Brown hackathon winners build HandyDaddy: AI that videos your clogged toilet, runs it through Gemini, RAG, and Groq—because calling dad was too low-tech

A team of developers has won a hackathon at Brown University with their project, HandyDaddy, an interactive home troubleshooting platform. The platform aims to provide clear, step-by-step guidance for household repairs, using live video to give personalized advice. Inspired by the need for reliable and safe repair guidance, the team built HandyDaddy as a personal real-time vision-first home repair assistant. The system uses a Gemini vision model to analyze live video feeds, identifying potential issues and safety levels, and then provides structured outputs and retrieves relevant documentation using Retrieval-Augmented Generation. The team overcame challenges in building a trustworthy system, separating the architecture into clear stages, and ensuring correctness and trust over speed and simplicity. HandyDaddy has the potential to help various individuals, including college students, families, and older adults, with its transparent and reliable guidance, and the team plans to expand the system to handle more household scenarios in the future.