(P1)
NeoCognition, a startup developing self-learning AI agents, has emerged from stealth with $40 million in seed funding to tackle what it calls the “50% problem”—the observation that current AI agents fail at complex tasks about half the time. The company’s approach moves away from generalist models to create specialized “expert agents” that learn on the job, aiming to provide more reliable and cost-effective automation for enterprise clients.
(P2)
"AI today is fundamentally unreliable when it comes to executing real work that requires deep expertise," said Yu Su, CEO and Co-Founder of NeoCognition. "Our approach mirrors how humans gain expertise on the job through building a structured model of their micro-world, and would eliminate the extensive manual customization required by current models."
(P3)
The oversubscribed round was co-led by Cambium Capital and Walden Catalyst Ventures, the firm of Intel CEO Lip-Bu Tan. The funding also saw participation from private equity giant Vista Equity Partners, along with A&E Investments, Salience Capital Partners, and Frontiers Capital. Other prominent angel investors include Databricks Co-Founder and Executive Chairman Ion Stoica and leading AI researchers Dawn Song, Ruslan Salakhutdinov, and Luke Zettlemoyer.
(P4)
The investment signals a significant shift in focus from large, general-purpose AI to smaller, specialized agents designed for high-stakes business applications. With backing from major players in both hardware (Intel) and enterprise software (Databricks, Vista), NeoCognition is positioned to challenge the dominance of frontier models from OpenAI, Google, and Anthropic in the B2B sector by offering a more dependable and efficient alternative.
Solving the ‘50% Problem’ in AI Agents
Unlike the massive, static models developed by larger labs, NeoCognition is building agents that continuously learn and adapt to their specific work environments. The technology is designed to create an internal “world model” of a domain, allowing the agent to understand workflows, causal relationships, and constraints much like a human expert.
The founding team, led by Yu Su from his AI agent lab at the Ohio State University, has a strong research background that predates the recent generative AI boom. Their work on projects like Mind2Web, MMMU, and SeeAct has been foundational to the AI agent field and is widely used in models from major industry players. This new class of expert agents is expected to be faster, cheaper, and safer for critical business functions in sectors like finance and law.
A Who’s Who of Strategic Backers
The composition of NeoCognition’s investors provides a strategic advantage beyond capital. The involvement of Lip-Bu Tan, CEO of Intel, points to the critical link between next-generation AI software and the hardware required to run it efficiently. Similarly, backing from Ion Stoica of Databricks and Vista Equity Partners opens a direct path to a vast portfolio of enterprise customers.
"Dr. Su and his team have already developed research that spans every piece of the agent puzzle, ranging from perception to memory, planning, evaluation, and safety," said Lip-Bu Tan. "We are confident NeoCognition is uniquely positioned to tackle the hardest challenges in agentic AI."
The Shift Toward B2B “Expert Agents”
While much of the market is focused on consumer-facing chatbots, NeoCognition is targeting the B2B enterprise sector. The company plans to sell its specialized agent systems directly to large corporations and software service providers.
By optimizing for specific tasks, these expert agents are designed to be more reliable and significantly more cost-effective to operate than their general-purpose counterparts. This focus on dependability and efficiency directly addresses the primary barriers to deploying AI in high-stake corporate environments, where the cost of failure is high. With $40 million in new capital, NeoCognition plans to expand its research team and accelerate the commercial rollout of its platform.
This article is for informational purposes only and does not constitute investment advice.
