News

Startups

Artificial Intelligence

Europe

Cognee Secures $7.5 Million Seed to Build Permanent Memory for AI Agents

Cognee closes a $7.5 million seed round led by Pebblebed to solve the problem of amnesia in autonomous AI agents.

Cognee closes a $7.5 million seed round led by Pebblebed to solve the problem of amnesia in autonomous AI agents.

Cognee closes a $7.5 million seed round led by Pebblebed to solve the problem of amnesia in autonomous AI agents.

NewDecoded

Published Feb 21, 2026

Feb 21, 2026

3 min read

Image by Cognee


Funding the Memory Layer\n\nBerlin and San Francisco-based startup Cognee has raised $7.5 million in a seed funding round led by Pebblebed. The investment includes participation from 42CAP, Vermilion Ventures, and prominent figures from Google DeepMind and n8n. Led by the founders of OpenAI and Facebook AI Research, Pebblebed is backing Cognee to create a foundational memory layer for the next generation of AI agents.\n\n## Solving the Amnesia Problem\n\nCurrent AI models often operate like a blank slate during every new session. This stateless nature leads to frequent hallucinations and an inability to retain user context over time. Cognee addresses these limitations by providing a structured memory engine that allows agents to learn and reason based on historical data.\n\n### The ECL Pipeline\n\nThe technology relies on a proprietary ECL pipeline which stands for Extract, Cognify, and Load. By transforming raw data into a dynamic knowledge graph, Cognee enables agents to connect facts and identify complex relationships. A unique Memify layer incorporates user feedback to sharpen these memories, ensuring that the most relevant information is prioritized during retrieval.\n\n## Proven Market Traction\n\nMarket adoption has accelerated rapidly, with the company recording over one million pipeline runs in the past year. High-profile organizations such as Bayer and the University of Wyoming already use the platform to manage complex research and policy workflows. The open-source community has also shown strong support, with the project surpassing 12,000 stars on GitHub.\n\n## Scaling the Infrastructure\n\nThe new capital will fuel the launch of a cloud platform for managed memory services. Development is also underway for a high-performance Rust engine designed specifically for edge devices and local privacy. Furthermore, the team plans to expand its research into cognitive science to translate human memory models into production-ready software tools.

Decoded Take

Decoded Take

Decoded Take

The rise of autonomous agents has hit a ceiling where raw compute and large context windows can no longer compensate for a lack of persistent state. By moving beyond simple vector search into structured knowledge graphs, Cognee is positioning memory as a distinct architectural layer rather than a mere database feature. This shift suggests that the industry is moving away from chatbots toward true digital coworkers that can develop long-term expertise and situational awareness. With backing from the architects of OpenAI and FAIR, this funding signals that structured cognitive architecture is becoming the new standard for enterprise-grade AI reliability.

Share this article

Related Articles

Related Articles

Related Articles