Evidence-First Website Facts Extractor

Extract compact, source-linked fact packs from public websites for AI agents, sales research, SEO audits, competitive research, RAG preparation, and due diligence workflows.

This Actor is intentionally evidence-first. It returns direct website snippets with source URLs and diagnostics. It does not use a private API, infer facts without evidence, enrich private individuals, or act as a broad unbounded AI web crawler.

Input

{
  "urls": ["https://apify.com"],
  "factPack": "features",
  "customTerms": [],
  "maxPagesPerSite": 8,
  "maxFactsPerSite": 20,
  "requestTimeoutSecs": 20
}

Supported factPack values:

• pricing
• faq
• features
• policies
• docs
• company
• custom

For custom, provide customTerms such as ["SOC 2", "API", "enterprise"].

Output

Each website produces one fact-pack record with:

• direct facts and evidence snippets;
• source URLs for every fact;
• the sanitized starting URL plus the site origin used as the crawl boundary;
• matched terms;
• pages scanned;
• confidence and completeness scores;
• missing fields and diagnostics;
• uncharged error records for blocked or failed sites.

Pricing

• $0.00005 when a run starts.
• $0.005 for each useful evidence-complete website fact pack.
• Failed, weak partial, duplicate, and empty records do not trigger useful-fact-pack-result.
• There is no apify-default-dataset-item charge, so writing an error record does not create a result fee.

Safety

• Public HTTP and HTTPS pages only.
• URLs with credentials, query parameters, or fragments are rejected.
• Clean path URLs such as /pricing, /docs, or /privacy are preserved as starting pages so users can target a specific fact page.
• Account, invite, reset, unsubscribe, or token-like paths are rejected and redacted to the site origin.
• Local and private-network addresses are blocked.
• Redirects must stay on the requested website.
• Each HTML page is limited to 3 MB.
• Crawling is bounded by maxPagesPerSite.