RAG typically assumes centralized access to documents, which breaks down when knowledge is distributed across private data silos. We propose a secure Federated RAG system built using Flower that performs local silo retrieval, while server-side aggregation and text generation run inside an attested, confidential compute environment, enabling confidential remote LLM inference even in the presence of honest-but-curious or compromised servers. We also propose a cascading inference approach that incorporates a non-confidential third-party model (e.g., Amazon Nova) as auxiliary context without weakening confidentiality.
@article{arxiv.2603.25374,
title = {Supercharging Federated Intelligence Retrieval},
author = {Dimitris Stripelis and Patrick Foley and Mohammad Naseri and William Lindskog-Münzing and Chong Shen Ng and Daniel Janes Beutel and Nicholas D. Lane},
journal= {arXiv preprint arXiv:2603.25374},
year = {2026}
}