Teams Race to Build AI App in 7 Hours at Google Hackathon

URGENT UPDATE: Teams at the Google Gemini 3 Hackathon in Singapore are racing against the clock to develop AI applications in just seven hours. As participants work feverishly to build prototypes, the pressure mounts with a deadline set for 5:30 PM today.

This intense competition kicked off early this morning, with participants arriving just after sunrise. A group of four Malaysian developers, united by their passion for AI, embarked on this challenge with the goal of creating a unique “feng shui” app that could assess personal surroundings and outfits to enhance luck—a concept deeply rooted in traditional Chinese culture.

By 10:30 AM, the team pivoted from their original idea to a more feasible app, now dubbed “Feng Shui Banana“, which uses real-time camera analysis to evaluate users’ wardrobe choices and workspace environments. Participants Chan Wei Khjan, Chan Ler-Kuan, Loh Wah Kiang, and Lee How Siem quickly began coding, utilizing the Gemini Live API to facilitate their project.

As the clock ticked down to 12:20 PM, the team skipped lunch, consumed by debugging their application. Initial tests revealed numerous bugs, but the team’s persistence paid off when they managed to get the camera feature operational. “One prompt matters,” said Wei Khjan, emphasizing the importance of clear communication with AI to refine their project’s capabilities.

By 1:00 PM, the app had made significant progress, accurately identifying clothing colors and suggesting enhancements for better feng shui alignment. The team took a brief moment for a pizza break, but quickly returned to work, realizing they had only four hours remaining.

With 3:00 PM approaching, the atmosphere grew tense as they filmed their project demo. Utilizing AI-generated storyboards, they crafted a compelling narrative to showcase their hard work. By 4:00 PM, the app was nearly complete, featuring an engaging voice output reminiscent of a feng shui master.

As the deadline loomed, the team scrambled to finalize their video and code repository. With just 15 minutes to spare, they submitted their project, relieved but exhausted. Although they did not win a prize from the $100,000 credit pool, they expressed satisfaction with their experience. “Sometimes, the best experiences come from saying ‘yes’ without overthinking,” Ler-Kuan remarked.

This hackathon, organized by Google DeepMind and 65labs, attracted 189 participants who collectively built 76 projects in a race against time, showcasing the innovative spirit of the tech community in Singapore.

As the event wraps up, the buzz surrounding the projects and their potential impact on AI application development is palpable. Stay tuned for further updates on the outcomes of the hackathon and the future of these innovative ideas.