ON-DEMAND CONVERSATION | 20 min 1 sec
Real-time world models: what changes when video becomes interactive?
Most AI video tools return a finished clip. Real-time world models introduce a different experience: video that responds as a user interacts. That changes what developers need to build.
Join Alberto Taiuti, CEO and Co-founder of Reactor, and Josh Liss of Nebius to explore how Reactor approaches world models. Hear why persistent sessions, state and code matter, and where Alberto sees opportunities for personalized media and interactive experiences.
What you’ll learn
- How Reactor defines a world model and how it differs from generating a video file.
- Why real-time interaction calls for persistent sessions, state and developer control.
- How a shortage of 3D training data led Alberto to rethink video models.
- Where Alberto sees potential for personalized advertising, branching shows and new application experiences.
Who should watch
For game and interactive-media developers, AI engineers, technical founders and product leaders exploring world models.
Featured speakers
Alberto Taiuti | CEO and Co-founder, Reactor
Josh Liss | Head of Media and Entertainment, Nebius
From the Nebius for Startups podcast.
The AI cloud for media and entertainment.
Build faster. Scale with confidence. Own your intelligence.
Nebius combines GPU infrastructure, managed services and hands-on engineering for game AI, media models and creative production, with control over your IP and deployment.