Berlin-based AI infrastructure firm Cognee has successfully closed a $7.5 million seed funding round to advance its pioneering knowledge engine. Led by pebblebed, a venture firm founded by former leaders from OpenAI and Facebook AI Research, the investment will scale Cognee’s mission to provide AI systems with long-term, structured memory. This technology enables AI to retain context, understand complex relationships, and learn from interactions over time, addressing a fundamental limitation in current systems.
Solving AI's Amnesia Problem
Most artificial intelligence systems today operate with a significant handicap: they lack persistent memory, effectively starting fresh with each new interaction. This stateless nature is a primary cause of common AI failures, such as hallucinations and an inability to maintain context in complex conversations. According to Cognee Founder and CEO Vasilije Markovic, AI systems fail not because they lack processing power, but because they are fundamentally unable to remember.
A New Architecture for AI Memory
Cognee tackles this challenge by transforming unstructured data into a dynamic and persistent memory layer built upon knowledge graphs. The company's proprietary pipeline ingests information from numerous sources, structuring it into a network of interconnected facts and relationships. A unique feedback mechanism then allows the memory to become more accurate with each use, enabling AI agents to continuously improve their understanding and performance.
This innovative engine unifies three distinct storage layers—relational, vector, and graph—into a single, cohesive system. By doing so, it empowers AI agents to reason across vast and interconnected datasets, significantly enhancing their reliability in production environments. The platform is designed for seamless integration with existing AI development tools, including the agent SDKs from OpenAI and Google, ensuring developers can easily incorporate its capabilities.
Strong Backing and Market Momentum
The funding round was led by pebblebed, a firm established by OpenAI co-founder Pamela Vagata and Facebook AI Research Lab founder Keith Adams, signaling strong confidence from industry pioneers. Their support underscores the critical need for a dedicated memory layer to advance AI from experimental tools to robust products. The round also drew participation from 42CAP, Vermilion Cliffs Ventures, and angel investors from Google DeepMind and other leading tech firms.
Cognee's successful raise is part of a broader investment trend across Europe, where capital is increasingly flowing into foundational AI infrastructure rather than just end-user applications. This strategic focus highlights the continent's strength in building deep technology and positions Cognee as a key contributor to the global AI stack. The company aims to provide a core building block that countless AI systems can depend on for reliable memory functions.
Rapid Growth and Future Ambitions
Since its inception as an open-source project in 2024, Cognee has experienced remarkable growth, transitioning into production infrastructure for over 70 companies, including major enterprises like Bayer. Its technology is being applied to complex tasks such as powering scientific research workflows and building evidence graphs from scattered policy documents. The project's popularity is further evidenced by its more than 12,000 stars on GitHub and a vibrant community of contributors.
With the new capital, Cognee plans to accelerate its product roadmap on several fronts. Key initiatives include launching a scalable cloud platform to make its technology more accessible, developing a high-performance Rust engine for on-device AI, and deepening its research into cognitive memory. The company will also continue to enhance its open-source core, with plans to add multi-database support and dozens of new data source connectors.
This $7.5 million investment marks a significant milestone for Cognee and the broader pursuit of more capable artificial intelligence. By focusing on the fundamental problem of memory, the company is providing a critical missing piece needed to build truly intelligent and reliable AI agents. As the industry moves toward more complex and autonomous systems, Cognee's knowledge engine is poised to become an essential layer in the next generation of AI.