When the agent scrapes websites
You share a URL as reference
When you provide a link and want the agent to understand what’s there, it visits the URL to see the content and structure. Example: “Build a pricing page like this one: [example.com/pricing]” The agent visits the URL, reads the content, and understands how the pricing page is organized.Understanding existing designs
When you want to recreate or take inspiration from a website, the agent examines how it’s structured. Example: “Make my homepage look like Apple’s product pages” The agent visits Apple’s site to see how sections are organized, what content appears where, and how information flows.Learning from examples
When building something similar to an existing website, the agent scrapes the site to understand the approach, then adapts it to fit your project. Example: “Create a portfolio site similar to [designer-site.com]” The agent looks at the reference site to understand the layout and content structure, then builds your portfolio with a similar approach but customized to your needs.What the agent learns
When the agent scrapes a website, it reads: Text contentAll the written content on the page—headlines, paragraphs, button labels, navigation items. Structure and organization
How the page is organized into sections, what order content appears in, and how information is grouped. Key components
What types of elements are used—forms, galleries, cards, testimonials, feature lists. Images
The agent notes what images are used and where, though it sees them as part of the overall structure.
What the agent does with this information
The agent uses scraped content as reference material, not as something to copy directly. It understands the structure and approach, then makes independent decisions about how to implement something similar for your project. The agent adapts what it learns to fit your app, your content, and your goals. The agent doesn’t:- Copy code directly
- Clone the exact design
- Replicate everything it sees
- Understand the approach
- Adapt useful patterns
- Make decisions based on your needs
What the agent doesn’t scrape
The agent only scrapes external websites you reference. It doesn’t scrape your own app, it already has direct access to your code and can read it without scraping.Limitations
Login-protected contentThe agent can’t access pages that require you to log in or sign up. Very slow sites
If a site takes too long to load, the scraping may fail. The agent will let you know if this happens. Dynamic content
The agent waits for pages to load, including content that appears through JavaScript, but some highly dynamic sites may not fully load in time.
How this helps you
Web scraping makes it easier to communicate what you want. Show, don’t tellInstead of describing a complex design or structure in words, you can share a link. The agent sees exactly what you’re referring to. Get inspired by real sites
When you see something you like on the web, you can use it as a starting point. The agent understands what makes it work and adapts it for you. Save time on explanations
Rather than breaking down every detail of how a page is structured, you provide a reference and the agent figures out the rest. You get results that match your vision because the agent can see examples of what you want, not just hear descriptions of it.