GDC 2025 Session Viewer
Game AI Summit: Multiagent Planning for Large-Scale Narrative Content
Trevor Santarra (Principal Engineer, Bitpart AI)
Pass Type: All Access Pass, Summits Pass - Get your pass now!
Track: Programming
Format: Lecture
Vault Recording: TBD
Audience Level: Intermediate
This session will explore the techniques used at Bitpart AI to create multiagent behavior in games, with a focus on Hierarchical Task Network (HTN) planning for multi-modal actions, including dialogue, movement, and object interactions. Attendees will learn how to structure complex behaviors between AI agents in dynamic, narrative-driven environments, allowing characters to make real-time decisions and engage in meaningful interactions with players. The talk will cover multi-modal planning, behavioral concurrency, and strategies for scaling AI systems in large, data-driven worlds. Developers will walk away with actionable insights into balancing AI complexity while ensuring immersive, responsive game worlds that feel alive and natural.
Takeaway
Attendees will learn practical strategies for managing multi-modal behaviors, real-time concurrency, and balancing AI complexity in dynamic, narrative-driven games. Concrete examples will show how to create responsive, immersive worlds where characters act naturally and enrich the player experience.
Intended Audience
This session is for AI and gameplay programmers, technical designers, and developers working on interactive, narrative-driven environments. Some familiarity with AI planning techniques, such as Hierarchical Task Networks, and game development processes is recommended. Attendees should have an intermediate to advanced understanding of AI and narrative design in games.