Backed by LinkedIn and Zynga Founders, Prentis Eyes Unicorn Status with Office Automation AI
Key takeaways
- Prentis is seeking $100M at a $1B valuation just months after launching in April, with $50M in enterprise contracts already signed.
- Its Hive-32B model claims to outperform GPT-5.4 and Claude Opus 4.6 on computer-use benchmarks at roughly one-tenth the cost per task.
- The startup faces direct competition from Anthropic, OpenAI, and Thinking Machines Lab, all of which are actively developing computer-use AI agents.
A startup called Prentis is quietly positioning itself as a major player in the race to automate everyday office work using AI agents, and it may be about to get a significant financial boost to do it. The company, which launched just this past April, is currently in talks to close a $100 million funding round at a $1 billion valuation, according to two sources familiar with the negotiations. If those talks succeed, Prentis will enter unicorn territory before most people outside Silicon Valley have even heard its name.
At the helm is Ritankar Das, a 31-year-old CEO who graduated from UC Berkeley at 18 as the university's youngest medalist in over a century, later earning a master's from Oxford and briefly pursuing an AI PhD at Cambridge as a Gates Scholar. Das dropped out of that program in 2014 to found Titan, a self-funded holding company that has since incubated multiple healthcare and AI ventures. Prentis is his latest project under that umbrella, and it carries the added credibility of two high-profile co-founders: Reid Hoffman, the LinkedIn co-founder and Greylock partner, and Mark Pincus, the founder of Zynga who now runs investment firm Reinvent Capital.
Prentis is training its models by observing how office workers actually navigate documents, software tools, and multi-step administrative processes — things like processing insurance claims or handling customs duty refund paperwork. The goal is to build AI agents that can perform these tasks end-to-end, without a human needing to intervene. The company has already inked contracts reportedly worth up to $50 million with a handful of enterprise customers, including a healthcare management organization and multiple goods and clothing manufacturers.
The startup's pitch centers on its Hive-32B model, which it claims outperforms OpenAI's GPT-5.4 and Anthropic's Claude Opus 4.6 on two computer-use benchmarks: WindowsAgentArena and ScreenSpot-v2. Crucially, Prentis argues its smaller model runs at roughly one-tenth the cost per task compared to frontier-model APIs, making it far more practical for repetitive enterprise deployment. Its pitch deck projects an annualized run rate of approximately $75 million by the third quarter of this year, though the company notes these figures are performance-dependent and based on a contracted fee tied to 20% of realized savings rather than recognized revenue.
The team Das has assembled includes researchers who previously worked at OpenAI, Google DeepMind, Meta, Tencent, and Alibaba — a roster that signals serious technical ambition. But Prentis enters a fiercely competitive landscape. Anthropic, OpenAI, and Mira Murati's newly formed Thinking Machines Lab are all pursuing computer-use AI agents. Anthropic went as far as acquiring Seattle-based computer-use startup Vercept earlier this year and folding its founders into its own team. TechCrunch notes it has not independently verified Prentis's benchmark claims, and the company did not respond to a request for comment.
The bigger picture
The timing of Prentis's emergence is telling. Computer-use AI — where models can directly operate software interfaces rather than merely answer questions — has quickly become one of the most hotly contested frontiers in applied AI. What makes Prentis's bet interesting is its contrarian emphasis on cost efficiency over raw capability. While competitors like Anthropic and OpenAI race to build the most powerful frontier models, Prentis is arguing that enterprise customers don't actually need the most powerful model — they need the most deployable one. If Hive-32B can deliver reliable task automation at a fraction of the cost, that's a compelling value proposition for large organizations running thousands of repetitive workflows daily.
Hoffman and Pincus's involvement is more than just a credibility stamp. Both have long track records identifying platform-level shifts early — Hoffman at LinkedIn and as an early OpenAI backer, Pincus at Zynga during the social gaming boom. Their participation signals a belief that automating non-technical office labor represents a commercial opportunity comparable to, or even larger than, AI-assisted coding. That framing matters because the coding-AI market is already crowded and arguably commoditizing; the administrative-workflow market is far less picked-over and far more expansive in terms of labor hours.
Investors and industry watchers should scrutinize a few key risks before treating Prentis as a foregone success. The $75 million ARR projection is explicitly performance-dependent, meaning it won't materialize unless the system actually delivers measurable savings to clients — a high bar for a company that's only been live for months. Additionally, the concentration of big-name co-founders who are clearly juggling multiple ventures (Hoffman is simultaneously backing Manas AI and stepping down from Microsoft's board) raises legitimate questions about bandwidth and long-term commitment. Prentis will need to demonstrate durable customer outcomes quickly, because in a market with Anthropic and OpenAI as rivals, runway and momentum are everything.
We decided to cover Prentis because it sits at the intersection of several converging stories we've been tracking closely here at LagPing — the rapid commercialization of agentic AI, the ongoing power consolidation among a small group of well-connected tech founders, and the question of whether smaller, cheaper models can genuinely compete with frontier giants on real-world enterprise tasks. What struck us about this story isn't just the unicorn valuation or the famous names attached, but the specific claim that office automation will surpass coding as AI's dominant use case. That's a meaningful industry thesis, not a throwaway line, and it deserves serious scrutiny. We also think the Ritankar Das narrative is worth paying attention to: a founder who built a self-sustaining holding company from scratch before age 30 is operating with a very different playbook than typical VC-backed AI startups. Whether Prentis's benchmark claims hold up under independent review will be the real test, and we'll be watching those results closely as the company moves through this funding round.
As an Amazon Associate, LagPing earns from qualifying purchases. Product links are affiliate links.
You might also like
Google Brain Founders Depart to Build Discovery Loop, an AI-Powered Scientific Research Engine
2d ago
LinkedIn Fights Back Against AI Spam With a Dedicated User-Reporting Tool
Jul 31
How a DJI-backed Motor Turned a Hardcore Trail Purist Into an E-MTB Believer
Jul 18
American Rare Earth Miners Backed by Billions in Federal Funds Are Selling to Japan and Korea
Jul 9