Banque des Territoires deploys sovereign RAG data solution with Hugging Face for EduRénov

AI Impact Summary

BDT, Polyconseil, and Hugging Face are delivering a sovereign RAG-based assistant for the EduRénov program, leveraging French-governed compute to meet data sovereignty requirements. The solution uses Hugging Face Text Generation Inference, Transformers, Sentence Transformers, and Tokenizers to index EduRénov documentation and assemble prompts via an Orchestrator, Retriever, and Reader workflow, with human-in-the-loop RLHF validation. This design focuses on in-country hosting and secure data handling, signaling a practical pattern for public-sector AI pilots that require strict governance. The outcome should reduce email handling time and accelerate response quality to local authorities while maintaining traceability to cited sources.