We’re building an AI-powered game platform to enable a new creator ecosystem for interactive entertainment. Stories that respond to your actions. Characters that remember what you’ve done and act on it. Experiences where you’re the protagonist — not a spectator, or just picking from a pre-written menu of options.
Every major technological shift has transformed how humans experience stories. The printing press brought them to the masses. Cinema made them visual spectacles. Podcasting and YouTube solved distribution and turned anyone with a microphone or camera into a creator. Each time, a new technology didn’t just improve the old medium — it created an entirely new one. Cinema isn’t recorded theater — it introduced entirely new affordances and techniques: cuts, close-ups, non-linear time, montage. Each new medium develops its own grammar for storytelling that simply didn’t exist before.
AI is creating another new medium. Traditional computer RPGs and interactive productions like Netflix’s Bandersnatch offer the illusion of choice, but they’re fundamentally limited — every path was scripted by someone in advance, every outcome pre-authored. AI removes that ceiling. The content adapts to your decisions in real time. No two playthroughs are the same because no playthrough is pre-written. This kind of experience has never existed before.
YouTube and affordable production tools enabled anyone to become a video creator, director, or producer. AI will do the same for games and interactive entertainment. Today, building a compelling game requires large teams and enormous budgets. We’re building a platform where anyone with a story worth telling can create an immersive interactive experience.
The result is entertainment that is both more diverse — because anyone can create — and more personal — because the story forms around the decisions you make.
I’ve been interested in narrative games for the past 20 years. About a decade ago, I tried to build a procedural story engine — and quickly got stuck. The technology simply wasn’t there.
When GPT-4 came out, everything changed. I became obsessed and started working on what would eventually become this startup. For the first time, AI could write coherent stories and follow basic instructions. It has many problems — long-term coherence, balancing player agency with narrative stakes, maintaining realism while still telling compelling drama — but the foundation was real. I explored the possibilities and limitations, building different prototypes for an interactive story engine and a story creation tool on top of AI.
Since then, the models have gotten better and faster. Many of the original limitations remain, but I believe it’s now a question of who builds the first great AI story engine, not whether it will happen. There is ample investment and competition in generating images, video, and audio. But there has been limited progress on what matters most: telling coherent and interesting stories.
We’re opening our Copenhagen office and hiring our founding team. If you want to define what the next generation of entertainment looks like, reach out on LinkedIn.
]]>First and foremost, I want to thank my amazing advisor, Cristian Sminchisescu, for his guidance throughout the years, as well as my co-supervisors, Elin Topp and Kalle Åström. I would also like to thank my excellent co-authors over the years: Mykhaylo Andriluka, Aleksis Pirinen, David Nilsson, Hongyi Xu, Erwin Coumans, Luke Metz, and C. Daniel Freeman. Finally, I would like to extend my gratitude to my opponent, Prof. Fahad Khan, and the thesis committee: Prof. Serge Belongie, Prof. Siyu Tang, and Prof. Hossein Azizpour, for their valuable feedback and insights.
My thesis is published here and the pre-recorded defense presentation can be found below.