​From Tokens to Robots

​A fireside discussion on VLAs, robotics, and world models.

​What does it actually take to build systems that can see, understand, and act?

​This event explores how developers and researchers can use web-scale data to power the next generation of vision, multimodal, and robotics applications. We’ll look at how teams are thinking about training data, data pipelines, and real-world use cases for Vision-Language-Action systems — and why access to the right data layer matters.

​Expect a fireside chat and capability showcase focused on what’s possible today, with concrete examples of how web data can support multimodal workflows. Rather than a hands-on workshop, this session is designed to help you understand the landscape, see example implementations, and leave with a clearer sense of how these systems are actually built.

​What you’ll get out of the event

• ​A practical understanding of how web data fits into VLA, multimodal, and robotics workflows

• ​Examples of the kinds of datasets, pipelines, and infrastructure needed to support these systems

• ​A look at real use cases and demos showing what these capabilities can unlock

• ​Perspective on how teams are approaching data collection, structuring, and downstream model workflows

• ​Connections with other developers, researchers, and builders exploring similar problems

​If you’re curious about where multimodal systems are going — and what it takes to support them with the right data foundations — this event is for you.