Skip to main content

Extract Web Page Data Workflow

This workflow asks an AI node to read a URL, use the Web Page to Markdown Tool, and return the main points of the page content.

Extract Web Page Data Workflow


Steps

  1. Provide the URL request: The workflow begins with a Data Source node containing a natural-language request about a specific web page.
  2. Call the page-reading tool: The AI Data Processing node decides that it needs the Web Page to Markdown Tool and passes the URL to it.
  3. Fetch and normalize the article: The tool opens the page in Chromium, follows redirects, handles common consent walls, extracts the main content, and converts it to Markdown.
  4. Summarize the result: The AI node reads the returned Markdown and produces a concise answer describing what the page says.
  5. Inspect the output: The Data Flow Spy node shows the final response for debugging or reuse.

Common Use Cases

  1. Summarizing a single article found through a search workflow.
  2. Reading redirect-heavy links such as Google News read/ URLs.
  3. Turning long-form web content into structured text before feeding it into later workflow steps.

Best Practices

  • Prefer direct article URLs when available, even though the tool can handle many redirects.
  • Keep maxToolRetries enabled so the AI has room to retry when a page initially fails to load.
  • Use Data Flow Spy while tuning prompts so you can inspect whether the AI is summarizing the page content or merely restating the input.