At Oracle CloudWorld 2024, our goal wasn’t to create just another AI product demo, but to offer a truly interactive and immersive experience that would make Oracle’s AI and cloud services tangible. We wanted something that attendees could engage with firsthand, bringing the power of AI to life by turning real-world actions into digital insights.
Foosball became our experience of choice for its simplicity, universal appeal, and portability. Our team set out to enhance the classic game using Oracle’s AI, capturing gameplay through computer vision and transforming the results into AI-generated sportscasts, all through an automated process. The result was Oracle Foosball Frenzy, an experience where analog competition seamlessly converged with cutting-edge AI and cloud technologies. Although we chose foosball as the platform for this demo, the underlying concepts are broadly applicable to real-world business scenarios across various industries.
Oracle’s foosball demo
In this blog series, we’ll explore the development of an innovative foosball demo centered around live gameplay on our foosball table that was showcased during Oracle CloudWorld in Las Vegas, shown in Figure 1.
Our team set out to create a fully automated system that tracks and analyzes each match in real-time, capturing game data and generating actionable insights, as shown in Figure 2. The project culminates in an AI-powered avatar delivering a video summary of the game, providing a dynamic overview based on the captured data.
By the end of the event, hundreds of players had competed, with every match sparking excitement as teams vied to climb the league standings. Each game brought the unpredictability and thrill of real sports, keeping players and spectators on edge. The anticipation grew as everyone eagerly awaited the AI-generated summary, delivered by an avatar and crafted by a large language model (LLM), adding an exciting digital twist to the traditional foosball experience.
The architecture (Figure 3) began with an overhead camera, utilizing computer vision to capture gameplay at 30 frames per second, tracking the ball’s movement throughout each match. This data was stored in a schemaless database designed for event capture. From there, the raw data was transformed using database views, converting it from a format optimized for data capture into one suited for analytics. We developed real-time dashboards to display live scores and ball possession during each game, while summary dashboards provided league-wide statistics, offering a comprehensive view of all games played. The following upcoming installments in this blog series will dive into the different aspects of the system:
- Part 1: Introductory blog (this blog)
- Part 2: Core engine and data processing layer
- Part 3: Achieving automation using Python
- Part 4: Translating operational data into analytical insights
- Part 5: Visualizing the data for players
The Oracle Foosball Frenzy demo at CloudWorld 2024 provided an exciting glimpse into how AI can transform even the simplest of games into rich, data-driven experiences. By integrating cutting-edge AI, computer vision, and automation, we created an engaging, fully automated system that generated real-time insights from live gameplay. In the upcoming blog series, we’ll explore the technical journey behind this innovative demo and share how it came to life step by step.
Continue reading about the Oracle AI Foosball Frenzy demo in part 2 of this series.
Get more information about Oracle’s:
- AI Infrastructure
- Database 23ai
- Kubernetes Engine (OKE)
- Generative AI using Llama 3
- Analytics Cloud (OAC)

