Investors are actively lobbying AI researchers to build startups that make AI more reliable and efficient.
Yu Su, the Ohio State University professor who heads the AI Agent Lab, said he initially resisted pressure from venture capitalists to commercialize his research. Last year, he finally took the plunge and spun his work out into a startup, believing that advances in the basic model could allow agents to be truly personalized.
NeoCognition, a startup that Su describes as a research institute developing self-learning AI agents, has just emerged from stealth 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 angels including Intel CEO Lip-Bu Tan and Databricks co-founder Ion Stoica.
“Agents today are generalists,” Hsu (pictured right) told TechCrunch. “Every time you ask them to do a job, you take a leap of faith.”
The problem, Su said, is a lack of consistency. He says that current agents, whether using Claude Code, OpenClaw or Perplexity computer tools, only successfully complete tasks as intended about 50% of the time.
Agents remain so unreliable that they are not ready to become reliable independent workers, Hsu told TechCrunch. NeoCognition aims to change this by developing agent systems that can self-teach to become experts in any field, just as humans learn.
Hsu argues that human intelligence is broad, but its true power lies in its ability to specialize. When we enter a new environment or profession, we quickly learn its unique rules, relationships, and outcomes.
tech crunch event
San Francisco, California
|
October 13-15, 2026
NeoCognition is building agents that reflect exactly this approach.
“For humans, the process of continuous learning is essentially a process of building a model of the world in every profession and in every environment,” Hsu said. “We believe that for an agent to become an expert, it must learn autonomously to build a model of a given microscopic world.”
Su believes this ability to specialize quickly is the key missing piece to ensuring AI works independently.
It is possible to train agents for autonomous tasks, but they must be custom designed for specific industries. NeoCognition is different because we are building agents that are generalists that can self-learn and specialize in any area.
NeoCognition plans to sell the agent system primarily to enterprises, including established SaaS companies, who can use the agent system to build agent workers or enhance existing product offerings.
For this reason, Su emphasized that the investment from Vista Equity Partners is particularly valuable. As one of the largest private equity firms in the software space, Vista can provide NeoCognition with direct access to a vast portfolio of companies looking to modernize their products with AI.
NeoCognition currently has approximately 15 employees, the majority of whom have Ph.D.
If you buy through links in our articles, we may earn a small commission. This does not affect editorial independence.
