How a check works
PehleCheck combines fixed security checks with careful, limited use of AI. Here is exactly what happens.
- 1
You share the suspicious thing
Paste a message or link, enter a phone number, or upload a screenshot, PDF or voice note. On WhatsApp, forward it to the PehleCheck number.
- 2
We pull out what matters
Links, phone numbers, UPI IDs, email addresses and amounts are extracted. Screenshots are read on our server; QR codes are decoded; voice notes are transcribed.
- 3
Fixed checks run first
Known scam wording, link structure, domain age, internal verified listings and external threat-intelligence lists. These give the same result every time for the same input.
- 4
AI reads the wording, if you allow it
Sensitive numbers are removed. The model must quote the exact words it relies on; anything it can’t back up is thrown away.
- 5
We score only what we observed
Each category of evidence has a cap, so no single weak signal can make something look critical. When the evidence is thin, the answer is ‘Inconclusive’.
- 6
You get the result and next steps
Keep it in your history, download it, share a safe summary, ask family to verify, or open an incident case if you’ve already paid.
What the score means
The score describes how much risk we observed, from 0 to 100. Below 25 is Low, 25–54 Suspicious, 55–79 High and 80 or more Critical.
A low score means the checks that ran didn’t find significant warning signs. It is never a guarantee that something is safe. New scam sites appear every day and may not be on any list yet.
When a provider is down
If a threat-intelligence or AI provider is unavailable, the result says so. We never fill the gap with a guess, and a missing check never counts as a clean one.