Fundamental research for AI
that companies can trust.
Most companies want to use AI on their own data, but can't rely on what it gives them. We study how mathematics and AI can work together so that the answers are right, the data stays private, and the system is affordable to run.
Deterministic computation, complemented by AI
Computation
Mathematics and algorithms do the parts that must be right: structure, counting, compression and symmetry. The same input gives the same output, every time.
AI
AI does what mathematics can't: understanding what people ask, and explaining results in their own terms.
Research at the boundary
We study where the line between the two should sit, so that each does what it is good at and the whole system can be trusted.
Two products from our research so far
Question answering over your own data
A neuro-symbolic system that keeps a very large knowledge graph in one small file and answers plain-language questions from it, checked against the data, on your own hardware.
Explore SACPKG →Network structure, found exactly and explained
Finds which parts of a network play exactly the same role. An insight agent explains what that means in your field, and a chat agent answers your follow-up questions.
Explore Automorph →Working with sensitive data or complex networks?
Tell us what you're trying to do. We're happy to run a live demo on data like yours.