Inspiration
The FIFA 2026 World Cup spans 16 cities across 3 countries. Millions of fans will travel internationally, navigate unfamiliar cities, and try to catch matches on time. I wanted to build something that actually helps fans take action — not just answer questions, but reason through a full travel plan in seconds.
What it does
WorldCup Pulse is an AI agent that helps FIFA 2026 fans plan their matchday experience. Ask it anything:
- "I'm attending a match in Dallas. What should I know before leaving?"
- "Show me all World Cup venues"
- "What's the Final match and where is it?"
The agent automatically calls multiple tools in sequence — match data, weather, travel alerts, and transport options — then combines everything into a clear, actionable recommendation.
How we built it
- Google Cloud Agent Development Kit (ADK) to build and orchestrate the agent
- Gemini 2.5 Flash as the reasoning engine
- MongoDB Atlas to store match schedules, venue data, and user preferences
- Google Cloud Run to deploy the agent as a public web service
- Python for all tool logic
Challenges we ran into
- Connecting MongoDB Atlas to Google Cloud Run required whitelisting dynamic IPs
- Getting the ADK agent to chain multiple tool calls automatically required careful instruction design
- Deploying with proper IAM permissions on Google Cloud took several iterations
What we learned
- How to build multi-step reasoning agents with Google ADK
- How to integrate MongoDB as a real-time data store for an AI agent
- How to deploy a fully functional AI agent to production in under a day
What's next for worldcup-pulse
- Add real FIFA 2026 official match schedule data as it gets released
- Add flight and hotel search integration
- Add push notifications for match reminders
Log in or sign up for Devpost to join the conversation.