Qdrant is a vector store for keeping document embeddings and retrieving documents similar to a question. Spring AI provides a common abstraction, so changing vector stores has little effect on application development. section 4.2 section 4.4
A practical setup runs Qdrant with Docker Compose. Expose ports 6333 and 6334, start the service, and configure Spring AI to use the local instance. section 4.2
services:
qdrant:
image: 'qdrant/qdrant:latest'
ports:
- '6334:6334'
- '6333:6333'
Use this Compose definition to start Qdrant and expose the HTTP and client ports used by the application. section 4.2
A document pipeline can use Qdrant to:
- Store chunks produced from game-rule documents. section 4.3.4
- Organize stored chunks in a named collection; Qdrant calls individual chunks points. section 4.3.4
- Support RAG by retrieving documents similar to a question and supplying them as prompt context. section 4.4
Verify an ingested collection
After loading Burger Battle rules, query Qdrant’s collections endpoint and count points in board-game-buddy. The example reports one collection containing two points.
