Local-First Knowledge Engine

Yourbrowsinghistory,searchablebymeaning.

Pages, videos, purchases, likes — Rekoll captures it all and makes it instantly findable with natural language. On-device ML means your data never leaves your machine.

0
Entities Extracted
0
Connections Made
0
Cloud Uploads
0ms
Recall Time

Core Architecture

Total Recall Vector Store

384-dimensional embeddings generated by all-MiniLM-L6-v2 stored in Dexie.js over IndexedDB. Sub-50ms cosine similarity search across your entire browsing history.

384-dim vectors all-MiniLM-L6-v2 IndexedDB Dexie.js

Zero Egress

Transformers.js runs ONNX models in WASM — zero network requests for inference. Your data never leaves the browser.

Transformers.js ONNX WASM

Smart Extraction

Site-specific extractors for YouTube, Amazon, Reddit, and X. Five event types capture rich context from every interaction.

5 event types 4 extractors

Cross-Platform Knowledge

Automatic interaction tracking across sites. Pages, videos, purchases, likes, and bookmarks unified into a single searchable knowledge base.

YouTube Amazon Reddit X Generic

See it in action

Watch the three-step workflow: capture your browsing, embed it semantically, then search by meaning.

Rekoll — Capturing Events...
Capture
Embed
Search
Search by meaning...
All Page VisitVideoPurchaseLike
Waiting for events...
Rekoll
Privacy Terms GitHub © 2026