Choose the right API endpoint for your job.

Ask a question in natural language; get a source-backed answer and optional structured JSON.

Use when

• Grounding LLMs on live web data
• Enrichment when the input is a question, not a URL list

You get

An answer object with validated fields, citations, and NOT_FOUND when the web cannot support a claim — not just raw page text.

Tip: Use Search when you need a list of links to explore; use Answers when you need a single synthesized response.

Semantic web search from a plain-English query — deduplicated links with titles and descriptions.

Use when

• Discovery without maintaining your own SERP scrapers
• Feeding candidate URLs into Scrapes or Batches

You get

A search object with result.links (url, title, description) plus optional hosted JSON for the full result set.

Tip: Search returns links across the web for a query; Map returns URLs discovered for one site you already chose.

Extract content from a URL you already know — markdown, HTML, text, JSON, screenshots, parsers, LLM extract.

Use when

• One page or a small set of known URLs
• Rich formats and structured extraction per page

You get

Per-URL content (markdown_content, html_content, json_content, etc.) ready for RAG, agents, or storage.

Tip: Use Batch when you have hundreds to thousands of URLs and want one job with predictable wall-clock time.

Discover URLs for a single domain — sitemaps, links, filters, and cursor pagination.

Use when

• Site audits
• Building URL lists before crawl or batch jobs

You get

A list of URLs on or under the target site, filterable with include/exclude patterns.

Tip: Map discovers URLs only; Crawl fetches content across many pages of a site.

Walk a site from a start URL with depth and page limits; poll until completed; list pages and retrieve content.

Use when

• Multi-page documentation or blogs on one site
• When you want exploration + content in one workflow

You get

Crawl status, per-page retrieve_ids, then markdown/HTML/JSON via /v1/retrieve — similar to batch but site-shaped.

Tip: Use Batch when your URLs come from anywhere (not one crawl frontier); use Crawl for hierarchical site walks.

Process large URL lists in one job — predictable completion time, parsers, webhooks, metadata, cursor over items.

Use when

• Thousands of arbitrary URLs (catalogs, SERPs, exports)
• Pipelines that need webhooks and batch-level parsers

You get

Batch status, paginated items, then per-item retrieve — ideal for scale without hand-rolling concurrency.

Tip: Parallel Scrapes are often faster for tiny sets; Batch wins for large lists and operational simplicity.