-
-
Escene 1: Emotional Mood
-
Escene 2: Cinematic Atmosphere
-
Escene 3: Dietary & Age Preferences
-
Escene 4: Executive Review & Launch
-
Cinema Night Package
-
Specialist Agents: Soundtrack Maestro & Cinema Sommelier
-
-
Collaborative Partner: Active Learning Loop
-
3D Emotional Constellation & Google Lyria Leitmotifs
-
Behind the Scenes: Multi-Agent Execution Trace
-
Real-Time Google ADK Execution Logs
-
AI Alignment & Constraint Fidelity Matrix
Inspiration
I love cinema because it can change how we feel. Sometimes we want to escape, sometimes we need comfort, and sometimes we simply want to discover something that matches our mood.
But finding that movie can be surprisingly difficult. Streaming platforms give us thousands of options, yet we often end up scrolling endlessly without finding anything that feels right.
That was the idea behind Feel And Film: instead of searching through endless filters, you tell the agent how you feel and what you're looking for, and it builds a complete movie experience around you.
What it does
Feel And Film is an agentic movie recommendation experience that goes beyond suggesting a title.
The agent considers things like mood, themes, directors, studios, eras, age restrictions and other constraints to find a movie that fits the user's needs.
It can also explore the movie's soundtrack, suggest food and drinks that match the experience, help find where to watch it, and send the complete movie-night plan by email.
After watching movies, they become part of an interactive Emotional Constellation: a 3D visualization where each movie becomes a star with its own musical identity generated with Google Lyria.
What it does
How we built it
Feel And Film uses Gemini and Google ADK to orchestrate several specialized agents for film curation, soundtrack exploration, food pairing, cinema planning and the emotional constellation.
The backend was built with Python and FastAPI, containerized with Docker and deployed on Google Cloud Run. ClickHouse Cloud is used to store and analyze movie sessions and preferences.
For the Emotional Constellation, I combined Google Lyria with the browser's Web Audio API to create short musical pieces associated with the movies in the user's history.
Challenges we ran into
One of the biggest challenges was coordinating several specialized agents while keeping the final result coherent. Each agent has a different responsibility, but the user should experience it as one single recommendation.
Another challenge was the musical side of the project. I wanted the constellation to feel connected to cinema rather than being just another visualization, so I experimented with using AI-generated music to give each movie its own identity.
I also had to deal with conflicting constraints. For example, a user might want something relaxing while specifically asking for a director known for intense films. The system needs to find a good balance instead of simply matching keywords.
Accomplishments that we're proud of
I'm proud that Feel And Film became more than a movie recommender.
It combines cinema, music, food and visualization into one experience, with the Emotional Constellation giving the user's movie history a visual and musical identity.
I'm also proud of bringing together several Google technologies in one project, including Gemini, Google ADK, Lyria and Cloud Run.
Accomplishments that we're proud of
I'm proud that Feel And Film became more than a movie recommender.
It combines cinema, music, food and visualization into one experience, with the Emotional Constellation giving the user's movie history a visual and musical identity.
I'm also proud of bringing together several Google technologies in one project, including Gemini, Google ADK, Lyria and Cloud Run.
What we learned
This project taught me a lot about building agentic systems with Google ADK, especially how to give specialized agents clear responsibilities while keeping a coherent final experience.
I also learned how interesting it can be to combine AI with areas that aren't usually part of recommendation systems, such as music, food and visual storytelling.
Most importantly, I learned that an AI recommendation system can feel much more engaging when it acts less like a search engine and more like a creative partner.
What's next for Feel And Film
The next step for Feel And Film is to take the idea beyond cinema.
I want to explore how the same approach could be applied to other areas where recommendations are usually based on simple filters, but where context, emotions and personal preferences matter just as much.
Log in or sign up for Devpost to join the conversation.