EdgeAI — Private AI Assistant
Overview
Your AI. Your device. Your data. Works offline once the models download. No API keys, no accounts, no data leaves your device.
EdgeAI is an AI assistant that runs inside Chrome on your own computer. Its language model, Phi-4-mini, runs on your GPU, so your chats, notes and PDFs are never sent to a server. Ask it general questions, or import your own notes, PDFs, bookmarks and web pages and ask about those. On first run, EdgeAI downloads its models from Hugging Face, about 2.4 GB in all. After that it works offline. It's free, and you don't need an account or an API key. WHAT YOU CAN DO - Chat with Phi-4-mini (3.8B parameters). On GPUs that look too small for it, EdgeAI switches to Llama 3.2 1B. - Import an Obsidian vault or PDFs and ask questions about them. Chrome bookmarks can be imported too; EdgeAI keeps each one's title and address and doesn't visit the pages. - Add the page you're reading with the "Index this tab" button, or right-click any page and choose "Index this page with EdgeAI". - Talk instead of typing. Moonshine transcribes your speech on your computer. - Have answers read aloud. EdgeAI only uses voices installed on your computer and skips Chrome's "Google" voices, which run on Google's servers. - Keep EdgeAI open in Chrome's side panel next to the page, or open it with Ctrl+Shift+E (Cmd+Shift+E on Mac). HOW IT ANSWERS FROM YOUR FILES Your documents are split into passages and indexed on your computer with bge-small-en-v1.5 (130 MB, running on your GPU). Each question runs a keyword search and a vector search, a reranker scores what comes back, and passages unrelated to the question are dropped before the model answers. CHECK THE PRIVACY CLAIMS YOURSELF Open the Privacy tab. It lists every network request EdgeAI's pages have made, and one row counts "Requests other than model downloads". It should say None. The same tab shows what EdgeAI stores and how much space it uses, a log of your questions and imports that you can export as JSON, and the version and git commit of the build you're running. The privacy policy explains how to compare the network log with Chrome's own DevTools. EdgeAI has no access to the sites you visit. It reads a page only when you ask it to index that page. REQUIREMENTS - Chrome 116 or later with WebGPU, which is on by default in Chrome on Mac, Windows and ChromeOS - About 5 GB of free disk space for the model download OPEN SOURCE MIT license, code at github.com/kamalkalwa/EdgeAI. No telemetry, analytics or tracking
0 out of 5No ratings
Details
- Version0.1.1
- UpdatedOctober 8, 2026
- Offered byVisgen AI
- Size17.19MiB
- LanguagesEnglish (United Kingdom)
- Developer
Email
kamal.kalwa.dev@gmail.com - Non-traderThis developer has not identified itself as a trader. For consumers in the European Union, please note that consumer rights do not apply to contracts between you and this developer.
Privacy
EdgeAI — Private AI Assistant has disclosed the following information regarding the collection and usage of your data. More detailed information can be found in the developer's privacy policy.
EdgeAI — Private AI Assistant handles the following:
This developer declares that your data is
- Not being sold to third parties, outside of the approved use cases
- Not being used or transferred for purposes that are unrelated to the item's core functionality
- Not being used or transferred to determine creditworthiness or for lending purposes
Support
For help with questions, suggestions, or problems, visit the developer's support site