NeoCognition raises $40 million seed round to develop self-learning AI agents

Investor interest in artificial intelligence continues to rise. In particular, venture capital firms are encouraging researchers to turn academic work into commercial products. This shift reflects a broader effort to improve how AI systems perform in real-world environments.

From academic lab to startup

Yu Su, a professor at Ohio State University, had initially avoided launching a company. However, he reconsidered as advances in foundational AI models created new opportunities. As a result, he spun out his research into a startup focused on building more adaptive AI systems.

NeoCognition, the company founded from this work, has now raised $40 million in seed funding. The round was co-led by Cambium Capital and Walden Catalyst Ventures. In addition, Vista Equity Partners and several angel investors participated, including industry leaders from the technology sector.

Addressing reliability challenges in AI agents

Current AI agents often struggle with consistency. Although they can complete tasks, their success rate remains limited. According to Su, many existing systems perform correctly only about half the time. Therefore, businesses remain cautious about relying on them for independent work.

This gap highlights a key limitation. While AI agents can handle general tasks, they lack the ability to consistently deliver accurate outcomes. As a result, their adoption in critical workflows remains restricted.

A focus on learning and specialization

NeoCognition aims to address this issue by developing agents that can learn continuously. Instead of relying on fixed programming, these agents are designed to build their own understanding of specific environments.

The approach draws from how humans learn. For instance, individuals can adapt to new roles by understanding rules, relationships, and outcomes within a given context. Similarly, NeoCognition’s agents are being built to develop expertise through ongoing interaction.

According to Su, this ability to specialize is essential. While human intelligence is broad, its effectiveness often depends on deep knowledge in a specific domain. Therefore, enabling AI to achieve similar specialization could improve reliability.

Building adaptable enterprise solutions

Unlike many existing tools, NeoCognition is not focusing on single-use applications. Instead, it is developing general-purpose agents that can adapt to multiple domains over time. This approach reduces the need for custom engineering for each industry.

The company plans to offer its technology to enterprises, including SaaS firms. These organizations can use the agents either to create autonomous digital workers or to enhance existing products.

Strategic backing and future direction

The participation of Vista Equity Partners may provide additional advantages. As a major investor in software companies, Vista offers access to a large network of potential enterprise customers. This could support faster adoption of NeoCognition’s technology.

At present, the startup has a team of around 15 employees. Notably, most of them hold doctoral degrees, reflecting its research-driven approach.