Semantic Search

Local Semantic Search

Paste documents, build a local index, and search by meaning using TF-vector similarity. Everything runs in your browser — no backend, no API key, no upload.

This MVP uses a lightweight local TF-vector fallback. Architecture is provider-abstracted, ready to swap in a WebGPU embedding model when browser support matures.

Document Index

Paste documents separated by blank lines. Each block becomes a searchable unit indexed locally using TF-vector similarity.

Text stays in your browser.