Skip to content

Embeddings

Configure embedding models for RAG (Retrieval-Augmented Generation), semantic search, and knowledge repository vector operations.

Overview

Embedding models convert text into numerical vector representations, enabling semantic search and similarity matching. They are a core component of the knowledge retrieval pipeline — when a user asks a question, the embedding model finds the most relevant document chunks from the knowledge repository.

Embeddings are configured at the workspace level under Settings > AI Models > Embeddings, and are used by:

  • Knowledge Repositories — for indexing and searching documents
  • Vector Stores — for storing and querying document embeddings
  • Retrieval Tool — for finding relevant context during conversations

Adding an Embedding Model

Embeddings List

  1. Navigate to Settings > AI Models > Embeddings
  2. Click "Create Embedding"
  3. Select a provider from the available options
  4. Configure the embedding settings
  5. Click "Save"

Supported Providers

Add Embedding

Provider Description
Google AI Embeddings Google's text embedding models
Voyage Embeddings Voyage AI's embedding models
OpenAI Embeddings OpenAI's text embedding models (e.g., text-embedding-3-small, text-embedding-3-large)
Azure OpenAI Embeddings Azure-hosted OpenAI embedding models

Configuration

Embedding Configuration

Each provider requires:

Field Required Description
Embedding Name Yes A descriptive name for this embedding configuration
Connect Credential Yes Select the provider credential configured under Settings > Credentials
Model Name Yes Select an embedding model from the dropdown
Azure OpenAI API Deployment Name Yes (Azure only) The deployment name of the embedding model in your Azure OpenAI resource. Only shown when using Azure OpenAI Embeddings

Model Availability

The list of available models is dynamic and depends on the selected provider and credential. New models are added as providers release updates.