Intel® AI for Enterprise RAG Documentation
The technical documentation for the RAG application layer. Start with Meet for the what and why, Get started to stand it up, Deploy to choose a pipeline and go further, then customize, operate, and look things up as needed.
Important
This layer is a component of Intel® AI for Enterprise Solutions. Clone that repository, run every command from its root, and init erag will fetch this one into ext/ for you. See Meet.
Meet
| Guide | What it covers |
|---|---|
| Meet Intel® AI for Enterprise RAG | What it is, why it exists, when to choose it, and which repository to start from |
| FAQ | Short answers to the questions that come up most |
| Glossary | Every term used in these guides, defined once |
Get started
Check your machine meets the bar, then deploy the stack and ingest a document.
| Guide | What it covers |
|---|---|
| Prerequisites | RAG hardware sizing, software access, and the limited single-user option |
| Quick Start | init erag to a working assistant, end to end |
Deploy
Install the layer, pick a pipeline, and take it onto a partner platform.
| Guide | What it covers |
|---|---|
| Deploy the RAG Layer | Step-by-step install, endpoints, credentials, and teardown |
| Pipelines | The five flavours, ChatQnA variants, and how to switch between them |
| Deploy on VMware | Running the chatbot on VMware vSphere |
| Deployment Layer Reference | What the deployment/ plug-in contributes and how the installer consumes it |
Customize
Change models, wire up identity and external data sources, and tune every knob.
| Guide | What it covers |
|---|---|
| Configuration | Every option in config.erag.yaml, and how overrides resolve |
| Models | Changing the LLM, embedding, and reranking models; multilingual and accuracy features |
| Authentication | Keycloak SSO with Microsoft Entra ID, multi-factor auth, and Active Directory federation |
| SharePoint | Ingesting from SharePoint Online, scheduled sync, and per-user site filtering |
| MCP Integration | Exposing retrieval and ingestion to AI agents over Model Context Protocol |
| Object Store | EDP storage backends, and serving documents from NetApp ONTAP S3 |
| Building Images | Building the component images locally and pointing the deployment at your registry |
Operate
Watch it, tune it, and fix it once it is running.
| Guide | What it covers |
|---|---|
| Telemetry | The RAG Grafana dashboards and where to find logs |
| Performance | Per-step resources, replica sizing, autoscaling, and vector-database tuning |
| Backup and Restore | Enabling it, running a backup or restore, what is and is not captured, and the identity realm |
| Troubleshooting | Common symptoms, their causes, and the debug tool |
Reference
How the layer is built and what it exposes.
| Guide | What it covers |
|---|---|
| Architecture | Components, roles, namespaces, pipeline composition, and request flow |
| Container Images | Every image the deployment pulls, with its build context |
| Service Mesh | How RAG namespaces join the Istio ambient mesh, and the current mTLS posture |
User guides
Feature and usage walkthroughs for each app, with screenshots.
| Guide | App |
|---|---|
| AudioQnA User Guide | Voice question answering |
| ChatQnA User Guide | Conversational retrieval |
| DocSum User Guide | Document summarization |