I created DesiCheck because of a personal observation: digital adoption in India has outpaced digital literacy. I watched my own grandfather struggle to distinguish between a legitimate banking alert and a sophisticated phishing message on WhatsApp. He, like millions of other non-tech-savvy users, was relying on visual cues that scammers easily fake (like a bank logo).
I realized there was a massive gap in the security market. Large corporations have "Security Operation Centers" (SOCs) and firewalls, but the common man has no protection against social engineering. Existing tools are too technical—they talk about "SSL certificates" or "DNS records," which mean nothing to a senior citizen or a student. I wanted to build a "Tier-0 Security Analyst" that fills this gap: a tool that doesn't just block threats but explains why something is dangerous in simple language, effectively acting as a knowledgeable grandchild who is always available to help
In practice, DesiCheck is a lightweight Progressive Web App (PWA) designed for extreme simplicity. It works like a digital second opinion.
When a user receives a suspicious message (e.g., "Your electricity will be cut tonight" or a "Job Offer"), they simply take a screenshot and upload it to DesiCheck. Behind the scenes, we use Gemini 1.5 Flash (Multimodal AI) to analyze both the text urgency and visual anomalies (like blurry logos or mismatched fonts). Within seconds, the user gets a clear, color-coded verdict: Green for Safe, Red for Scam.
Crucially, the app provides a "Risk Score" and a layman's explanation (e.g., "This message creates false urgency and the link does not match the official bank website"). If it is a scam, the app also offers a "Troll Reply"—a pre-generated, safe, and sarcastic message the user can send back to the scammer to waste their time without revealing personal data.
Currently, DesiCheck is spreading through organic "trust networks"—primarily family WhatsApp groups. Because the problem is universal across generations, younger users are setting it up for their parents and grandparents.
The "Viral Counter-Measure" (Troll Agent) feature is designed to be our primary growth engine. When users generate a funny, sarcastic reply to a scammer using our AI, they often share that interaction on social media. This turns a negative experience (being targeted by a scam) into a positive, shareable moment, naturally increasing awareness of the tool. We are also planning to partner with local community centers and digital literacy drives to reach users who are most at risk but least likely to search for security tools app stores.
Initially, DesiCheck was just a text analyzer where users had to copy-paste message text. We quickly realized this was a friction point—elderly users struggle with copy-pasting on touchscreens, and many scams rely on visual imagery (fake posters/logos) that text analysis missed.
We modified the core engine to be Multimodal. We integrated Gemini 1.5 Flash to process images directly. We also realized that latency was an issue on Indian mobile data networks, so we added a client-side image compression layer to ensure screenshots upload instantly even on 3G/4G connections. We also pivoted the language model to better understand "Hinglish" (a mix of Hindi and English), as most scams in India utilize this hybrid language to build rapport.
You don't need to install anything from an app store. Since DesiCheck is a Progressive Web App (PWA), you can simply visit our secure URL on any mobile browser.
Once there, you will see two simple options: "Check Screenshot" or "Check Link." You can test it immediately by uploading any screenshot of a message you've received recently, or even a picture of a legitimate email to see how the AI verifies its authenticity. The app is free to use and requires no login for basic checks, ensuring privacy by design.
