Multi-Channel AI Movie Curation Taste Intelligence Platform

Inspiration

Modern entertainment discovery suffers from a severe decision fatigue crisis. Viewers spend an average of 18 18 minutes scrolling through repetitive catalogs before selecting a film—with nearly half of all sessions ending in choice paralysis. Existing recommendation engines operate as opaque, black-box algorithms that isolate viewers. We built DARK to transform movie discovery into an interactive, explainable, and collaborative experience. Our goal was to leverage the cognitive reasoning of the Google Gemini API to create Nyx, a centralized AI Operating System that turns natural language queries into native, real-time user interface controls.

What it does

Conversational Discovery: The central AI assistant, Nyx, processes natural language requests, reasons about user preferences, and executes structured layout changes directly on the page. Dynamic Analytics: Tracks user behavioral telemetry across 8 statistical dimensions to display an interactive Movie DNA chart and assign taste personas. Collaborative Curation: Users can organize watchlists into custom Favorite folders, sync profiles with a partner using Co-Viewing sharing codes, or query recommendations inside team channels using our Slack Bolt adapter.

How we built it

We engineered the platform using a decoupled Hexagonal Ports & Adapters architecture to separate core business logic from presentation layers. The backend is built on Node.js, Express, and MongoDB, while the frontend uses a glassmorphic Vanilla JS layout.

Challenges we ran into

Our primary technical hurdle was maintaining model latency and system availability under strict API rate limits. Sending raw user profiles consumed excessive quota. We solved this by developing a Dynamic Context Builder

Accomplishments that we're proud of

Developing the real-time Movie DNA Engine, which dynamically recalibrates telemetry metrics, donut charts, and unlocks progressive avatar styles based on live interaction scoring. Building a secure, production-ready Slack integration from scratch, utilizing Bolt event listeners, timing-safe HMAC-SHA256 signature verification, and dynamic Block Kit layouts.

What we learned

We learned that generative AI is most effective when integrated as a structured logic director rather than just a text generator. By training Gemini prompts to return deterministic JSON commands instead of conversational prose, we converted a standard chat bot into a dynamic visual curation system.

What's next for Multi-Channel AI Movie Curation Taste Intelligence Platform

MCP Server Port: Expose Taste DNA profiles to native desktop clients via a Model Context Protocol tool server. Group Curation Blending: Implementing collaborative team taste DNA profiles to blend multiple members' interactions into a unified movie recommendation feed.

Built With

• ai
• bolt-framework
• chart.js
• collaboration
• css3
• express.js
• full-stack
• gemini-api
• html5
• javascript
• json
• mongodb
• mongoose
• node.js
• render
• rest-api
• slack-api
• software-architecture
• vercel