Vector Database¶
Vector Database is the RAG preparation and retrieval-validation area.
It gives you an end-to-end environment for document ingestion, chunking, embedding, storage, and similarity search.
What It Supports¶
This area acts as a vector database playground built on Spring AI vector store integrations.
That includes:
- switching between vector providers without changing application code
- using a unified Spring AI retrieval model
- validating retrieval quality before relying on it in chat
Support for Major Vector Database Providers¶
Spring AI Playground follows the Spring AI vector store ecosystem and can be used with providers such as Apache Cassandra, Azure Cosmos DB, Azure Vector Search, Chroma, Elasticsearch, GemFire, MariaDB, Milvus, MongoDB Atlas, Neo4j, OpenSearch, Oracle, PostgreSQL/PGVector, Pinecone, Qdrant, Redis, SAP Hana, Typesense, Weaviate, and others supported by Spring AI.
Major Capabilities¶
- Custom Chunk Input: enter raw text and test chunking directly
- Document Uploads: ingest PDF, Word, and PowerPoint-style content
- End-to-End Processing: extraction, chunking, embedding, and indexing
- Search and Scoring: run vector similarity search and inspect scores
- Spring AI Filter Expressions: narrow searches using metadata conditions
Why It Matters¶
RAG often fails quietly when chunking, embeddings, or indexing are misaligned. This screen exists so those problems become observable:
- you can see whether ingestion completed
- you can inspect chunk quality
- you can verify retrieval relevance
- you can catch embedding-model changes that invalidate old vector data
That is why the desktop launcher warns users about changing embedding models after indexing content.
In practice, this is what turns the Vector Database page into a real RAG validation surface rather than a generic upload page. You can inspect ingestion quality, retrieval quality, and filter behavior before trusting the same data inside chat.
→ Next: Agentic Chat — compose tools and retrieved documents into a single chat session.