Founded by alumni of the École Polytechnique, HEC and Cambridge University – Louis de Benoist, Victor Plaisance and Sacha Ichbiah – Retab is the AI infrastructure platform for the reliable processing of unstructured data. The company’s mission is to become the benchmark solution for the developers who will build tomorrow’s world. Retab has assembled a team of researchers and operators from ENS, Polytechnique and Oxford, Retab intends to make AI trustworthy for the most critical needs of global enterprises.
Retab’s AI platform is based firstly on structured generation, a method that provides mathematical guarantees for each output format. The platform then relies on two distinct systems that work in concert to guarantee maximum performance:
- Self-Optimising Schemas: A dedicated AI agent automatically tests and refines the various model parameters, based on the user’s dataset, maximising accuracy before deployment.
- Intelligent Model Routing: Once deployed, the platform dynamically redirects tasks to the best performing LLM for that use case, whether it comes from OpenAI, Google or elsewhere. With this technology, Retab is sometimes 100x cheaper than today’s AI solutions.
This choice process is governed by Retab’s ‘k-LLM Consensus’, which acts as a committee of AI experts debating answers to quantify uncertainty.
