AI Management - RAG documentation#24968
Conversation
Add a tip to the "Adding RAG Dependencies" section clarifying that RAG is entirely optional and that other AI Management features work without installing RAG-related dependencies. This clarifies expectations for users about required packages.
There was a problem hiding this comment.
Pull request overview
This PR enhances the AI Management (Pro) module documentation by expanding and clarifying how to set up and use RAG (Retrieval-Augmented Generation) features, including dependencies, workspace configuration, data source management, and related permissions.
Changes:
- Added a new section describing required RAG dependencies (embedding + vector store providers) and the built-in provider names.
- Expanded workspace/UI docs with RAG-related fields, a Workspace Data Sources page description, and a detailed RAG configuration + indexing flow.
- Updated RAG provider support details (MongoDb/Pgvector/Qdrant) and expanded the documented permission set.
|
|
||
| #### Chat Integration Behavior | ||
|
|
||
| When a workspace has embedder configuration, AI Management wraps the chat client with a document search tool function named `search_workspace_documents`. |
There was a problem hiding this comment.
Within this new section, you state RAG is enabled only when both embedder and vector store are configured, but here the condition is described as only “has embedder configuration”. Please align the wording (e.g., require both embedder + vector store / “RAG enabled”) to avoid confusing readers about when the search_workspace_documents tool is available.
| When a workspace has embedder configuration, AI Management wraps the chat client with a document search tool function named `search_workspace_documents`. | |
| When RAG is enabled for a workspace (that is, it has both embedder and vector store configuration), AI Management wraps the chat client with a document search tool function named `search_workspace_documents`. |
| | `VectorStoreSettings` | No | Vector store connection string or settings | | ||
| | `EmbedderProvider` | No | Embedding provider name (e.g., "OpenAI", "Ollama") | | ||
| | `EmbedderModelName` | No | Embedding model identifier (e.g., "text-embedding-3-small") | | ||
| | `EmbedderApiKey` | No | API key for embedding provider | |
There was a problem hiding this comment.
The workspace properties table documents EmbedderApiKey, but the later “Configuring Embedder” section doesn’t mention how/where to set the embedder API key (it only lists provider/model/base URL). This leaves the RAG embedder configuration steps incomplete/inconsistent within the same page; please update that section to include the API key (or clarify when it’s not required).
| | `EmbedderApiKey` | No | API key for embedding provider | | |
| | `EmbedderApiKey` | No | API key for embedding provider (only needed if different from the main `ApiKey` or provider defaults) | |
|
|
||
| #### Workspace Data Sources | ||
|
|
||
| Workspace Data Sources page is used to upload and manage RAG documents per workspace. Uploaded files are processed and indexed in the background. |
There was a problem hiding this comment.
This sentence is missing an article and reads awkwardly (“Workspace Data Sources page is used…”). Consider changing it to “The Workspace Data Sources page is used …” for grammatical correctness.
| Workspace Data Sources page is used to upload and manage RAG documents per workspace. Uploaded files are processed and indexed in the background. | |
| The Workspace Data Sources page is used to upload and manage RAG documents per workspace. Uploaded files are processed and indexed in the background. |
Description
Enhances documentation for RAG usage
Checklist