WorldCup App
A football prediction app built entirely with AI, from spec to deployed product. Now used by 250+ people in production.
"What does AI-first development actually look like, end to end? I built this to find out."
Every office does it. Someone makes an Excel, shares it on WhatsApp, half the team forgets to fill it in, nobody knows the scores by day three, and the whole thing quietly dies. We'd done this too many times.
I wanted something that actually made the tournament worth following, not just a prediction sheet but a real shared experience. A proper competitive game with live results, a leaderboard everyone can see, each person's picks visible to the group. Friends outside the office wanted in too. So it grew.
An end-to-end web application that replaced the spreadsheet entirely.
- + Match prediction system for all participants
- + Real-time leaderboard with automated scoring
- + Live visibility of other users' predictions
- + Social layer with chat between participants
- + Admin panel for result management and updates
Design, code, testing, deployment: AI tools were the primary instrument throughout. From spec to shipped in under two weeks. Not a prototype, a working product used by a real team, and now backed by MaaliMetrics as a launch partner.
- · AI compresses timelines drastically on boilerplate, but the hard parts stay hard
- · Prompt quality matters more than model choice when the logic gets complex
- · Debugging AI-generated code requires genuine understanding, not just running it
- · You ship faster, but you also ship bugs faster. Testing discipline is non-negotiable
- · A simple need turned into a product creates engagement a spreadsheet never could