was co-founded by Ritankar Das, alongside LinkedIn co-founder Reid Hoffman and Zynga founder Mark Pincus. Prentis is focused on “computer use models” — systems that can understand screens, locate controls, move through applications and complete business processes in the way a human operator would.

Key points

• A large funding round is being discussed: TechCrunch cites two people familiar with the talks who say Prentis is seeking $100 million at a $1 billion valuation. The company did not respond to TechCrunch’s request for comment.
• The product target is enterprise workflow automation: Prentis wants to build agents tailored to customers’ internal processes. Examples cited include insurance claims handling and automating customs duty refund exceptions, where workers often need to hunt through paperwork and multiple systems.
• Customer contracts come with caveats: The company has reportedly signed contracts worth up to $50 million with customers including a healthcare management service organization and manufacturers. Investor materials viewed by TechCrunch also point to an estimated $75 million annualized run rate by the third quarter, but the deck notes that the figure is based on a fee equal to 20% of realized savings and is performance-dependent, not recognized revenue.
• Prentis is emphasizing cost efficiency: The company says its Hive-32B model outperforms rivals on WindowsAgentArena and ScreenSpot-v2, two benchmarks for completing tasks in Windows applications and identifying on-screen controls. It also claims roughly 10x lower cost per task than frontier APIs. TechCrunch has not independently verified those benchmark results.
• The category is crowded: Anthropic, OpenAI and Mira Murati’s Thinking Machines Lab are also working on computer-use AI agents, according to the report. Anthropic has already acquired talent in the area through its purchase of Vercept.

Why it matters

Prentis highlights a broader shift in AI commercialization. Enterprises may be less interested in chatbots that simply produce text and more interested in systems that can remove labor from back-office operations. If an agent can reliably process claims, check documents, reconcile information or escalate exceptions, the business case can be tied directly to measurable savings.

That opportunity also raises the bar. Computer-use agents must operate in messy real-world software environments, not controlled demos. They need to manage permissions, audit trails, security constraints, errors, handoffs and edge cases. A high benchmark score is useful, but enterprise adoption will depend on reliability over time and the ability to prove that savings are real.

The involvement of Hoffman and Pincus gives Prentis immediate visibility, while Das brings experience from Titan, a holding company that has launched and operated several AI-focused businesses. Still, the company’s most ambitious claims remain early: the financing is not finalized, the projected revenue metrics are conditional, and the benchmark comparisons have not been independently confirmed. Prentis is entering one of AI’s most promising markets, but also one where proof will matter more than pitch decks.

Source: TechCrunch AI