NeoCognition Emerges with $40M to Build AI That Teaches Itself

A new AI research startup, NeoCognition, has launched with a substantial $40 million seed investment. The company, founded by former Ohio State professor Yu Su, aims to develop a new class of self-learning AI agents designed for reliability.

Su explained the current problem with AI agents, noting that tools from major providers only complete tasks as intended roughly half the time. This inconsistency makes them unsuitable as independent workers. "Every time you ask them to do a task, you take a leap of faith," Su told TechCrunch.

NeoCognition's approach is modeled on human learning. While people have broad intelligence, their real strength is the ability to rapidly specialize—to enter a new field and master its specific rules and relationships. The startup is building agents that can do the same: autonomously learn and build a detailed model of any given domain to become an expert.

"We believe for agents to become experts, they need to learn autonomously to build a model of any given micro world," Su said. This self-specialization, he argues, is the key to creating AI that can work reliably without constant human oversight.

The funding round was co-led by Cambium Capital and Walden Catalyst Ventures. Participants included Vista Equity Partners and angels like Intel's Lip-Bu Tan and Databricks co-founder Ion Stoica. Su highlighted Vista's involvement as strategically important, providing a direct channel to a large portfolio of enterprise software companies seeking AI integration.

NeoCognition, which currently employs about 15 people, most with PhDs, plans to sell its agent systems to established SaaS companies and other enterprises looking to build AI-powered workers or enhance their existing products.

Source: TechCrunch