Studio Atelico

Today, I’m excited to unveil a project that’s incredibly close to my heart. As a lifelong gamer, I’ve always dreamed of pushing the boundaries of what’s possible in video games using my expertise in AI. Over the past year, the Studio Atelico team has been blending the magic of gaming with AI to create something special and groundbreaking: an on-device AI engine for games that  empowers NPCs, agents, and natural language interactions in ways that were previously impossible or prohibitively expensive. To showcase its potential, we built the Generative Agents Realtime Playground (GARP), a tech demo inspired by Generative Agents. GARP demonstrates how our AI engine runs in real time, locally on-device—all without relying on the cloud and for free. Can’t wait to see what game developers will build with the Atelico AI Engine! Last year at GDC, I spoke with many developers trying to add generative AI to their games and they all shared a common frustration: - Cloud/SaaS generative AI solutions are prohibitively expensive and offer little control. - Trade-offs between quality and latency are outside their control. - Outputs feel overly sanitized - The pay-per-interaction Cloud/SaaS model is unsustainable. That feedback stuck with me. Drawing from my experience building task-specific SLMs at Predibase, I realized we could solve these problems by making game specific AI models run locally. So, I joined forces with Paul Szerlip, JP Chen, Ennio De Nucci, and Patrick John Chia to build an MVP AI Engine that seamlessly integrates with Unreal Engine, Unity, and Godot Engine. To showcase the engine, we took inspiration from Joon Sung Park and Percy Liang’s Generative Agents paper and created GARP: a real-time, on-device simulation. In GARP, over 20 agents inhabit a vibrant village, each with unique personalities, backgrounds, and behaviors. These agents interact, form memories, build relationships, plan, learn, and even exhibit emergent behavior. It wasn’t easy, but the results speak for themselves. Check out the video demo below and let us know what you think! We’ve also captured our journey and key insights in a detailed blog post. As we gear up for the first release of the Atelico AI Engine, we’d love your feedback. Let’s build the future of video games together! Thanking everyone who has helped out so far on this journey: Chris Re, Kenneth Stanley, Thomas Wolf, Chris Olson, Rob Bartholomew, David Berger, Jack Soslow, J. Aaron Farr, T. Carl Kwoh, Stephen Stagliano, Michael Stagliano, Ramón Iglesias, Jacopo Tagliabue, Mattia Pavoni, Ciro Greco #ai #gamedev #llm #slm Blog post and video links below 👇