Vector Search Feature

[inalbilal]

Is your feature request related to a problem? Please describe.
Quickwit currently does not support vector search, which limits its ability to perform similarity-based searches. Given its high-speed search capabilities, integrating vector search could unlock new use cases, especially in AI and machine learning applications.

Describe the solution you'd like
I would like to see native support for vector search in Quickwit, enabling efficient similarity searches. This could include an integration with existing vector database solutions or direct support for storing and querying vector embeddings. Additionally, making Quickwit compatible with LangChain’s vector store integrations (such as FAISS, ChromaDB, Weaviate, and others) would allow developers to use it seamlessly in AI-driven applications, including chatbots, recommendation systems, and retrieval-augmented generation (RAG) pipelines.

Describe alternatives you've considered
Currently, the alternatives involve using Quickwit for traditional keyword-based search while relying on external vector search engines like FAISS, ChromaDB, or Milvus for embedding-based search. However, having built-in support for vector search within Quickwit would provide a more streamlined and efficient solution, reducing system complexity and improving performance.

Additional context
Vector search is becoming increasingly important in modern AI applications, particularly for NLP tasks, recommendation systems, and image similarity searches. If Quickwit were to support vector search, it could become a compelling alternative to existing solutions by combining high-speed indexing with efficient similarity searches. Compatibility with LangChain's vector store APIs would further enhance its usability within the AI ecosystem.