CrewConnect

Logo
Desktop

Inspiration

As student filmmakers, we realized the industry is fragmented. We use one app for chatting, another for scriptwriting, and a third for storing footage. It’s chaos. We wanted to build a "GitHub for Film"—a central operating system where creative teams can go from ideation to final cut without ever switching context.

What it does

CrewConnect is a comprehensive web platform that manages the entire film production lifecycle. It features:

Community Hub: A social space to recruit cast and crew (Directors, Editors, Cinematographers) and showcase portfolios.
Project Workspace: A dedicated dashboard for managing scripts, storyboards, and call sheets in one place.
AI Co-Pilot: Powered by Gemini 3, the platform analyzes scripts for plot consistency, automatically tags raw footage using vision capabilities, and offers creative suggestions. ## How we built it We built the frontend using Next.js and React for a responsive, modern UI, hosted on Google Cloud Workstations. The backend utilizes Firebase for real-time data synchronization and authentication. The core innovation lies in the Gemini 3 API integration:
Multimodal Analysis: We use Gemini's vision capabilities to scan uploaded storyboards and raw footage to generate metadata and organize assets automatically.
Long-Context Reasoning: We feed entire scripts into Gemini to check for character consistency and generate breakdown summaries for the crew. ## Challenges we ran into Integrating real-time collaboration (live updates when a crew member changes a status) alongside heavy AI media processing was difficult. Managing the state between the asynchronous Gemini API responses and the live project dashboard required careful handling to ensure the UI remained responsive. ## Accomplishments that we're proud of We successfully built a working "Studio" environment where users can actually invite collaborators and assign roles. Getting the AI to "understand" the context of a film script and offer relevant production advice—rather than just generic text—feels like a huge step forward for creative tools. ## What we learned Building CrewConnect taught us how to bridge the gap between creative workflows and technical implementation.
Multimodal Power: We learned how to effectively use Gemini 3's multimodal capabilities to interpret visual data (storyboards) alongside text (scripts), which opened up new possibilities for automated asset management.
Async Complexity: We gained a deeper understanding of handling asynchronous AI responses within a real-time React application, ensuring the user interface remains snappy while the model processes heavy data.
Prompt Engineering: We discovered that precise prompt engineering is crucial for getting consistent, structured JSON outputs from the LLM, which is essential for integrating AI logic into our application's state management. ## What's next for CrewConnect We plan to implement the "AI Editor" fully—where Gemini doesn't just suggest cuts but actually creates a rough assembly edit in the browser. We also aim to launch a mobile companion app for on-set management.