AI research startup NeoCognition raises $40M seed to develop self-learning domain agents

NeoCognition, a startup built from the AI research lab of Ohio State University professor Yu Su, has completed a $40 million seed funding round. Cambium Capital and Walden Catalyst Ventures co-led the round, with additional backing from Vista Equity Partners, Intel CEO Lip-Bu Tan, and Databricks co-founder Ion Stoica. Su described his initial reluctance to commercialize his academic work, but moved forward after observing how advances in foundational AI models had created new opportunities for personalized agent systems.

The startup's core thesis centers on a perceived limitation in current AI agents. Su characterizes existing systems—including those from Anthropic, OpenAI, and Perplexity—as unreliable generalists that succeed at completing tasks as specified roughly 50% of the time. This inconsistency, he argues, prevents agents from functioning as trustworthy independent workers on enterprise tasks.

NeoCognition's approach focuses on agents that can autonomously learn and adapt to specific domains, building what Su calls a 'world model' for any given vertical. The company positions this capability as analogous to human specialization: while humans possess broad intelligence, our practical power derives from the ability to rapidly master the unique rules, relationships, and consequences of new professional environments. The startup aims to embed this specialization capacity into AI agents.

The business model targets enterprises and established SaaS providers seeking to modernize existing products with agentic capabilities or build new agent-based workflows. The participation of Vista Equity Partners, a major private equity player in enterprise software, reflects this positioning—Vista's portfolio would provide NeoCognition with direct customer access. NeoCognition currently operates with approximately 15 employees, most holding doctoral degrees.