Tool · Verify a website
Check any website like a KYB reviewer would.
We crawl the homepage, extract the facts, and grade the site with a strict AI reviewer trained on the checks Stripe, Paddle and Adyen use during merchant onboarding.
What you get
Trust score (0-100)
A conservative score based on transparency, policies, contact info, and consistency of claims.
Missing pages check
Flags standard pages a reviewer expects: Terms, Privacy, Refund, Contact, Pricing, About.
Marketing claims review
Every bold claim on the homepage is graded supported, unverifiable, or exaggerated — with a note explaining why.
Contact + identity extraction
Pulls out email, phone, address and company number if the site publishes them.
How it works
- 1
Crawl
We fetch the homepage through Firecrawl and extract clean markdown plus the site's internal link graph.
- 2
Analyse
A strict compliance-style AI reviewer scores the site against the same checks payment gateways run during KYB.
- 3
Report
You get a plain-English trust score, strengths, risks, missing pages, and prioritised fixes.
Who uses this pattern
The crawl → extract → LLM-verify workflow is standard in fintech, compliance, and trust-and-safety. KYB providers such as Middesk, ComplyAdvantage and Onfido use it during business onboarding. Payment gateways including Stripe Radar, Adyen RevenueProtect and Braintree run it against merchant websites. Visa and Mastercard rely on G2 Web Services and LegitScript, which crawl millions of merchant URLs weekly and apply ML / LLM models to classify risk. StackswiseSoft packages the same technique — Firecrawl for the crawl layer, Lovable AI for the reasoning layer — into a tool anyone can run in seconds.
Limits & pricing
- Free — 3 verifications / day, homepage only. No account needed.
- Free account — 30 verifications / month, multi-page crawl, downloadable PDF report.
- Paid plans — bulk verifications, scheduled re-checks, and an API. See pricing.
- We never publish the results of your verifications and do not store the crawled page content beyond the request.