Every 8 minutes, an Indian family loses money to a phone scam.
We built Kavach after watching the same thing happen to people we know — a retired schoolteacher in Hyderabad who transferred ₹4 lakh to a man impersonating a CBI officer. A working professional in Mumbai who shared her OTP to someone claiming to be her bank. A small business owner in Delhi whose UPI collect request was disguised as a refund.
The common thread isn't gullibility. It is fear. Scammers are psychologically sophisticated. They manufacture emergencies, impersonate authority, and exploit the same human instincts that make us good people — the instinct to comply with law enforcement, to protect our family, to act quickly in a crisis.
Google's phone app will tell you "Spam likely." That is not enough when someone is threatening your elderly parent with digital arrest. Kavach is built to interrupt the psychology of the scam — not just flag the phone number.
Most spam apps match phone numbers against a database. Kavach does that too. But the features that matter most are the ones that work after the number gets through.
During the call: A persistent notification with one button — Alert my guardians. One tap. Your family knows immediately, silently, without the scammer hearing anything.
When WhatsApp opens: Scammers routinely move victims from a phone call to WhatsApp video to show fake police stations and arrest warrants. Kavach detects this exact pivot and shows a full-screen warning before the camera turns on.
After the call: 90% of financial loss happens in the 30 minutes after a scam call ends, when the victim is still in a panic state. Kavach shows a mandatory post-call message — "Government agencies NEVER arrest anyone over phone or video call. This is legally impossible." — before you can open a banking app.
Neighbourhood awareness: When 10 Kavach users report the same number within 15 minutes, everyone nearby gets an alert with the exact script the scammer is using. The scammer's illusion of a personalised emergency collapses.
This is the difference between a spam filter and a psychological firewall.
Kavach runs three AI models entirely on your device — no internet required for fraud detection:
For cases the on-device models find ambiguous, Kavach sends the text (never your identity) to Gemini 2.5 Flash for a second opinion.
The 7-stage call screening pipeline runs in under 1.5 seconds: blacklist lookup → STIR/SHAKEN verification → spoofing detection → IVR bot challenge → AI threat score → decision → overlay.
Kavach makes money from subscriptions, not data. We cannot record your calls — Android prohibits this for third-party apps. We never read your contact list. We never see your phone number in plaintext — only a one-way cryptographic hash that cannot be reversed.
When you report a scam, your report is anonymous. When your guardian receives an alert, they receive a notification — not a recording, not a transcript, not a location.
We built the privacy architecture first and the features second. That is the only order that makes sense.
Kavach is built in Hyderabad by a team that grew up watching digital arrest scams, KYC phishing, and UPI fraud become a daily reality for Indian families.
The AI is trained on Indian fraud patterns — the specific vocabulary of CBI impersonation, the Hindi phrases used in OTP phishing, the Hinglish scripts of fake investment groups. It understands the context that a global spam database cannot.
We are registered in India and process data in compliance with the Digital Personal Data Protection Act, 2023.
Kavach launched with protection for individuals and families. The same technology — velocity burst detection, cross-platform pivot alerts, psychological circuit breakers — scales to protect entire organisations, banks, and telecom networks.
If you are interested in enterprise or partner integrations, write to us at support@kavachprotect.com.