Odyssey is developing a new class of generative AI called interactive world models, which produce and stream interactive video environments in real time rather than rendering them through a traditional game engine. Where most generative video systems output passive clips, Odyssey's models generate the next frame in response to live user input, creating the illusion of moving through and interacting with a living digital world. The company unveiled a research preview of this capability in May 2025.
The company was founded by Oliver Cameron and Jeff Hawke, both veterans of the autonomous-vehicle industry, where world models are used to simulate and predict driving environments. Odyssey applies that lineage of spatial AI to entertainment, with the thesis that game and film production can be radically streamlined when explorable worlds can be generated and modified on demand instead of built asset-by-asset. The London-based lab pairs deep learning research with a content and tooling roadmap aimed at creators.
Odyssey's interactive video can be controlled with a keyboard, phone, or controller, and the team envisions voice-driven interaction over time. Because frames are generated and streamed by a model, worlds can in principle be reshaped through prompts and parameters, opening possibilities for procedurally generated, infinitely variable game environments and interactive films. The technology sits at the frontier where generative video, real-time interactivity, and game design converge.
The company's ambitions are reflected in its backers and advisors. Odyssey has raised roughly $27 million across multiple rounds from investors including EQT Ventures, GV, and Air Street Capital, and its board features Ed Catmull, co-founder of Pixar and former president of Walt Disney Animation Studios. As world models mature, Odyssey is positioning itself as foundational infrastructure for a future where interactive entertainment is generated rather than hand-built.