NeoCognition is an AI research lab building a new class of agents that continuously learn the structure, workflows, and constraints of the environments they operate in, then specialize into domain experts by learning what the team calls a "world model of work." The goal is agents that are faster, cheaper, and more reliable than general-purpose LLM agents in real enterprise settings.
The company emerged from stealth in April 2026 with a $40 million seed round 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. AI researchers Dawn Song, Ruslan Salakhutdinov, and Luke Zettlemoyer are among the founding advisors, signalling NeoCognition's positioning at the frontier of agent research.
NeoCognition was founded by Yu Su (CEO), Xiang Deng, and Yu Gu, who grew out of one of the most established AI agent labs in the United States at Ohio State University. The team currently sits at about 15 people, the majority holding PhDs, and is headquartered in Palo Alto, California while maintaining deep research ties to academia.
Commercially, NeoCognition plans to sell its agent systems to enterprises and to SaaS companies that want to embed expert-level agent workers in their own products. Where many startups wrap a foundation model in prompts, NeoCognition's bet is that agents should keep learning on the job — accumulating skills, conventions, and policies inside a specific workflow — to reach reliability levels that general-purpose models cannot reach alone. The product is still early, but the team, capital, and advisor bench position the company as one of the more credible early-stage agent labs of the cycle.