Reid Hoffman and Mark Pincus Launch Prentis AI Lab, Seeking $100 Million

Reid Hoffman, co-founder of LinkedIn and a partner at Greylock, has teamed up with Zynga creator Mark Pincus to launch a new artificial intelligence lab called Prentis. The company, which has been operating quietly until now, is working on so-called computer-use models, AI systems that can see a screen on their own and operate software instead of just generating text.

A New Lab in a Crowded Market

Prentis marks Hoffman's third venture in AI, following earlier involvement with Inflection AI and Manas AI. Pincus, known chiefly as Zynga's founder and the creator of FarmVille, has spent the past several years investing in AI companies and previously funded a grant for an AI lab focused on organizational innovation at the Wharton School, a separate undertaking unrelated to Prentis. The third co-founder is Ritankar Das, an entrepreneur who also runs the investment firm Titan Holdings.

Models That Operate Computers Instead of Just Talking

Prentis is betting on computer-use models, technology that lets AI click, fill out forms, and navigate applications on its own the way a person would. That's a different direction from classic text chatbots, though similar to what Anthropic is already building into Claude and what OpenAI is developing with its own computer-operating agents.

According to the company's website, the team trains its models through continuous interaction with real work tasks rather than on static datasets collected in advance. The goal is for the models to learn actual, evolving company procedures rather than just patterns scraped from the internet.

Their progress is limited less by raw intelligence and more by experience, meaning exposure to how work actually gets done. - Prentis AI Lab

The Team and the Money

The lab already employs more than 25 people, including researchers and engineers from OpenAI, Google DeepMind, Meta, and Chinese giants Tencent and Alibaba, with part of the team coming from MIT, Berkeley, and Carnegie Mellon. That hiring pattern suggests Prentis wants to compete directly with labs building top-tier agentic models, not just wrappers around existing APIs.

According to reports, the company is in talks for a funding round of around $100 million at a valuation close to $1 billion. The exact round type, the list of investors, and the final terms have not yet been officially confirmed by Prentis, Hoffman, or Pincus, so these figures remain reported rather than a closed deal.

What It Means for the Agent Market

The market for computer-operating agents keeps getting more crowded. Alongside major labs like Anthropic and Google DeepMind, specialized startups building nothing but computer-use models are competing for the same territory, and each new mega-round raises the bar of entry for smaller players. A near-$1 billion valuation for a company still building its product shows how much capital investors keep funneling into agentic AI, even though many earlier projects of this kind have yet to prove a durable business model.

For Polish companies and developers tracking the agent market, this means another potential supplier of models that could automate office work, software testing, or enterprise system operations without requiring a separate API integration for every task. For now, Prentis hasn't announced any commercially available product, so the practical relevance for companies outside Silicon Valley remains distant.

The next step to watch for is the formal close of the funding round and the first public demo of Prentis's computer-use models. Only then will it be possible to judge whether the new lab actually brings something that sets it apart from existing agents at Anthropic, OpenAI, or Google, or whether it's just another player rehashing the same idea with a different founding team.