CineVerse

Inspiration

We love movies, but discovering cinema beyond the usual recommendations is surprisingly difficult. Most movie platforms help users find what to watch, but not understand why a movie matters, how it connects to other films, or how a genre has evolved. That inspired us to build CineVerse — The Ultimate Movie World: a platform that combines the cinematic experience of a streaming service with the intelligence of an AI research assistant.

What it does

CineVerse allows users to explore cinema using filters such as genre, themes, tone, language, era, and country. Instead of returning a simple list of movies, CineVerse creates an AI-powered Cinema Landscape containing relevant films, recurring themes, narrative patterns, visual styles, and interesting connections. Users can also search for individual movies, explore detailed movie pages, save favorites and watchlists, view their history, and interact with Ask Cinema Intelligence for deeper questions about cinema. At the core is a multi-agent architecture built with Google Gemini and Google ADK:

• Discovery Agent — discovers relevant movies using TMDB.
• Research Agent — researches the broader genre, themes, history, cultural context, and cinematic trends.
• Analysis Agent — analyzes movies in the context of the user's current exploration.
• Synthesis Agent — combines the findings into a curated cinema landscape and ranked results. We use Parallel for web research and combine its findings with structured movie data and Gemini's reasoning capabilities.

How we built it

The frontend was built with React + Vite, with a cinematic interface inspired by modern streaming platforms. The backend uses Node.js/Express, while Supabase handles authentication and user data. The application integrates:

• Google Gemini
• Google ADK
• TMDB
• Parallel
• Supabase We focused on keeping the AI agents genuinely functional rather than creating simulated agent animations or static AI-generated responses. ## Challenges we ran into One of our biggest challenges was maintaining a balance between depth and performance. Researching cinema broadly can require many API calls, while users expect results quickly. We addressed this through parallelized agent execution, focused research tasks, progressive filtering, and caching where appropriate. Another challenge was designing an experience that feels like a premium movie platform while still exposing meaningful AI intelligence. We wanted the AI to enhance movie discovery rather than overwhelm the user with technical complexity. CineVerse is our attempt to make cinema discovery feel less like searching a database and more like exploring an intelligent map of the movie world.

Accomplishments that we're proud of

We are proud of turning the idea of an AI-powered cinema intelligence platform into a functional product with a polished, streaming-platform-inspired experience. Our biggest accomplishment is building a multi-agent architecture where different agents have clearly defined responsibilities—from movie discovery and web research to contextual analysis and final synthesis. We also successfully integrated real movie data with AI-generated insights instead of relying on static or simulated responses. We are especially proud of creating an experience that makes complex cinema research feel simple and engaging for movie enthusiasts.

What we learned

We learned that building an effective agentic system requires much more than simply adding an LLM. Agent responsibilities, data flow, API integration, reliability, and response quality all need to be carefully designed. We also learned how to combine structured movie data from TMDB with broader web research and Gemini's reasoning to generate more meaningful, context-aware insights. Most importantly, we learned how to balance AI capability, user experience, and performance while building a real-world application under hackathon constraints.

What's next for CineVerse

We want CineVerse to evolve into a deeper AI-powered cinema intelligence ecosystem. Future versions could include personalized cinema journeys, deeper filmmaker and screenplay research, cinematic trend discovery, stronger recommendation intelligence, collaborative exploration, and richer connections between films, creators, movements, and cultures. Our long-term vision is for CineVerse to become a place where anyone can not only find a movie to watch, but explore the world behind it.

Built With

• adk
• agent
• agents
• api
• artificial
• builder
• cloud
• css
• discovery
• express.js
• gemini
• google
• movie
• multi-agent
• node.js
• parallel
• react
• search
• supabase
• systems
• tailwind
• tmdb
• typescript
• vite
• web