Skip to main content

Documentation Index

Fetch the complete documentation index at: https://docs.llmgrid.ai/llms.txt

Use this file to discover all available pages before exploring further.

​
Overview

The Vector Store Management page allows administrators to register and manage vector stores that hold embeddings used for semantic search, retrieval, and grounding. Vector stores are typically used in RAG (Retrieval‑Augmented Generation) scenarios and can be attached to agents, prompts, and workflows. Vector stores are configured centrally and referenced by ID during request execution.

​
Vector Stores List

The main Vector Stores page displays all configured vector stores for the tenant.

​
Primary Actions

  • Add Vector Store
    Opens the configuration dialog to register a new vector store.

​
Vector Stores Table

Each row represents a configured vector store. Columns include:
  • Vector Store ID
    Unique identifier of the vector store (provided by the underlying provider).
  • Name
    Human‑friendly name for easier identification.
  • Description
    Optional description of the store’s purpose.
  • Provider
    Provider backing the vector store.
  • Created At
    Creation timestamp.
  • Updated At
    Last update timestamp.
  • Actions
    Edit or delete the vector store.

​
Vector Store Details

Selecting a Vector Store ID opens the detail view.

​
Details Tab

The Details tab displays:
  • ID – The provider‑assigned vector store ID
  • Name – Display name
  • Description – Optional description
  • Provider – Provider backing the store
  • Metadata – Arbitrary JSON metadata
  • Created – Creation timestamp
  • Last Updated – Most recent update timestamp
Select Edit Vector Store to modify editable fields.

​
Test Vector Store

The Test Vector Store tab allows you to validate connectivity and availability of the configured vector store without impacting production traffic. Use this test during:
  • Initial setup
  • Credential rotation
  • Troubleshooting retrieval issues

​
Add New Vector Store

Select Add Vector Store to register a new vector store.

​
Required Fields

​
Provider

Select the provider that backs the vector store. The selected provider determines which additional fields are required.

​
Vector Store ID

Enter the vector store identifier as defined by the provider. This value is required to reference the existing vector store during retrieval operations.

​
Optional Fields

​
Vector Store Name

A human‑friendly name to identify the vector store in the UI.

​
Description

Optional text describing what data the vector store contains or how it is used.

​
Existing Credentials

Optionally select previously stored credentials instead of entering new connection details. This simplifies reuse across multiple vector stores.

​
Metadata

Optional JSON metadata associated with the vector store. Metadata can be used for:
  • Classification
  • Internal annotations
  • Integration context

​
Provider‑Specific Configuration

Depending on the selected provider, additional fields may appear, such as:
  • API key
  • API base URL
  • Custom connector endpoints
All sensitive values are stored securely and masked after saving.

​
Create Vector Store

Select Create to save the vector store configuration. Once created, the vector store becomes available for:
  • Agents
  • Search and retrieval workflows
  • Prompt and tool invocation scenarios

​
How Vector Stores Are Used

Vector stores are used to:
  • Store embeddings for documents or records
  • Perform semantic similarity search
  • Ground model responses with contextual data
  • Enable retrieval‑augmented generation (RAG)
Vector stores do not generate embeddings themselves—they store and retrieve them for use during requests.

​
Best Practices

  • Use clear naming to reflect stored data or use case
  • Test vector stores immediately after creation
  • Reuse credentials where possible
  • Attach vector stores only to trusted workflows
  • Review metadata usage for maintainability
  • Periodically remove unused vector stores

​
Common Use Cases

  • Knowledge‑base retrieval
  • Document search
  • Contextual grounding for agents
  • Internal or external semantic search
  • Enterprise RAG implementations
  • Agents – Use vector stores for retrieval
  • Search Tools – Combine search with vector retrieval
  • Prompts – Include retrieved context in prompts
  • Models – Generate embeddings and responses
  • Logs – Debug vector‑backed requests