Intelligence, Redefined
Symbolica is an artificial intelligence research lab that is pioneering the application of category theory and type theory to build a symbolic reasoning engine.
We have launched Agentica, an open-source library for composable agent workloads. Try it now.
Current state-of-the-art models do not directly incorporate code execution as a first class primitive at the architectural level. In order to build a general purpose reasoning engine we must combine symbolic and neural intelligence within the architecture itself. We believe this is the most important outstanding problem in artificial intelligence research. Several fundamental breakthroughs are required to achieve this goal.
Applied Category Theory
Symbolica is a pioneering AI research lab founded in 2022. We are a team of software engineers and mathematicians leveraging a form of abstract mathematics, known as category theory, to deliver exceptional breakthroughs in modern artificial intelligence research.
Semantics
Our models integrate formal logic and semantics at the architectural level, providing provable invariants of model behavior under all conditions.
Reasoning
Unlike large language models, our category-theoretic architectures embed reasoning into the architecture itself. Program execution, search, and synthesis all occur in the same architecture, fundamentally blurring the line between neural, symbolic, and reasoning.
Categories
By leveraging category theory to inform the construction of our architectures we are able to seamlessly unify discrete concepts like types and programs with fuzzy concepts like vector spaces and optimization with a degree of sophistication inaccessible to other architectures.
Introducing Agentica
Agentica provides agents with the ability to iteratively decompose, investigate, and dynamically solve problems by equipping them with the single most expressive mechanism: arbitrary code execution.
Leveraging types and composition, two of the fundamental building blocks of category theory, we enable users to build multi-agent systems that deliver state of the art performance on top of any model.
Careers
Symbolica is a world-class team of researchers and engineers, with offices in San Francisco and London.
We are united by deep experience in building mathematically grounded AI systems, bringing experience from leading institutions and companies including Tesla AI, Meta, HFT Funds, Snap Research, TikTok, Wolfram, Imperial College London, Cambridge, and Oxford.