Weaviate is overkill for my embedding search use case

Probably yes—if all you need is “store vectors + do nearest-neighbor search”, Weaviate can be more than you need.

When Weaviate is overkill

If your use case is simple, like:

• a few thousand to a few million embeddings
• basic cosine / dot-product similarity search
• minimal metadata filtering
• no hybrid search, RAG pipeline, multi-tenancy, schema management, or built-in modules
• you’re comfortable managing your own index/storage logic

then lighter options may be a better fit.

Simpler alternatives

Depending on your stack, consider:

• FAISS
  Best for local/offline vector search. Fast, lightweight, no server required.

• pgvector
  Great if you already use PostgreSQL and want vector search without adding another system.

• Qdrant
  Good middle ground: lighter than Weaviate, but still a dedicated vector DB with strong filtering and APIs.

• Pinecone / managed services
  If you want to avoid infrastructure but don’t need Weaviate’s richer feature set.

• Elasticsearch / OpenSearch vector search
  Useful if you already use them for keyword search and want hybrid retrieval.

When Weaviate makes sense

Weaviate is more attractive if you want:

• a full-featured vector database
• hybrid semantic + keyword search
• metadata filtering at scale
• schema-driven organization
• RAG-friendly features and integrations
• production deployment with richer query patterns

Rule of thumb

• Prototype / small app / simple similarity search: use FAISS or pgvector
• Need vector DB features but want simplicity: Qdrant
• Need integrated search platform + semantics + filters + modules: Weaviate

If you want, I can help you choose the best alternative based on your exact requirements.