Local-first
RAG with Chroma
Run retrieval-augmented generation entirely on your machine. Chroma embeds, stores, and queries vector data locally — no cloud, no API keys, no latency.
Embeddings
Convert text into dense vectors using sentence-transformers or OpenAI-compatible models running locally.
Storage
Chroma persists collections to disk. Restart your app and your vectors are still there.
Query
Cosine similarity search over thousands of documents in milliseconds. Filter by metadata.
Why local Chroma?
- ▸Zero network calls — your documents never leave your machine.
- ▸No vendor lock-in. Chroma is Apache 2.0 licensed.
- ▸Embedding model agnostic — swap models without changing your pipeline.
- ▸Perfect for offline-first desktop apps and privacy-sensitive workloads.