Crawl4AI Plus icon

Crawl4AI Plus

Crawl web pages, extract data, and ask questions using Crawl4AI

Actions4

Extract Data

AI-generated

Summary

Extract structured data from a specified URL by crawling one or multiple pages of the same domain using configurable crawling options and optional LLM-enhanced extractions such as organisation info, contacts, and custom data.

Inputs

  • URL (required) — The URL to extract data from.
  • About Organisation — If enabled, generates a concise description of what the organisation does and who it serves using an LLM.
  • About Organisation Prompt — Custom instructions for the LLM to generate the organisation description, applicable only if About Organisation is enabled.
  • Custom (LLM) — If enabled, runs a custom LLM extraction based on user-provided instructions.
  • Custom Field Name (required) — Key name used in the output JSON for the custom extracted value. Required when Custom (LLM) is enabled.
  • Custom Extraction Prompt (required) — Instructions for the LLM describing what to extract from the page. Required when Custom (LLM) is enabled.
  • Email Addresses — If enabled, extracts email addresses from the page, annotated with associated person or office names when LLM credentials are available.
  • Locations — If enabled, extracts all physical locations including addresses, phone numbers, and emails per location; also extracts global phone numbers and emails. Requires LLM credentials.
  • Official Org Name — If enabled, identifies the official registered or trading name of the organisation. Requires LLM credentials.
  • Phone Numbers — If enabled, extracts all phone numbers found on the page.
  • Crawl Scope — Defines how many pages to extract data from: single page, follow same-domain links (depth 1), or full site (depth 3).
  • Max Pages — Maximum number of pages to crawl. Applies only when Crawl Scope is set to follow links or full site.
  • Smart URL Selection — If enabled, uses LLM to select the most relevant pages for crawling by first extracting links and then selecting by relevance. Requires LLM credentials.
  • Options — Additional crawling options such as crawl mode (Standard, Anti-Bot for Cloudflare, or Auto), default country code for phone number parsing, exclude URL patterns to skip during multi-page crawling, and LLM model override.

Output shape

a single JSON object containing extracted structured data from the target URL and optionally linked pages, keyed by extraction type (e.g., organisation name, about organisation text, emails, locations, phone numbers, custom extracted data).

Output includes combined crawl results respecting max pages and crawl scope, with partial extraction for each enabled data type. LLM credential-required extractions (official org name, about organisation, locations, custom LLM extraction) depend on proper credential configuration. Email extraction includes context annotation if LLM is available. Smart URL selection narrows crawl targets when enabled. Multi-page extracts deduplicate pages and aggregate data accordingly.

Examples

Example 1: Extract just email addresses and phone numbers from a single web page

Set URL to the target page, enable 'Email Addresses' and 'Phone Numbers', leave Crawl Scope on 'Single Page'.

Example 2: Extract organisation info, multiple page locations, and custom product list across the full domain

Set URL to seed page, enable 'About Organisation', 'Locations', and 'Custom (LLM)', provide custom field name and prompt, set Crawl Scope to 'Full Site', adjust Max Pages and enable Smart URL Selection if focusing extraction to relevant pages.

Links

Discussion