Conversational AI Museum Guide

We built a conversational AI museum guide that transforms the traditional fixed audio-tour experience into an interactive system controlled through natural conversation. Instead of selecting predefined audio tracks or navigating complex menus, visitors can speak directly with an AI guide to request a recommended tour, ask questions about artworks, explore specific visual details, receive contextual explanations, ask for directions, and navigate the museum using natural language. The key idea is that the conversation directly controls the experience. The AI is also connected to the visual interface, allowing visitor requests to trigger actions such as zooming into artwork details, creating visual hotspots, displaying contextual artwork information, opening the museum floor plan, and generating routes to destinations. For the technology stack, we use the AssemblyAI Voice Agent API for real-time voice interaction, turn-taking, and tool calling, with an AI agent using structured tool calls to control the application. Groq (Qwen) or Gemini can be used for image analysis depending on availability, while the frontend is built with Next.js. This architecture allows natural spoken requests to produce real actions inside the application rather than simply generating a text or voice response. The goal is to make exploring physical spaces feel natural, conversational, and personalized, giving visitors a guide that can follow their curiosity and adapt to their needs in real time.