About us
Symbolica was founded on a simple idea: intelligence is about structure and reasoning, not just pattern matching.
For decades, AI has been split between symbolic systems that are reliable but brittle and neural systems that are flexible but unpredictable. We’re building a bridge between the two.
We’re an AI research lab combining deep mathematics and machine learning to create systems that truly reason. Using category and type theory as a unifying framework, we develop interpretable, reliable and scalable foundations for intelligence and turn them into real products, working in a tight feedback loop where research directly shapes application.
Agentica is the latest expression of this vision: an agent framework for tool use and multi-agent orchestration through arbitrary code execution. It recently achieved 85.28% on ARC-AGI-2, setting a new public SOTA.
Founded in 2022, we’ve raised over $30M from investors including Khosla Ventures, General Catalyst, Abstract Ventures and Buckley Ventures, backing us to rethink the mathematical foundations of machine learning.
If you’ve ever been kept awake wondering what comes after deep learning, you’ll fit right in.