NeoCognition, a cutting-edge AI research lab, has secured $40 million in seed funding to pioneer self-learning AI agents that mimic human cognition.
The round was co-led by Cambium Capital and Walden Catalyst Ventures, with additional backing from Vista Equity Partners and prominent angels including Intel CEO Lip-Bu Tan and Databricks co-founder Ion Stoica.
From Academia to Frontier AI
Founded by Yu Su, a professor at Ohio State University who leads an AI agent lab, NeoCognition emerged from stealth after Su recognized the potential of recent foundational model advances.
Su, who initially resisted commercializing his research, spun out the company last year amid rapid progress in personalized AI agents.
Building Agents That Learn Like Humans
The lab's mission centers on developing generalist AI agents capable of autonomous self-learning by constructing world models for any domain or environment.
Unlike today's unreliable agents with roughly 50% task success rates, NeoCognition's approach enables rapid specialization into trusted, independent workers.
"Today's agents are generalists; every time you ask them to do a task, you take a leap of faith," Su stated.
Enterprise Impact and Future Horizon
The company plans to sell these agent systems primarily to enterprises and SaaS providers for building autonomous workers or enhancing products.
With about 15 employees, mostly PhDs, NeoCognition leverages Vista's software portfolio for real-world AI modernization.
This funding marks a pivotal step toward reliable AI that could transform industries by bridging the gap between generalist models and specialized expertise.
Looking ahead, such human-like learning agents promise to redefine automation, fostering a future where AI excels in micro-worlds like professionals in niche fields.
Andrew Lee
