Vector Upsert

End-to-end pipeline: dataset → embeddings → vector DB upsert.

When to invoke

• User wants semantic search over their data.
• database-guide recommended a vector backend.
• Existing embeddings need to be refreshed.

Supported backends

• Pinecone (managed cloud).
• Qdrant (self-hosted or cloud).
• Weaviate (self-hosted or cloud).
• Milvus / Zilliz.
• pgvector (Postgres extension).
• ChromaDB (local / embedded).