A
Vector DBTier 2

Azure AI Search

Policy-enforced retrieval on Azure AI Search with native RRF hybrid and ranked BM25 keyword support.

Search capabilities

Vector (ANN)Keyword (ranked)Keyword (filter)Hybrid (native)Grep (substring)Grep (regex)

Azure AI Search S1+: <100ms p95 typical. Policy overhead <25ms p95. Use Basic tier for development only.

sidecar metadatainline metadata

Azure AI Search is Microsoft's fully managed search service supporting vector, keyword, and hybrid retrieval. Gateco connects to Azure AI Search as a Tier 2 retrieval connector using the 2025-05-01-preview REST API.

Azure AI Search's native hybrid mode uses Microsoft's RRF fusion of vector and keyword results. Gateco passes queries to the vectorQueries + search parameters of the Azure REST API. The alpha parameter is accepted by Gateco but does not map to an Azure-side control — native RRF fusion proceeds unchanged.

BM25 keyword search is fully ranked (not filter-only). Azure's lexical analysis pipeline (analyzers, tokenizers) applies to keyword queries. Semantic ranking fields are reserved in Gateco search_config for future support.

Grep is not supported for Azure AI Search in Gateco v1. Azure Lucene supports wildcards and regex, but behavior is analyzer-dependent and not deterministic across index configurations.

Sample search config

{
  "service_name": "my-search-service",
  "index_name": "knowledge-base",
  "vector_field": "content_vector",
  "content_field": "content",
  "api_version": "2025-05-01-preview"
}

Sample policy

{
  "name": "Azure AD group access",
  "effect": "allow",
  "rules": [{
    "conditions": [
      {"field": "principal.groups", "operator": "contains", "value": "az-knowledge-readers"},
      {"field": "resource.classification", "operator": "in", "value": ["internal", "public"]}
    ]
  }],
  "selectors": [{"connector_type": "azure_ai_search"}]
}

Policy conditions reference resource.* and principal.* fields. Policy reference →

Frequently asked questions

Does Gateco use service_name or a full URL?
service_name only — pass the subdomain (e.g. my-search-service), not the full https:// URL. Gateco constructs the endpoint as https://{service_name}.search.windows.net.
Which api_version should I set?
Gateco defaults to 2025-05-01-preview. Older versions use incompatible vector query syntax. Do not set api_version to 2024-07-01 or earlier.
Can I pair Azure AI Search with Azure Entra ID?
Yes — this is the natural Microsoft-stack pattern. Configure Entra ID as your IDP to sync users and groups. Configure Azure AI Search as your vector DB connector. Gateco policies then gate search results based on Entra group membership.

Ready to connect Azure AI Search?

Follow the step-by-step setup guide or talk to the team for help with your specific configuration.

Setup guide →Talk to the teamAll integrations