Over 860 Participants, 63,000 Predictions and One Winner: How We Built a World Cup Prediction Platform with AI
For every FIFA World Cup or UEFA European Championship over the past 15 years, there has been a good chance that Hallstein Brøtan, Senior Developer at Novanet, has been behind the office prediction competition.
What began as a simple Excel-based competition for colleagues and friends has grown with every tournament. The dream was always to move beyond spreadsheets, and with the help of AI, that finally became possible.
From Excel to a Digital Platform
During Euro 2024, the setup was still relatively simple. Participants submitted their predictions in Excel, results were updated manually, spreadsheets were shared and maintained, and much of the work happened behind the scenes.
Then came the 2026 FIFA World Cup. For the first time in many years, Norway was part of the tournament.
That was when Hallstein, together with fellow Novanet colleagues Jan Ødegård and Hanna Kjus, decided it was time to do more than update another spreadsheet. The goal was to build a platform that could host the entire competition in one place and be accessible to anyone.
The result was an AI-powered prediction platform where users can not only predict match outcomes but also create private leagues for friends, family, and colleagues.
What was once an internal initiative has evolved into a complete digital product with hundreds of users and tens of thousands of registered predictions.
Why We Made the Change
The objective was simple: spend less time on administration and more time creating a great user experience.
Two years ago, we managed:
Prediction submissions via Excel
Manual result updates
Limited real-time visibility
This year, we built:
A dedicated web application
Automatic updates of results and standings
The difference goes far beyond design. The entire process is now automated. Predictions are submitted directly through the platform, results are updated continuously, and leaderboards are generated automatically.
Participants can follow the competition in real time, while the administration requires minimal effort.
How We Used AI Throughout Development
The solution was developed by Hallstein Brøtan and Jan Ødegård, both Senior Developers at Novanet, with AI playing a central role throughout the development process from concept to launch.
Using tools such as Lovable and GitHub Copilot, they were able to move from idea to functioning product in record time. AI was actively used throughout development, including for:
Generating code and components
Sparring around structure and architecture
Testing, iteration, and continuous improvement
The result is a platform that automatically handles everything from user registration and predictions to results and leaderboards.
Making the Competition More Social
One of the key goals was to make the competition more personal and social.
Users can create their own leagues and invite anyone they want to join. Participants have used the platform to:
Run internal competitions at work
Challenge friends and family
Compete in private leagues with custom rankings
This is where much of the engagement emerged. When league standings determine who has to buy the next coffee or endure some friendly teasing at lunch, every match suddenly becomes a little more important.
Technology That Creates Engagement
Although the solution is built on technology, the outcome is ultimately about people.
Throughout the tournament, the platform sparked activity and conversations across teams, companies, and groups of friends. Matches became topics of discussion during lunch breaks, in Teams chats, and around coffee machines, while participants closely followed their positions on the leaderboard after every game.
For us, this is a great example of what happens when technology does more than streamline a process. It creates engagement, strengthens communities, and contributes to meaningful experiences.
More Than 750 Users. 63,000 Predictions. One Winner.
What started as a simple Excel spreadsheet became a digital gathering place throughout the World Cup.
By the end of the tournament, the numbers told the story:
More than 750 registered users
Nearly 63,000 submitted predictions
Over 100 leagues created
An average of eight members per league
The NOVA League attracted 120 participants
And after weeks of excitement, surprises, and some remarkably bold predictions, we were finally able to congratulate Janne Britt Hauge as the winner of the NOVA League.
For us, however, this project is about more than who made the most accurate predictions. It demonstrates what is possible when curiosity, technology, and AI come together with a desire to build something people genuinely want to use.
What started as an idea inspired by a football championship became a platform with hundreds of users, built in record time with significant support from AI-powered tools.
And who knows? The next tournament might be even bigger.
