NeoCognition Emerges from Stealth With $40 Million Seed Round to Advance Specialized Intelligence and Expert Agents

NeoCognition, an Agent Lab for Specialized Intelligence, led by Yu Su, Xiang Deng, and Yu Gu, has announced its emergence from stealth with a $40 million seed funding round. The oversubscribed round was co-led by Cambium Capital and Walden Catalyst Ventures, with participation from Vista Equity Partners and others. Angel investors and founding advisors include Lip-Bu Tan, CEO of Intel, Ion Stoica, Co-Founder and Executive Chairman of Databricks, A&E Investments, Salience Capital Partners, Nepenthe Capital, Frontiers Capital, and leading AI researchers like Dawn Song, Ruslan Salakhutdinov, and Luke Zettlemoyer, among others.

Yu Su, CEO and Co-Founder of NeoCognition and a Sloan Research Fellow, leads a leading AI agent lab at Ohio State University. The team began developing large language model-based agents well before the rise of ChatGPT, contributing foundational work such as Mind2Web, MMMU, and SeeAct.

Their research has played a key role in shaping the modern AI agent field and is widely used across frontier models from OpenAI, Anthropic, and Google.

“Dr. Su and his team have developed research spanning the full agent stack, from perception and memory to planning, evaluation, and safety,” said Lip-Bu Tan, Founding Managing Partner of Walden Catalyst Ventures. “NeoCognition is well-positioned to tackle some of the most complex challenges in agentic AI.”

Coming out of stealth, NeoCognition is developing a new class of AI agents that learn the structure, workflows, and constraints of their environments, evolving into domain experts through learned world models of work.

This deeper understanding enables faster, more cost-effective, and more reliable performance, while also improving safety and responsibility in high-stakes settings.

“AI today remains unreliable for executing complex, expertise-driven work,” said Yu Su. “A more effective approach mirrors how humans build expertise, by continuously learning and developing structured models of their environments, reducing the need for extensive manual customization.”

“As general-purpose agents become table stakes, the core challenge shifts to achieving expert-level intelligence,” said Ion Stoica, Professor at the University of California, Berkeley and Co-Founder of Databricks. “NeoCognition’s approach to building agents that learn to become experts has the potential to deliver the reliability, efficiency, and cost-effectiveness required for high-stakes applications.”

“At the core of NeoCognition is a novel learning mechanism that enables rapid agent specialization,” said Landon Downs, Managing Partner at Cambium Capital. “The team’s research is charting a path toward specialized intelligence that broadens access to advanced agent capabilities.”