Muse | Devpost

Inspiration

When we first saw this prompt, we wanted to create something that could address a realistic challenge for scriptwriters. We realized that writers often use references and easter eggs to make their scripts more engaging, but coming up with recognizable movie references that fit naturally into a scene can be difficult. This inspired us to create Muse, a tool that helps scriptwriters discover movie-inspired lines that they can incorporate into their own scenes.

What it does

Muse is a program that takes user input describing the context of a movie scene and generates an idea for a line that could be incorporated into the scene as a movie easter egg. By considering the genre and situation described by the user, Muse suggests a reference inspired by a well-known movie while keeping it relevant to the scene.

How we built it

We used IBM Bob to develop and connect the application’s main components. The frontend uses React, TypeScript, and Vite, while the backend uses Python and FastAPI.

Muse searches screenplay scenes stored in ChromaDB and queries an SQLite database based on the IMDb Top 1000 dataset. The retrieved context is passed to Google Gemini, which produces personalized screenplay guidance.

Challenges we faced

One challenge was connecting the notebook workflow to a full web application. We also had to manage local database paths, securely configure the Gemini API key, and update the backend when an older Gemini model became unavailable.

Accomplishments that we're proud of

We transformed an experimental notebook into a complete, interactive application. Muse combines semantic search, structured movie data, generative AI, and a polished cinematic interface in one workflow.

What we learned

We learned how to build an end-to-end AI application with a React frontend, FastAPI backend, vector database, traditional SQL database, and a large language model. We also learned the importance of protecting API credentials and handling changing AI model availability.

What's next for Muse

Next, we want to deploy Muse publicly, expand the screenplay database, add more advanced filters, improve quote attribution, support screenplay uploads, and let teams save and organize their favorite references.