NeoCognition Raises $40M for Self-Learning AIs

⚡

Key Takeaways

1NeoCognition, led by Yu Su, secures $40 million to develop self-learning AI agents.

2These agents aim to outperform current generalists, which only succeed in 50% of the tasks requested.

3NeoCognition targets SaaS companies to integrate its specialized agents into their products.

💡Why it matters — NeoCognition could transform the reliability of AI in business, making automation more effective and personalized.

NeoCognition Attracts Investors with an Ambitious Vision

In a context where artificial intelligence researchers are increasingly pressured to turn their discoveries into commercial applications, Yu Su, a professor at Ohio State University, has long resisted this pressure. Head of a lab specializing in AI agents, Su ultimately decided to take the plunge last year. He realized that advancements in foundational models now allowed for the creation of truly personalized agents, leading him to found NeoCognition.

NeoCognition, a startup that Yu Su describes as a research lab focused on developing AI agents capable of self-learning, has just secured an initial funding of $40 million. This funding round was co-led by Cambium Capital and Walden Catalyst Ventures, with notable participation from Vista Equity Partners, as well as influential individual investors such as Lip-Bu Tan, CEO of Intel, and Ion Stoica, co-founder of Databricks.

The Challenge of Current AI Agents

According to Yu Su, the AI agents available today are essentially generalists. “Every time you ask them to perform a task, you take a risk,” he told TechCrunch. The main issue lies in their lack of consistency and reliability. Agents from platforms like Claude Code, OpenClaw, or Perplexity manage to complete their tasks correctly only about 50% of the time.

This low reliability prevents these agents from being considered trustworthy autonomous workers. NeoCognition aims to change this by developing agent systems that learn autonomously and become experts in any field, thereby mimicking the human learning process.

The Human Approach to Learning

Yu Su emphasizes that human intelligence, while vast, derives its true strength from our ability to specialize. When an individual enters a new environment or profession, they can quickly assimilate its unique rules, relationships, and consequences. NeoCognition strives to create agents that reflect this human capability.

“For humans, our continuous learning process is essentially the process of building a model of the world for any profession, any environment,” explains Su. He believes that for AI agents to become reliable experts, they must autonomously learn to construct a model of any given micro-world.

Towards Specialized AI Agents

While training agents for autonomous tasks is possible, it generally requires custom design for each sector. NeoCognition stands out by developing generalist agents capable of autonomously learning and specializing in any field.

The startup plans to market its agent systems primarily to businesses, particularly those in the SaaS sector, which could use them to create agent workers or enhance their existing product offerings.

Strategic Support for NeoCognition

Vista Equity Partners' investment is particularly strategic for NeoCognition. As one of the largest private equity firms in the software sector, Vista provides NeoCognition with direct access to a vast portfolio of companies looking to modernize their products through AI.

Currently, NeoCognition employs about 15 people, the majority of whom hold PhDs, reflecting the expertise and potential of the startup to transform the landscape of artificial intelligence in business.