BrowserBee is an open-source Chrome extension that allows you to control your browser using natural language commands.
It uses a Large Language Model (LLM) to understand your instructions and Playwright to carry out the browser automation tasks.
This means you can tell BrowserBee to do things like “navigate to Twitter, check my notifications, and summarize them,” and it will work through those steps.
Since most of BrowserBee’s operations happen directly within your browser (the LLM communication being the main exception), it can interact with sites where you’re logged in—think social media or email—without your sensitive data going through some third-party backend.
See It In Action
Features:
- LLM Support: It works with major LLM providers like Anthropic, OpenAI, Gemini, and even local instances via Ollama.
- Cost Tracking: It keeps an eye on token usage and the associated costs, so you’re aware of how much you’re spending for each task. This, I find, is a really practical feature.
- Browser Interaction Tools: It has a comprehensive suite of tools for navigation (going to URLs, back, forward), tab management (opening, closing, switching), and page interactions (clicking elements, typing text, handling dialogs).
- Observation Tools: BrowserBee can fetch page titles, snapshot the DOM, query specific elements, read visible text, and take screenshots.
- Mouse & Keyboard Tools: It can perform mouse movements, clicks, drags, and press keys or type text directly.
- Playwright Powered: Under the hood, it uses Playwright, which is known for robust browser automation.
- Memory Function: BrowserBee can remember sequences of actions for specific tasks on websites. These “memories” are stored locally and can make future, similar tasks much more efficient.
- Approval Workflow: For actions like making purchases or posting social media updates, the agent will ask for your explicit approval first.
Use Cases
- Social Media Butler: Have it check your social media accounts, summarize new notifications and messages, and even help draft responses.
- News Curator: Ask it to gather and summarize the latest headlines from your favorite news sites or blogs for a quick, personalized update.
- Personal Assistant: Use it for everyday online chores like reading and sending emails, booking flights, or finding products. I’ve found it surprisingly helpful for sifting through product pages.
- Research Assistant: It can aid in gathering and organizing information on companies, job listings, market trends, or academic papers.
- Knowledge Bookmarking & Summarization: Quickly get summaries of articles, extract key information, and save useful bits for later.
How To Use BrowserBee
1. Install the BrowserBee extension from GitHub.
- Download the latest release from GitHub
- Unzip the downloaded file
- Open Chrome and navigate to
chrome://extensions/ - Enable “Developer mode” (toggle in top-right)
- Click “Load unpacked” and select the unzipped directory
- Set up your LLM API key(s) in the options page
2. Basic Usage:
- Click the BrowserBee icon in your Chrome toolbar (or press Alt+Shift+B)
- Type your instruction in natural language (e.g., “Go to Twitter, check my notifications, and summarize them”)
- Hit Enter and watch BrowserBee complete the task
3. Tips for Best Results
- Leave BrowserBee attached to a base tab throughout your session
- If you close the attached tab, use the reattach button to connect to a new tab
- BrowserBee can’t attach to new tabs without URLs or tabs with URLs starting with ‘chrome://’ or ‘chrome-extension://’
Pros
- Privacy-focused – Runs locally in your browser, minimizing data exposure
- Open source – Transparent codebase you can inspect and modify
- Multiple LLM options – Freedom to choose your preferred AI provider
- Memory feature – Gets more efficient over time by remembering successful task patterns
- No backend infrastructure – Works with your logged-in sessions safely
- Extensive tool suite – Rich set of browser interaction capabilities
Cons
- Setup complexity – Requires manual installation and API key configuration
- Chrome-only – Currently limited to Chrome browser
- API costs – Requires your own LLM API keys with associated usage costs
- Learning curve – May take time to understand optimal instruction phrasing
- Limited to web tasks – Cannot perform actions outside the browser environment
FAQs
Q: How does the memory feature work?
A: When BrowserBee completes a task you’ve assigned, you can instruct it to save_memory. It then stores the sequence of tool uses (like “navigate here,” “click this button,” “read that text”) locally in your browser’s IndexedDB, associated with that specific website. The next time you’re on that site, you can ask it to lookup_memories, and it can try to apply a saved sequence, often much faster and with fewer LLM calls.
Q: Is BrowserBee secure to use with my logged-in accounts?
A: BrowserBee is designed with privacy as a priority. It runs primarily within your browser. The main external communication is with the LLM provider you configure. Because it operates locally, it can interact with your logged-in sessions without needing to send your credentials elsewhere, which is generally more secure for personal use than solutions requiring backend access to your accounts. Always ensure you trust the LLM provider you are using.
Q: What happens if I close the tab BrowserBee is attached to?
A: If the tab BrowserBee is actively controlling is closed, its connection to that page will be lost. However, the BrowserBee extension panel has a “reattach” button. You can simply open a new tab (or navigate to the one you want to work with) and use this button to re-establish BrowserBee’s control.
Q: Can BrowserBee work across multiple tabs?
A: Yes, BrowserBee has tab management tools that allow it to open new tabs, switch between existing tabs, and close tabs. This enables it to perform tasks that might span multiple web pages or require information from different sources.
