SC
SilentCanoe Utilities /
💬 Chat with My Documents
☕ Buy me a coffee
100% Local — AI runs on your device

Free AI Document Chat — Ask Questions About PDFs Offline

🔒

The AI model runs entirely inside your browser using your device's GPU. Your documents and questions are never sent to any server. No internet connection needed after the model loads.

AI Model

Downloaded once, then cached on your device permanently.

Your Documents

📂

Drop files here or browse

PDF · TXT · DOCX · Markdown

Ask your documents anything

Load a model and upload documents to start

👋 Hello! I can answer questions about your documents. Here's how to get started:

  1. Choose a model size (Fast or Smart)
  2. Click Load Model — it downloads once and is saved on your device
  3. Upload your documents (PDF, DOCX, TXT)
  4. Ask me anything about them!

AI can make mistakes. Always verify important information.

Frequently Asked Questions

What file types can I chat with?

You can upload PDF, DOCX (Word), and plain text (TXT) files. The tool extracts the text content and sends it to the on-device AI as context for your questions.

How does the AI answer questions about my document?

The document text is extracted in your browser and passed as context to Llama 3.2 3B, running locally via WebGPU. The model reads your document and generates answers based on its content — similar to asking a ChatGPT-style assistant, but entirely offline.

Is my document kept private?

Yes — your documents are never uploaded. All text extraction and AI processing happens locally in your browser using WebAssembly and WebGPU. Nothing leaves your device.

Is there a document size limit?

Very long documents may exceed the AI model's context window. For best results, use documents up to ~50 pages. For longer documents, ask specific questions about sections rather than broad summaries.

What browsers are required?

AI chat requires Chrome 113+ or Edge 113+ on desktop (WebGPU). Document text extraction works in all modern browsers. Firefox and Safari users can still extract and read document text but won't get AI answers without WebGPU support.