Skip to main content
Start from what you have and what you want back. Every endpoint uses the base URL https://api.olostep.com and the header Authorization: Bearer <YOUR_API_KEY> (see Authentication).

​
By goal

​
Common decisions

  • Scrape vs. Crawl: scrape when you know the exact URL. Crawl when you want many pages from one site and don’t want to list them yourself.
  • Map vs. Crawl: map returns URLs only and is fast. Use it to pick pages first, then scrape or batch just those.
  • Crawl vs. Batch: crawl discovers pages by following links. Batch takes a list of URLs you already have, up to thousands at a time.
  • Search vs. Answers: search returns links for you (or your agent) to read. Answers reads the pages for you and returns a single answer with citations.
  • Parsers vs. LLM extraction: use a parser when one exists for the site. It’s faster and returns the same fields every time. Use LLM extraction for anything else.

​
Async endpoints

Crawls and batches run in the background. Each returns an id right away. Poll GET /v1/crawls/{id} or GET /v1/batches/{id} until the status is completed, then fetch the results. Stop polling if status is failed. A crawl can still end as completed with 0 pages (check pages_count), and a batch can end as completed with some failed items (list them with ?status=failed). Each feature page shows the full flow.

​
Other ways to call Olostep