Use cases
memista is a small, embeddable vector search library. It shines when a single binary and two files on disk are the right shape — and it's honest about the ~100k-vector ceiling it's been tested to.
Embedded semantic search in a Rust app
Add similarity search to a Rust agent, CLI, or desktop app without running a separate database — memista compiles into your binary.
Builders of RAG pipelines that don't need a clusterA RAG chunk store
Store and retrieve document chunks for retrieval-augmented generation — text, metadata, and vectors in one small library.
Teams deduping content or flagging repeatsNear-duplicate detection
Find near-duplicate text by embedding it and querying for close neighbours — dedup a corpus or flag repeats with a small in-process index.
Product teams adding 'more like this' to a small catalogueSmall-scale recommendation
Serve 'more like this' recommendations from item embeddings with a single-binary index — no recommendation platform required.
Engineers prototyping retrieval before committing to a stackPrototyping vector search
Stand up a working vector search endpoint in minutes to prototype retrieval, then decide whether you need a heavier system.