# Extracto > Extracto is a web data extraction API. You send a URL and a JSON schema; it renders the page, extracts the fields you asked for with an LLM, validates them against your schema, and returns clean JSON. No CSS selectors, no XPath, no scraper maintenance. Built for AI agents and data teams. How it works: requests go through a real headless browser (so client-rendered pages are visible), the page text and metadata are passed to a language model with your schema as a strict contract, extraction runs deterministically, and the result is validated before it is returned. Fields that are genuinely absent come back as `null` rather than a guessed value. Extracto is most reliable on cooperative, server-rendered, structured pages (for example GitHub repositories, documentation sites, news articles, and pages with embedded structured data). It has a free plan (100 pages one-time, no credit card) and paid plans from $14/month (Hobby, 5,000 pages/month), with Standard at $69/month (30,000 pages) and Growth at $299/month (120,000 pages). Pages are billed in credits: a scrape is 1 credit, an AI extraction is 5 credits, and a map is 1 credit; anti-bot pages cost the same flat price, and failed requests and cache hits are free. ## Core pages - [Home and live demo](https://getextracto.dev/): What Extracto is, the API in one snippet, and a live demo that runs real extractions against a public URL. - [Docs](https://getextracto.dev/docs): API reference — endpoints, request and response shape, schema format, authentication. - [Pricing](https://getextracto.dev/#pricing): Free plan and paid tiers. ## Guides and blog - [Why we built a structured web-scraping API](https://getextracto.dev/blog/hello-world): The bet behind describing data instead of writing selectors, and what it took to make it production-grade. - [How to scrape a website without writing a single selector](https://getextracto.dev/blog/scrape-without-selectors): Why selectors break and how schema-based extraction stays stable across redesigns. - [Turning GitHub repositories into clean JSON](https://getextracto.dev/blog/scrape-github-to-json): A worked schema for pulling repo metadata into structured data. - [How we keep an LLM scraper from making things up](https://getextracto.dev/blog/stop-llm-scrapers-hallucinating): Schema enforcement at the boundary, null over guesses, and temperature-zero extraction. - [Pulling structured data out of documentation sites](https://getextracto.dev/blog/extracting-from-documentation-sites): Turning reference pages into a queryable dataset. - [JSON schema vs CSS selectors, compared](https://getextracto.dev/blog/json-schema-vs-css-selectors): An honest comparison of the two approaches and when each one wins. ## Company - [About](https://getextracto.dev/about): Who builds Extracto and why. - [Contact](https://getextracto.dev/contact): How to reach the team.