Local AI Browser Assistant

On-device AI in the side panel via Chrome Built-in AI (Gemini Nano).

A private AI assistant in Chrome's side panel — powered by Gemini Nano and running entirely on your device. Local AI Browser Assistant puts on-device AI next to the page you're reading. Chat, write, summarize, translate, proofread, and rewrite without sending your text to a cloud service. No account, no API key, no data collection. NEW IN VERSION 1.8 • Translate this page, in place. Right-click → "Translate this page" for block-by-block translation, with progress, Stop, and Show original. • Original text on hover. Hover any translated block to see what it said before. • Smart re-runs. On dynamic pages, only new or changed text goes through the model. • Pick another target language. The card offers alternatives when a pair is unavailable or the page already matches. • System, Light, or Dark theme. Applied to the side panel and in-page toolbars. FEATURES • Six modes: Prompt (chat), Summarize, Translate, Write, Rewrite, and Proofread. • Use the current page as context. Flip one toggle and ask questions about what you're reading, or summarize it on the spot. Available in Prompt, Summarize, and Write modes. • Quick actions on any selection. Select text on a page and choose Summarize, Translate, Proofread, Rewrite, New Prompt, or Existing Prompt from the floating toolbar. Actions are configurable in Settings. • Persistent conversations. Threads are saved locally, survive panel reloads, and can be reopened, renamed, or deleted. • Web search suggestions. When a question needs current information, the assistant offers a search chip. Only that search query leaves your device, and you can turn the feature off in Settings. • More than English. The interface is available in English, French, Simplified Chinese, and Traditional Chinese. Text translation supports 16 target languages, with the source language detected automatically. OPEN SOURCE The full source code is on GitHub: https://github.com/JLtheCoder/local-ai-browser-assistant SETTING UP GEMINI NANO IN CHROME The extension uses Chrome's built-in AI, so Chrome itself must have Gemini Nano ready: 1. Use desktop Chrome 138 or newer (Windows 10/11, macOS 13+, Linux, or a Chromebook Plus). 2. Keep at least 22 GB free on the volume that contains your Chrome profile, and have either a GPU with more than 4 GB of VRAM or 16 GB of RAM with 4 or more CPU cores. The model itself downloads a few GB, so an unmetered connection is recommended. 3. Check status at chrome://on-device-internals (Model Status tab). If the model isn't installed, open chrome://components and check for updates on "Optimization Guide On Device Model" — or just use an AI feature in the extension and Chrome will fetch it in the background. Keep Chrome open while it downloads. 4. For Write, Rewrite, and Proofread, enable the developer flags at chrome://flags/#writer-api-for-gemini-nano and chrome://flags/#proofreader-api, then relaunch Chrome. (Proofreader requires Chrome 142+. Prompt, Summarize, Translate, and page translation work out of the box.) 5. Not sure what's ready? The side panel shows a banner listing any unavailable API with a hint for the matching flag, plus a Download button for models that haven't been downloaded yet. PRIVACY Chat history and settings are stored locally inside Chrome. Page text and selected text are processed on-device and are only stored if they become part of a thread you save. The only thing that ever leaves your device is the web search you explicitly trigger — everything else runs on Gemini Nano on your computer. No analytics, no tracking, no ads, and no external server.