Skip to main content

​
Overview

The BrowserUse Block is a critical block and a major contributor to Spine being state of the art on benchmarks. It allows agents to spin up actual browser sessions and interact with the real web: clicking, navigating, scrolling, filling forms, and extracting content. Multiple BrowserUse blocks can run in parallel, which means agents can browse several sites simultaneously. These blocks can also access web archives, which is essential for research tasks where current web content has changed since the source was published. BrowserUse is what makes Spine a genuine agentic platform rather than just an LLM wrapper with search. The agent is actually navigating and interacting with web pages the way a human would, not just calling a search API.

​
Key Features

  • Real browser sessions — Agents click, navigate, scroll, fill forms, and extract content from live websites
  • Parallel browsing — Multiple BrowserUse blocks run simultaneously, gathering information from many sources at once
  • Web archive access — Access archived versions of web pages for research where current content has changed
  • Context integration — Extracted content flows into connected downstream blocks automatically
  • Dynamic page interaction — Interact with JavaScript-heavy pages, SPAs, and authenticated flows

​
How Agents Use It

BrowserUse blocks are typically created and managed by L2.5 persona agents as part of larger workflows. A persona agent might:
  1. Do a web search to find relevant URLs
  2. Spin up a BrowserUse block for each URL
  3. Run them in parallel to gather information simultaneously
  4. Read the results and use them as input for the next step in the workflow

​
How to Use

  1. Add a BrowserUse Block — Click on the canvas → Select “BrowserUse Block”
  2. Describe the goal — Natural language instructions for what to do on the web (e.g. “Pull pricing from these three competitor sites”)
  3. Connect context — Link blocks that provide URLs or parameters
  4. Run — The agent spins up a browser session and executes the task
  5. Use results downstream — Connect to Prompt, Memo, Table, or artifact blocks to process what was gathered

​
Example Use Cases

  • Competitive intelligence — Browse competitor websites, pricing pages, and product changelogs in parallel
  • Data gathering — Navigate data portals, government sites, and databases to extract specific information
  • Research verification — Cross-reference claims by visiting original sources
  • Market monitoring — Check multiple sites for pricing changes, new products, or news updates
  • Web archive research — Access historical versions of pages for research where content has changed