Anti-scam & citizen cyber emergency portal
Thagi Alert
India's independent anti-scam portal: AI screenshot checks, a fund-freeze assistant and bots, in 8 Indian languages.
My role: Founder and sole full-stack and AI engineer
- AI
- Next.js 16
- Payload CMS
- Gemini 2.5 Flash
- PWA

Overview
Thagi Alert gives citizens real-time intelligence on scams such as Digital Arrest, UPI QR fraud and APK trojans — and practical help in the first hours after being defrauded. I built it as founder and sole engineer.
The challenge
Cyber fraud in India moves fast, from Digital Arrest calls to UPI QR scams and fake electricity-bill APKs. Victims often do not know whether a message is a scam, and if money has gone, the first hours matter most: a quick call to 1930 and the bank improves the chance of freezing the funds.
Thagi Alert had to help people at both moments: checking a suspicious message in seconds, and guiding a victim through what to do right now, in their own language, even on a phone with no data left.
How it works
- 1Screenshot / number
- 2Gemini 2.5 Flash
- 3Risk score + red flags
- 4Recovery playbook
- 5Freeze + legal draft
- 6WhatsApp / Telegram
Architecture
The portal runs on the Next.js 16 App Router with Payload CMS 3.x and PostgreSQL. Alerts are published through the CMS with editorial workflows and citizen-report moderation, and over 50 localised routes are pre-rendered for fast loads. A dual-layer translation cache serves pages in eight languages: Hindi, English, Gujarati, Marathi, Bengali, Telugu, Tamil and Kannada.
The screenshot checker sends WhatsApp or SMS screenshots and links to Gemini 2.5 Flash, which returns a bilingual risk score, red flags, counter-measures and a recovery playbook. A public lookup API checks mobile numbers, UPI handles and domains against verified citizen reports. A 30-second wizard triages incidents into categories.
The Golden Hour assistant shows a 180-minute countdown, bank-specific freeze shortcuts for 20+ banks, and a one-click drafter for police complaints and RBI zero-liability bank disputes. A WhatsApp webhook on the Meta Cloud API verifies messages around the clock, and a Telegram bot is triggered by CMS lifecycle hooks so new alerts broadcast automatically. An offline-first PWA keeps 1930 dialling and freeze codes available without internet, and next/og generates branded share cards.
What I built
Multimodal threat check
Gemini 2.5 Flash reads suspicious WhatsApp/SMS screenshots and links and returns a bilingual risk score, red flags, counter-measures and a recovery playbook.
Golden Hour assistant
A 180-minute countdown, bank-specific freeze shortcuts for 20+ banks, and a one-click drafter for police complaints and RBI zero-liability bank disputes.
Fraud lookup engine
A public API that checks mobile numbers, UPI handles and domains against verified citizen reports and known scam patterns.
Omnichannel bots
A 24×7 WhatsApp verification webhook (Meta Cloud API) and a Telegram alert bot fired by CMS lifecycle hooks.
8 languages, offline
Hindi, English, Gujarati, Marathi, Bengali, Telugu, Tamil, Kannada — with an offline-first PWA so 1930 dialling and freeze codes work without data.
Fast and shareable
Static generation across 50+ localised routes and branded OpenGraph warning cards generated with next/og.
Engineering decisions
- Make the emergency path work offline, because victims often call from a phone with no data left.
- Publish alerts through the CMS and fan them out to Telegram with lifecycle hooks — no manual cross-posting.
Results
- Live at thagialert.in in 8 Indian languages.
- 50+ pre-rendered localised routes and an offline-first PWA.
- Freeze shortcuts for 20+ banks and one-click legal drafting.
- Automated alert broadcasting to Telegram and 24×7 WhatsApp verification.
Stack
- Frontend
- Next.js 16, React 19, TypeScript, Tailwind CSS v4, PWA
- CMS & data
- Payload CMS 3.x, PostgreSQL, REST + GraphQL
- AI
- Google Gemini 2.5 Flash (@google/genai)
- Integrations
- Meta WhatsApp Cloud API, Telegram Bot API, next/og
Need something like this?
I can build a version of this for your product, your data and your stack.
Start a project