Hi, I'm Rayyan, founder of CodexAI, a small software company in Kurnool, India. I'm 18, and since 2025 I've shipped six products. This post covers the one I care about most, and the lessons I'd pass on to anyone building their first real product.

Where I started

When I was 13, I set a world record, recognised by the International Book of Records, by solving a customised Indian-flag 3x3 Rubik's Cube in 1:29. The lesson stuck: pick a hard problem, drill it until it's fast, then ship.

I work the same way now. Before building anything, I try to name the one problem it solves.

Safero: a personal safety app

Safero is a free personal safety app with a women's safety focus, built for anyone who might need help. It runs on Android and in the browser. The design problem is simple: in a bad moment, nobody wants to unlock a phone and hunt through menus.

So an alert can start in three ways:

  • Hold the SOS button.
  • Shake your phone. This starts a countdown you can cancel with one tap, so an accidental shake doesn't alarm anyone.
  • Tap the Safero tile in quick settings.

Your contacts then see your live location as it moves, along with your battery level and any medical details you chose to share.

Design decisions I'd defend

  • Contacts don't need the app. They get an email with a private link, so the person who might help you doesn't have to install anything first. App users also get a push notification.
  • Location is shared only when you raise an alarm, and alert links stop working after 24 hours.
  • The check-in timer runs on the server. If you don't end it in time, Safero alerts your contacts even if your phone is off or dead. A timer that lives only on the phone fails exactly when you need it.
  • Alarms survive bad connections. If you raise one offline, the app holds it, tells you it's waiting, and sends it when you're back online.
  • Shake detection stays on the device. Motion data is never uploaded.
  • A duress PIN. Entering it makes the screen show the alarm as ended, while your contacts keep being alerted.

There are also safe zones, a fake incoming call, audio recording, photo evidence, and "verified responders," who see your location only after saying they're coming.

Safero isn't an emergency service, and the app says so. It includes a one-tap 112 button for real emergencies.

What else I've shipped

  • Promptify: an AI toolkit with prompt generation for image and video models, AI retouch, and TruthLens, which scores whether an image or video looks AI-generated. You start free with 50 credits and no card.
  • Kurnool Shoppy, Hyderabad Shoppy and Hyderabad Directory: local commerce and discovery platforms for two Indian cities.

3 lessons from building six products

  1. Identify the problem first, keep the UI minimal. Safero's main action is one held button. Everything else is optional.
  2. Optimise for performance, architect to scale. The check-in timer on the server and the alarm queue for bad connections both come from asking what happens when things go wrong.
  3. Design for the worst moment. People use safety tools when they're scared and rushed. Every extra tap is a cost, and every failure case has to be planned for in advance.

What's next

Safero is Android and web today, with no iPhone app yet. That's the obvious gap.

If you want to follow along or talk about a project, find me on GitHub, LinkedIn or my site.