This solution accelerator provides architecture templates and deployment assets to help organizations build secure, scalable, and enterprise-ready agentic RAG solutions on Microsoft Foundry. Orchestration runs on the Microsoft Agent Framework, and Foundry IQ is the default retrieval backend, blending Blob, Azure AI Search, Work IQ, Fabric, SharePoint, OneLake, Web and MCP sources with permission trimming. Azure AI Search direct remains fully supported as a rollback path. It applies proven Azure design patterns with Zero-Trust security, Responsible AI, and end-to-end observability.
For full documentation, visit the GPT-RAG documentation site.
GPT-RAG is built on a Zero-Trust architecture to ensure that all components operate within a controlled, isolated environment. Network access is tightly governed, and communication between services follows least-privilege principles.
Head to the documentation site for the complete guides:
- Grounding sources overview start here: Foundry IQ (default), Azure AI Search direct, Work IQ, Fabric, SharePoint, OneLake, Web and MCP sources, and how to choose.
- Deployment Guide covers Basic, Zero Trust, and network-isolated deployments, preflight checks, jumpbox workflow, and container image builds.
- What's New highlights the notable features added over time.
Zero-Trust Architecture
Hosted continuity uses Responses protocol 2.0.0 with delegated
x-ms-user-identity ownership from the trusted UI BFF. Activation fails closed
until the UI identity has the exact Foundry Agent Consumer role and GPT-RAG
user-identity impersonation role directly at the individual agent scope. The
hosted container receives neither role, no ownership key, and no Cosmos DB in
the no-panel topology. See the shared ownership contract
for role IDs, validation rules, and the disabled capability fallback.
The accelerator supports a broad range of enterprise scenarios, from customer support to decision automation, by enabling systems to process complex queries across large data collections. It is designed for seamless integration into existing environments and can be adapted to both straightforward and advanced operational patterns.
The agent layer is built with the Microsoft Agent Framework on Microsoft Foundry. It supports scenarios such as NL2SQL query generation, multi-source grounding through Foundry IQ, and tool integration via MCP servers, so organizations can build workflows that retrieve, interpret, and act on enterprise data with contextual precision.
GPT-RAG UI
We welcome contributions! See the contribution guidelines for details on how to contribute.
This project may contain trademarks or logos. Authorized use of Microsoft trademarks or logos must follow Microsoft’s Trademark & Brand Guidelines. Modified versions must not imply sponsorship or cause confusion. Third-party trademarks are subject to their own policies.