Azure OpenAI Embeddings

Azure OpenAI Embeddings can be utilized natively within the Lyntaris V2 platform via the Document Stores (/vector-databases) configuration tab.

Because Lyntaris employs a securely isolated Shared Credentials architecture, there is no need to manually copy and paste API keys into individual configurations.

🏗️ Configuration Protocol

To assign Azure OpenAI Embeddings to a vector database:

1. Navigate to Vector Databases in the left-hand menu.
2. Select your desired Document Store (e.g., Weaviate, Qdrant) and enter the Configure Embeddings step.
3. If not already selected, click Other Embedding Providers and choose Azure OpenAI Embeddings.
4. In the configuration popup, input your specific deployment details:
   • Instance Name: Enter your unique Azure regional deployment instance name (e.g., lyntaris-eastus).
   • Deployment Name: Enter the embedding model designation you provisioned (e.g., text-embedding-ada-002 or text-embedding-3-small).
   • API Version: Select the stable Microsoft Azure release iteration (e.g., 2024-02-15-preview).

🔑 Shared Credentials (Crucial Step)

The most critical step in V2 architecture is binding the proper environment-level credential.

In the Connect Credential dropdown, locate and select the Shared Azure Speech or Shared Azure OpenAI credential that was provisioned by your deployment sync (see Credentials).

[!WARNING] DO NOT click "Create New" or manually enter raw API keys unless specifically instructed for pipeline-scoped overriding. Always use the pre-loaded Shared Credentials.

Resources

• Azure OpenAI Service REST API reference
• Lyntaris Credentials Architecture