NeoCognition secures $40M seed funding to develop human-like learning AI agents

Investors are increasingly targeting AI researchers to back startups building more reliable and efficient AI systems.

Yu Su, an AI professor at Ohio State University who leads an AI agent research lab, said he initially resisted pressure from venture capital firms to commercialise his academic work. However, he eventually decided to spin out his research into a startup after observing rapid advances in foundational models that could enable more personalised AI agents.

The resulting company, NeoCognition, describes itself as a research-driven startup building self-learning AI agents. The company has now emerged from stealth mode with $40 million in seed funding. The round was co-led by Cambium Capital and Walden Catalyst Ventures, with participation from Vista Equity Partners and several angel investors, including Intel CEO Lip-Bu Tan and Databricks co-founder Ion Stoica.

“Today’s agents are generalists,” Su said. “Every time you ask them to do a task, you take a leap of faith.”

According to Su, the core issue with current AI systems is inconsistency. He argues that existing AI agents—whether in tools like Claude Code, OpenClaw, or Perplexity’s computer-based systems—complete tasks as intended only roughly 50% of the time, making them unreliable for independent execution.

Because of this inconsistency, Su believes current agents are not yet suitable as autonomous workers. NeoCognition aims to address this limitation by developing AI systems that can continuously self-learn and evolve expertise in specific domains, similar to human learning processes.

Su explained that human intelligence is not just a general ability but also the capacity to specialise quickly when entering new environments. Humans can adapt to new roles by learning rules, relationships, and consequences within a short time.

NeoCognition’s approach is designed to replicate this mechanism in AI systems.

“For humans, our continued learning process is essentially the process of building a world model for any profession, any environment,” Su said. “We believe that for agents to become experts, they need to learn autonomously to build a model of any given micro world.”

The company believes this ability to specialise rapidly is the missing foundation needed for AI agents to operate reliably in real-world applications.

While some AI systems can already be trained to perform autonomous tasks, Su noted that they typically require extensive customisation for each industry or use case. NeoCognition differentiates itself by focusing on general-purpose agents that can independently learn and adapt across multiple domains.

The startup plans to primarily target enterprise customers, including established SaaS companies that can integrate AI agents into their products or deploy them as digital workers.

Su emphasised that backing from Vista Equity Partners is strategically important, as the firm’s portfolio includes numerous software companies that could benefit from AI-driven modernisation.

NeoCognition currently employs around 15 people, most of whom hold PhDs, reflecting its research-heavy approach to AI development.