As part of the KXSB HackXelerator 2025, creatives and developers teamed up to explore how AI can enhance – not replace – human creativity. One of the standout projects to emerge from this initiative is Trumbo, the Film category winner.
Trumbo is a tool designed to streamline creative collaboration across the film production lifecycle. Developed over a focused 20-day sprint, it merges AI-driven automation with an intuitive, semantically aware workspace that serves everyone involved in filmmaking – from screenwriters to production crews.
We caught up with project lead Dmitry to hear how the idea came together, what the team built, and where they're headed.
Meet the team behind Trumbo
Dmitry led the charge on Trumbo – from the first sketch of the idea to prototyping features and producing the final demo. “I came up with the concept, shaped the vision, and guided it through the sprint,” he says. He was soon joined by collaborators Yerson, Peter, and Ryan, working asynchronously via Discord and GitHub, with occasional check-ins over Zoom.
Dmitry first heard about the event through the London AI Engineering Meetup and Paul Dowling. But what really kicked things off? “Honestly, it was Yerson. Once he appeared interested in the idea and ready to collaborate, I thought, why not? Let’s build something together.”
The big idea: Smarter collaboration for creative teams
Trumbo began as a long-dormant concept. The initial spark came five years ago after Dmitry read The Tools of Screenwriting by David Howard and Edward Mabley. “Film production is rooted in a script – a document – but delivering that vision requires layers of semantic interpretation and a ton of manual work,” he explains. "Breaking down a script into characters, props, locations – all of that is still very hands-on."
Enter generative AI. With large language models now capable of handling those semantic layers, Trumbo finally had the technological fuel to come to life.
The goal? Create a shared, intelligent workspace that reduces friction in creative collaboration – a system that understands the underlying relationships between elements in a script and enhances productivity with AI-powered assistance.
What makes Trumbo unique
Most screenwriting tools focus on just one stage of the process: writing, analysis, or pitching. Trumbo is different. It aims to support the full creative lifecycle.
Beyond just documents and project boards, Trumbo introduces:
- AI assistance for tagging and interpreting scripts
- Semantic relationships between creative elements
- Holistic execution that unites writers, directors, and producers in one intelligent space
Dmitry sums it up: “What sets Trumbo apart from tools like Notion is that it’s built for the creative industry from the ground up, with AI Assisted Checks and Semantic Relations as first-class features.”
Under the hood: Tech and AI stack
To bring Trumbo to life, the team used:
- Vultr cloud and AMD Instinct MI325X Accelerator, which the team explored as a high-performance inference platform for future storyboard generation
- Kotlin for core development, chosen for its ability to unify backend and frontend logic across platforms (web, desktop, mobile)
- Python for AI prototyping
- LLMs like o4 and Mistral Large, integrated via API, to analyze scripts and generate semantic markup
One technical breakthrough? “Discovering AMD’s Instinct hardware with 256GB vRAM was huge – it opens the door to running heavier models without compromise.”
What’s next for Trumbo?
With the prototype now in hand, Dmitry and the team are forging ahead. Their next milestone is building an alpha version and testing it with real-world creative teams.
“The vision is strong, and the feedback so far has been encouraging,” Dmitry says. “We’re refining the UX and reaching out to film professionals to validate our approach. There’s a long road ahead, but the foundation is solid, and we’re excited to keep building.”
Want to learn more? Watch the demo video here.