Skip to main content

Knowledge Base

The Knowledge Base page manages enterprise RAG (Retrieval-Augmented Generation) data sources and vector search datastores.

By linking enterprise search indices to team credentials, Aegis enables AI agents to ground their responses in verified document repositories.

Supported Source Types​

Aegis supports enterprise vector search integration:

Source TypeIdentifierConfiguration Parameters
Google Vertex AI Searchvertex_ai_searchgcp_project_id, location (e.g., global, us), datastore_id

Cloud Credential Integration​

Knowledge Bases link directly to Team Cloud Credentials (TeamCloudCredential), allowing credential rotation without breaking individual RAG datastore configurations:

  • Service Account Key (sa_key): Authenticates using uploaded GCP Service Account JSON key.
  • Application Default Credentials (adc): Secretless authentication relying on GCP workload identity.

Managing Knowledge Bases​

Adding a Knowledge Base​

  1. Navigate to Agent Workspace > Knowledge Base in the Aegis Console.
  2. Click + Add Knowledge Base.
  3. Fill in the connection properties:
    • Name & Alias: Descriptive display title (e.g., HR Policies Datastore).
    • Source Type: Select Vertex AI Search.
    • GCP Project ID: Target Google Cloud project ID.
    • Location: Datastore location (e.g., global or us).
    • Datastore ID: Vertex AI Search datastore identifier.
    • Team Credential: Select active GCP cloud credential from your vault.
  4. Click Save Knowledge Base.

Interactive Search Query Testing​

Select any Knowledge Base card and enter a search query in the Test Query panel to evaluate document retrieval snippets, grounding scores, and source document links before binding to agents.