Nomic Layout v1 parses document structure, then OCR and embeddings power semantic search. Inference runs locally in your browser, powered by Muna.
Extracting pages and text layers with pdf.js, entirely in your browser.
Nomic Layout v1 detects headings, text blocks, tables, and images entirely in your browser.
Nomic Embed v1.5 builds a semantic vector index locally in the browser.
Finding the most relevant regions using vector similarity search.
SmolLM 2 135M creates a conversational summary based on the search results.
All processing runs locally via WebAssembly. Your documents never leave your device.