Atmos is an immersive music experience that turns a simple human moment into a visual, playable soundtrack journey. The user starts with a short context, such as “cooking”, “coding late at night”, or “driving to a wedding”. Atmos sends this context to an agentic backend built with n8n and Google Vertex AI. The agents analyze the moment, extract mood, genre, energy, atmosphere, and intent, then use Gemini embeddings to search a prepared music catalog and return tracks that match the feeling of the scene. Instead of showing a normal playlist, Atmos presents music as a visual gallery where images, moods, sponsor context, and audio playback are combined into one cinematic interface. Music discovery should not feel like typing keywords into a search box. Movies have posters, worlds, and atmosphere, but music browsing is often blind. Atmos makes the entry point emotional and visual: a moment becomes a world, and the user explores it by playing tracks, saving favorites, and opening overlays. The MVP is built with Flutter Web and Dart, with n8n webhooks for backend integration. Google Vertex AI acts as the intelligence layer: agents interpret user input, create structured music concepts, generate/search embeddings, route the request to the right musical direction, and prepare prompts for Lyria 3 Pro pipelines. This connects human context, atmosphere, semantic search, and AI-generated music into one flow. Atmos also explores a business layer where sponsored context does not interrupt like a traditional ad. Instead, a sponsor becomes part of the generated visual and musical world when relevant to the user’s moment. For the hackathon, we focused on a working interactive MVP with strong visual identity: moment input, gallery search, audio playback, track switching, favorites, cache fallback, and sponsor-aware overlays. The next step is to connect each track with a real story source, making every song feel less generic and more like music with a world behind it.
Category tags: