cognee Joins UC Berkeley Xcelerator's 2026 Agentic AI Cohort
We're excited to share that cognee has been selected for the Berkeley Xcelerator 2026 Spring Cohort, run by Berkeley RDI — UC Berkeley's Center for Responsible, Decentralized Intelligence.
The Xcelerator is a non-dilutive program built to help founders turn frontier AI ideas into venture-backable companies. Over three years it's incubated more than 110 global teams, who've gone on to raise over $650M in follow-on funding, with founders spanning 100+ countries. The 2026 cohort is focused on agentic AI, and cognee joins a strong lineup that includes Narada AI, RELAI, and Headroom.
Berkeley RDI described cognee as "an ontology-grounded context engine for your agents, enabling them to reason and improve over time." Reading that back, it's a pretty fair summary of what we've spent the last two years building: a durable, structured memory layer that gives agents something to actually reason over, not just retrieve from.
Why Berkeley Selected cognee
Berkeley RDI has been thinking hard about what's missing from the current generation of AI agents, and a lot of it comes down to memory. Without structured context that persists across sessions, shared memory they can coordinate around, and a way to reason over how facts connect, agent systems struggle to support the more complex use cases emerging across today's industries.
Simple vector stores and short context windows lose state the moment a session ends. They were never built to model relationships across everything an agent has learned, which is exactly what we've designed cognee for.
cognee was founded in Berlin in 2024 by Vasilije Marković, who came to this problem through a mix of data engineering and cognitive science, with prior roles at Zalando, Taxfix, and Omio. The platform turns raw documents and data into vectors and knowledge graphs — an ontology-grounded memory layer built for persistent context and multi-agent coordination. That's the architecture Berkeley pointed to specifically.
"Proud to announce cognee has joined Berkeley Xcelerator," said Vasilije, Founder and CEO in a recent statement.
What the Program Gives Us
Beyond the recognition, the Xcelerator comes with real resources: direct access to Berkeley's frontier research, mentorship from faculty and top investors, and a community of 40,000+ builders and researchers to learn from. We're also picking up cloud and GPU credits — up to $10,000 each from OpenAI, Google Cloud and DeepMind, and Nebius — without giving up any equity.
Who cognee Is For
If you're building AI agents that need to hold onto context, reason over connected data, or coordinate across a multi-agent system, we built cognee for you. Berkeley's selection is independent, research-grade validation of our approach, and we're just getting started.
Star us on GitHub, read our docs, and try cognee Cloud today.
