Description
Expand is an applied research lab that builds tools enabling AI agents to access and understand the web in a human-like manner. Traditional web scraping tools often rely on parsing raw HTML, which can miss crucial visual context, layout, and hidden data that are only apparent in a rendered browser page. Expand addresses this by opening any given URL in a real browser, allowing its Iris model to analyze the fully rendered page.
Read More
The core product, Fetch, takes a URL and returns the page content as structured Markdown, preserving tables, icons, and overall layout. Crucially, it also extracts hidden JSON data, referred to as State JSON, which often contains the underlying data powering dynamic web applications. Every line of the Markdown output is linked back to its exact source on the original page, providing agents with verifiable context for their readings.
Expand offers various building blocks beyond Fetch. Iris and X-Ray work on browser pages, while the Wayback Machine creates replayable Snapshots of fetched pages. Highlights identifies relevant snippets on a page to answer specific questions, reducing the need for agents to process entire documents. Search helps agents find pages worth fetching, and Crawl enables agents to process entire websites. WebSync continuously monitors sites for changes, keeping agents updated.
The platform is designed for developers and agents, offering integration via a CLI, TypeScript SDK, and a direct API. It automatically handles common bot checks and CAPTCHAs like Cloudflare, Turnstile, and reCAPTCHA for public pages, though it does not bypass sign-ins or paywalls. Pricing is credit-based, with 1,000 credits costing $1. A standard Fetch operation costs 1 credit, with additional costs for features like Highlights, converted documents, vision models, video, and transcribed audio. The system only charges for delivered content, not for failed attempts.
Expand aims to empower AI agents with a more complete and accurate understanding of the web, moving beyond simple HTML parsing to interpret the visual and data-rich context of rendered pages. This capability is vital for agents to perform complex tasks and interact with online information as effectively as humans.
- CategoryAI Agent Builders
- Pricing ModelCredit-based
- DeploymentCloud
- PlatformsWeb
Expand's Core Features
Real Browser Rendering: Opens URLs in a real browser to process rendered pages, not just HTML.
Structured Markdown Output: Returns web pages as Markdown, preserving tables, icons, and layout.
Hidden JSON Extraction (State JSON): Extracts underlying data from dynamic web pages.
Source Linking: Every output line links back to its exact spot on the original page.
API Access: Provides a robust API for agents to interact with web content.
SDKs & CLI: Offers TypeScript SDK and Command Line Interface for integration.
Bot & CAPTCHA Bypass: Automatically handles common bot checks and CAPTCHAs for public pages.
Page Snapshots: Creates replayable Snapshots of fetched browser pages.
Content Highlighting: Identifies and extracts relevant snippets that answer specific questions.
Website Crawling: Enables agents to crawl entire websites for comprehensive data collection.
Web Change Monitoring (WebSync): Continuously monitors websites for updates and changes.
Credit-Based Billing: Charges based on prepaid credits for delivered output, not tokens.
Progressive Disclosure: Returns main content first, with pointers to held-back data for efficiency.
Citation Links: Provides citation URLs for highlighted snippets for verification.
Agent Setup Guides: Offers /start.md for easy agent configuration.
How to use Expand?
Connect Agent: Use /start.md, CLI, TypeScript SDK, or API to integrate Expand with your agent.
Send URL: Submit a page URL to Expand's Fetch API.
Receive Structured Output: Get the rendered page back as Markdown, State JSON, and a Snapshot.
Process Content: Utilize the structured output for agent tasks, leveraging source links for context.
Extract Specifics: Use Highlights to find snippets answering questions or JSON Mode for separate fields.
Monitor Websites: Implement Crawl for full site processing or WebSync for continuous change detection.
Expand's Use Cases
- AI Agent Web Access
- Structured Data Extraction
- Automated Web Scraping
- Content Summarization
- Website Change Monitoring
- Historical Web Content Analysis
- Developer Tool Integration
- Enhanced Agent Context
- Research and Data Collection
- Competitive Intelligence





