Two kinds of user arrive here. Developers building retrieval into an AI application need somewhere to store embeddings and filter them quickly. Analysts and managers want to ask a database a question in plain language and get rows back without writing SQL. The category serves both, along with notebook and spreadsheet tools that generate the Python or SQL for you.

Read more about this category

Storing embeddings

Milvus and Chroma are vector databases for similarity search, with Milvus adding GPU acceleration. Pinecone is a managed vector database, while Weaviate is open source and supports hybrid search. Choose by how you will host the store and how much metadata filtering you need.

Asking in plain language

AI2SQL turns natural-language questions into SQL for MySQL, PostgreSQL, SQL Server, Oracle and others, and Bagel AI accepts questions in English or Arabic. Hex takes a notebook route: you write SQL, analyse with Python and share the result with your team.

  • Read the generated query before trusting the answer.
  • Connect with a read-only account, never an admin one.
  • Test Arabic column names and Arabic text values, not just English samples.

For analysis once the data is out, see Data analysis.