The Scrape API fetches a page (rendering JavaScript when needed) and returns structured JSON. With the extract_rules parameter you map field names to CSS selectors - for example {"title":"h1","price":".price"} - and get back exactly those fields. Rules support nested objects, arrays, per-field output types (text, HTML, attributes), and pagination across list pages. Because it is selector-based, results are deterministic and repeatable, so the same rules return the same JSON on every run.

Hand-written extract_rules give you full control and are ideal once you know a site's HTML structure - production pipelines that need 100% deterministic results, maximum speed, and the lowest cost per request. When you are scaling across many new site types, Guess-er can generate CSS rules for you to review and reuse instead of writing them from scratch. Either way you run the same Scrape API at the same low cost per request, and the selectors return identical results every time with no model variability.

Guess-er analyzes a page's HTML structure and generates CSS extract_rules that you can reuse with the Scrape API. Provide a target JSON schema or an example of the fields you want, and it returns selectors. The key advantage is cost efficiency: you generate the rules once, then reuse them with the cheaper Scrape API for all subsequent requests to similar pages. Use Guess-er when you are scaling extraction across many pages with similar structures, when you want the reliability of CSS selectors without writing them manually, or when you need to quickly bootstrap extraction templates for new site types.

Yes. Guess-er accepts a standard JSON Schema when generating rules - type definitions, required fields, enums, and nested structures - so the selectors it returns target the fields your schema expects. The schema format supports $ref and $defs for Pydantic-compatible schemas, anyOf/oneOf/allOf combinators, and all standard JSON Schema types including nested objects and arrays. And because Scrape API output is deterministic JSON, you can validate every response against the same schema (or your Pydantic model) on your side.

The Scrape and Extract APIs include full headless browser rendering with Chromium. JavaScript is executed, SPAs are rendered, and dynamic content is loaded before extraction begins. For bot-protected sites, our system includes stealth mode with realistic browser fingerprints, automatic proxy rotation across residential and datacenter pools, and automatic CAPTCHA solving. You can configure wait_for to pause extraction until a specific CSS selector appears, or set timeouts for pages with complex loading sequences. Set js=true to render before extraction.

Both endpoints return clean JSON by default with properly typed fields. The Scrape API with extract_rules returns data matching your defined rule hierarchy - nested objects, arrays, and per-field output types. The Extract API returns article fields with consistent naming. You can integrate results directly into databases, data warehouses, or downstream applications via our REST API. For batch processing, results can be delivered via webhooks or exported as CSV through our dashboard. All responses include metadata such as extraction time for monitoring and optimization.