You Don't Need a Vector Database
Summary
The article argues that a vector database is often overkill for most teams and that semantic search can be achieved with a managed search API. It clarifies what a vector database does, when its flexibility is needed, and when it is better to use an API that handles embedding, indexing, and multilingual search. It also discusses vendor lock-in, build traps, and provides a decision framework for choosing between a vector database and a search API.