Runi - AI Browser Agent: Autofill Forms, Automate Pages, Summarize

Runi is a controllable AI agent in your browser side panel. It helps you understand and work with the page you are currently viewing without switching context. Your page, your way. Use Runi to: • Summarize the current page and ask page-grounded questions. Attach text files, images, and PDFs to a request. PDF text is extracted locally before analysis. • Inspect page evidence such as readable text, HTML, DOM structure, scripts, stylesheets, and computed styles for technical analysis. • Request page transformations and browser-page actions. Known actions run automatically; detected form submissions ask for approval every time. • Bring context across multiple tabs in the same window. Type `@` in the composer to pick other tabs and send their content together to the AI. Referenced tabs are read-only — the model can read and analyze them, but write operations are refused. • Start common tasks from a shortcut chip: Summarize page, Translate selection, Fill this form, Polish selection, and Focus mode are built in, and you can add your own. • Select text on a page and click the Ask Runi bubble to ask about exactly that selection. • Let the model look at a screenshot when the answer depends on what the page looks like — canvas-rendered content, iframe content, whether a button appears disabled, layout problems. Enable it by marking the model as image-capable in Settings. • Take over at any time while Runi is working: if it detects your own mouse or keyboard input, it pauses and asks whether to continue or stop there. Redact page content before it is sent: phone number, email, ID number, and bank card rules are built in, and you can add your own. A match is replaced with a placeholder in full, so no original characters reach the model. Screenshots are images and are not covered by these text rules. • Connect your own OpenAI-compatible or Anthropic-compatible AI provider, including a custom endpoint. Before first use, configure your own AI provider and API key. Using DeepSeek as an example: 1. Sign in to the DeepSeek Platform, then create and copy an API key. 2. Open the Runi side panel and use the Settings link in the setup banner, or open Settings from the top-right menu. 3. Under Model providers, select Add provider and choose DeepSeek from Quick preset. 4. The preset fills in Base URL https://api.deepseek.com and model deepseek-v4-pro; leave Protocol at its default, OpenAI-compatible. Paste the key into API Key and select Add. 5. Return to the side panel and confirm that the model chip below the composer reads deepseek-v4-pro. You can now start a conversation. DeepSeek may charge for API usage. Runi is local-first: provider settings, API keys, interface preferences, and conversation history are stored in your browser. Text and image attachment contents may be stored with local conversation history; PDF text is extracted locally for the current request and is not persisted as PDF content. Runi has no developer-operated backend, analytics, or advertising SDK. When you initiate an Agent request, you direct Runi to transmit your API key, current prompt, recent conversation context, relevant current-page tool results, and the contents of files you selected for that request directly from the extension to the AI provider endpoint you configured as needed to fulfill that request. That provider processes the request under its own terms and privacy policy. Runi reads or changes a page only after you open the product and initiate an action. Known page actions run automatically; detected form submissions ask for approval every time.