We are building an AI you can trust by design: a new foundational model based on a fundamentally new non-transformer architecture. Inspectable. Explainable. Varifiable.
Current Status
PoC. This is a deep-tech bet on new foundational architecture.
Problem or Opportunity
Current AI generates, hallucinates and goes of the rails. We cannot inspect what it learned and why it answer the way it did. This makes its use in domains like legal, regulatory, compliance, medical or similar problematic.
Solution (product or service)
We are building an AI you can trust by design: a new foundational model based on a fundamentally new non-transformer architecture. Inspectable. Explainable. It learns customer's domains and ground all answers in learned facts. Customers can inspect what it learned and easily fix it if decission comes out worng.
Business model
Anthropic for lawyers, tax advisors, accountants, regulatory and compliance people. We will sell them AI like existing providers, but much better suited to their business. Think of TypeSafe.AI but for regualted industries.
