AI Agent Startup Prentis Seeks $100M Funding

A new artificial intelligence venture, Prentis, is reportedly in advanced discussions to secure $100 million in funding. This potential capital infusion would value the nascent company at approximately $1 billion, signaling strong investor confidence in its mission to automate complex computer-based workflows. The startup, which officially launched in April, is developing sophisticated AI agents designed to understand and execute routine office tasks, aiming to significantly boost productivity across various industries.

Prentis focuses on training AI models to interpret how professionals navigate daily operations involving documents and software systems. The ultimate goal is to create autonomous agents capable of managing these tasks without human intervention. Early applications are envisioned in areas such as processing insurance claims and resolving exceptions in customs duty refunds, tasks that often involve extensive manual data handling and cross-referencing.

The company's ambitious trajectory is already supported by significant commercial traction. Sources indicate that Prentis has secured contracts valued up to $50 million with a diverse clientele, including organizations in healthcare management, manufacturing, and the apparel sector. Investor materials suggest an annualized run rate projection of $75 million by the third quarter of this year, though this figure is based on a performance-dependent fee structure tied to cost savings realized by clients.

At the core of Prentis's technological claims is its proprietary Hive-32B model. The company asserts that this model surpasses leading competitors, such as OpenAI's GPT-5.4 and Anthropic's Claude Opus 4.6, on key benchmarks for computer interaction. Specifically, it reportedly excels in the WindowsAgentArena, which assesses end-to-end task completion within real Windows applications, and ScreenSpot-v2, measuring the ability to identify and interact with on-screen controls. A key differentiator highlighted by Prentis is the model's efficiency, claiming a cost per task approximately ten times lower than comparable frontier APIs, making it a more economical solution for widespread deployment in everyday business processes.

The broader market for AI agents is intensely competitive, with major players like Anthropic, OpenAI, and Mira Murati's Thinking Machines Lab also investing heavily in this domain. Anthropic, for instance, has actively acquired talent and technology, including the recent integration of the Seattle-based computer-use startup Vercept. Prentis, however, is betting that the automation of standard office functions will emerge as a more significant AI application than code generation. The company's founding team includes prominent tech figures: Reid Hoffman, co-founder of LinkedIn and partner at Greylock, and Mark Pincus, the founder of Zynga. The venture is also led by CEO Ritankar Das, who previously founded Titan, a holding company that has successfully launched and exited several AI ventures, including Tala Health and Forta Health.

Prentis has already assembled a strong technical team, reportedly hiring over 25 employees with prior experience at leading AI research institutions and corporations such as OpenAI, Google DeepMind, Meta, Tencent, and Alibaba. This strategic recruitment underscores the company's commitment to pushing the boundaries of AI agent capabilities and capturing a significant share of the rapidly expanding market for intelligent automation solutions.