Can I use Qdrant payload filters alongside Gateco policies?

Yes. Qdrant payload filters and Gateco policies are complementary. Payload filters are applied at the vector DB level for static data partitioning. Gateco policies are applied at the retrieval layer using live principal data from your IDP, with a complete audit trail. Both can be active simultaneously.

Is Qdrant keyword search ranked?

No. Qdrant keyword search uses match.text, which assigns uniform score 1.0 to all matches. There is no BM25 or TF-IDF ranking. If ranked keyword relevance matters, use pgvector or OpenSearch.

Does Qdrant support regex grep?

No. Substring grep (match.text) is supported. Regex grep is not — Qdrant's payload matching does not support arbitrary regular expressions. Use pgvector, Supabase, Neon, or OpenSearch for regex pattern matching.

Vector DBTier 1

Qdrant

Policy-enforced retrieval on Qdrant with rich payload filtering and native sparse+dense hybrid search.

Search capabilities

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

Qdrant in-memory: <10ms p95 ANN. On-disk: <50ms p95 at 1M vectors. Policy overhead <25ms p95.

Ingestion (Tier 1)Retroactive registrationsidecar metadatainline metadata

Qdrant is a high-performance vector database with first-class payload filtering, named vectors, and native sparse+dense hybrid search. Gateco connects to Qdrant as a Tier 1 connector with ingestion, retroactive registration, and policy enforcement.

Qdrant's payload filtering is useful for static data partitioning, but it operates on payload fields set at index time. Gateco adds the principal dimension: the same Qdrant collection returns different results depending on the requesting user's IDP-synced groups and attributes, with the decision recorded in the audit log.

Keyword search on Qdrant is filter-only: it matches payload text fields with match.text (uniform score 1.0, no BM25 ranking). Hybrid search uses Qdrant's native sparse+dense fusion engine. Substring grep is supported; regex grep is not.

Retroactive registration on Qdrant handles non-existent collections gracefully — the scan returns an empty list (no error) on HTTP 404, so you can register the connector before the collection is created.

Sample search config

{
  "collection_name": "knowledge_base"
}

Sample policy

{
  "name": "Team document access",
  "effect": "allow",
  "rules": [{
    "conditions": [
      {"field": "resource.classification", "operator": "neq", "value": "restricted"},
      {"field": "principal.department", "operator": "in", "value": ["engineering", "product"]}
    ]
  }],
  "selectors": [{"connector_type": "qdrant"}]
}

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

Frequently asked questions

Can I use Qdrant payload filters alongside Gateco policies?: Yes. Qdrant payload filters and Gateco policies are complementary. Payload filters are applied at the vector DB level for static data partitioning. Gateco policies are applied at the retrieval layer using live principal data from your IDP, with a complete audit trail. Both can be active simultaneously.
Is Qdrant keyword search ranked?: No. Qdrant keyword search uses match.text, which assigns uniform score 1.0 to all matches. There is no BM25 or TF-IDF ranking. If ranked keyword relevance matters, use pgvector or OpenSearch.
Does Qdrant support regex grep?: No. Substring grep (match.text) is supported. Regex grep is not — Qdrant's payload matching does not support arbitrary regular expressions. Use pgvector, Supabase, Neon, or OpenSearch for regex pattern matching.

Ready to connect Qdrant?

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

Setup guide →Talk to the team All integrations