One POST in, structured data out. Name the fields you want — the extraction is AI-driven, so it keeps working when the site's markup changes. Pay per URL. No subscription, no scraper to babysit.
No SDK required — it's one HTTP call. Up to 50 URLs per request, fetched concurrently, one response with per-URL status.
curl -X POST https://api.mentalpros.org/extract \ -H "Authorization: Bearer YOUR_API_KEY" \ -H "Content-Type: application/json" \ -d '{"urls": ["https://example.com/product/123"], "fields": ["title", "price", "rating"]}'
{
"results": [{
"url": "https://example.com/product/123",
"status": "success",
"data": { "title": "Acme Anvil", "price": "$49.99", "rating": 4.7 }
}],
"credits_charged": 1,
"credits_remaining": 499
}
Name the fields you want and get exactly those JSON keys back — or omit them and get the page's main structured content.
Extraction keys off meaning, not CSS selectors. When the target site ships a redesign, your integration keeps working.
Pass "format": "markdown" and get the page as clean markdown — ready to feed to an LLM or a knowledge base.
One credit per successful URL. Failed fetches are detected and refunded automatically, in the same request.
Up to 50 URLs per call, fetched concurrently with per-URL status — one round trip instead of fifty.
Every request fetches the page at call time with a well-behaved, self-identifying bot. Nothing served from a stale cache.
The live page is fetched with redirect re-validation, size caps, and private-network protection.
Markup, scripts, and chrome are stripped down to the meaningful text content.
An AI model fills your exact schema — or produces clean markdown — with null for anything genuinely absent.
Most tools in this space push you into monthly tiers. We don't: buy credits when you need them, they never expire, and failures never cost you.
5 free credits, key delivered to your inbox. No card, no forms beyond this one.
Any publicly reachable HTML page: product pages, articles, listings, docs, profiles. Pages behind logins or rendered entirely client-side may come back thin — those fetches fail cleanly and are refunded.
A scraper needs you to write and maintain selectors per site. Here you describe the output you want, and the extraction adapts to whatever markup the page has today. The trade-off is per-URL cost, which stays fair via auto-refunds.
Every request fetches the live page at call time. Nothing is served from a cache.
The model is constrained to the page's text and told to use null for missing values, but AI output can still contain errors — verify before relying on it for high-stakes use. Per-URL status makes failures explicit.
The API fetches only publicly accessible pages, identifies itself with a bot user-agent, and blocks private-network targets. You're responsible for having the right to extract what you request — see the terms.
Up to 50 URLs per request; your credit balance is the only other limit. Need sustained volume? Email us.