Silicon Valley heavyweights Reid Hoffman and Marc Pincus are betting big that AI's next frontier isn't writing code - it's handling the mundane computer tasks that drain knowledge workers' time. The duo has quietly launched Prentis, a new AI lab now in active talks to raise $100 million, TechCrunch has learned. The move marks a strategic pivot from the coding-assistant arms race that's dominated AI investment, instead targeting the broader market of routine digital work that affects every office worker, not just developers.
LinkedIn co-founder Reid Hoffman and Zynga creator Marc Pincus are making a contrarian bet on AI's future. While the industry pours billions into coding assistants and developer tools, their new venture Prentis is laser-focused on something far more universal: automating the repetitive digital busywork that consumes hours of every knowledge worker's day.
The company has emerged from stealth with ambitious fundraising plans, seeking $100 million to build what it's calling a "neolab" - a term that signals something between a research outfit and a product company. According to sources familiar with the discussions, investor interest has been strong, drawn both by the founders' track records and the scope of the problem they're targeting.
Hoffman's involvement is particularly noteworthy. The Greylock Partners investor has been vocal about AI's transformative potential, but his decision to co-found rather than simply invest suggests conviction that existing players are missing a critical market segment. Pincus, who built Zynga into a social gaming giant before its eventual sale, brings expertise in creating addictive, user-friendly experiences at massive scale.
The thesis driving Prentis cuts against the grain of current AI development. While OpenAI, Anthropic, and Microsoft have focused heavily on GitHub Copilot-style coding assistants, Prentis argues that developers represent a fraction of the potential market. Every office worker spends hours each week on repetitive tasks - scheduling meetings, data entry, document formatting, email triage - that don't require coding skills but desperately need automation.
This isn't entirely new territory. Robotic process automation companies like UiPath and Automation Anywhere have built billion-dollar businesses on similar promises. But those tools typically require IT departments to configure complex workflows. Prentis appears to be betting that modern large language models can make task automation as simple as describing what you want done in plain English.
The timing aligns with broader industry signals. Google recently demonstrated AI agents that can navigate websites and fill out forms autonomously. Anthropic has been testing Claude's ability to control computers directly. Both indicate growing confidence that AI can reliably handle multi-step digital tasks without constant human supervision.
But execution risk remains high. Task automation demands extraordinary reliability - a coding assistant that makes mistakes is annoying, but an AI that schedules the wrong meeting time or enters data incorrectly creates real business problems. Prentis will need to solve for accuracy, security, and user trust simultaneously while competing against both incumbent RPA vendors and the tech giants now entering this space.
The $100 million target, if achieved, would give Prentis substantial runway to build foundational models, recruit top talent, and establish enterprise partnerships. For context, that's roughly double what typical seed or Series A AI startups raise, but a fraction of the multi-billion dollar rounds going to foundation model companies. It positions Prentis as application-layer focused rather than competing to build the next GPT.
Neither Hoffman nor Pincus has publicly commented on Prentis yet, and the company maintains minimal online presence. That stealth approach mirrors how many well-funded AI labs have launched recently - staying quiet until they have compelling demos rather than announcing vaporware.
The competitive landscape is getting crowded fast. Microsoft is embedding automation capabilities directly into Office 365 through Copilot. Startups like Adept and Induced AI are building similar task automation agents. And every major enterprise software vendor is racing to add AI-powered workflow features. Prentis will need to move quickly to establish differentiation before the market consolidates.
What gives Hoffman and Pincus credibility is their history building platforms that achieved massive adoption. Hoffman didn't just invest in social networking - he architected LinkedIn's growth from zero to hundreds of millions of professionals. Pincus turned casual gaming into a cultural phenomenon. Both understand how to make complex technology feel simple and habitual for mainstream users, which is exactly what workplace AI needs to break out beyond early adopters.
Prentis represents a high-profile bet that AI's biggest impact won't come from helping the 30 million developers worldwide write better code, but from liberating the 1 billion-plus knowledge workers trapped in digital busywork. If Hoffman and Pincus are right, the companies winning the coding assistant wars may be fighting over a niche while the real prize - automating work itself - goes to whoever makes task automation reliable, secure, and dead simple. The $100 million they're raising suggests they're serious about being that company. Now they just have to build it before Microsoft, Google, and a dozen well-funded startups beat them to market.
