Training models on large-scale data has given us powerful generative capabilities for text, images, and video. However, this success has not yet extended to training generalist embodied agents. This talk tackles this gap by focusing on a potential solution to this problem: scalable world models. We'll trace the idea of planning in predictive models, from its origins to modern efforts on building world models directly from pixels. I'll discuss the primary challenge of scaling these models and present our work, Genie, which enables us to learn world models without explicit action labels at scale, demonstrating a new path forward for training the generalist agents of the future.

The aim of the project is an educational platform that helps women and their partners understand mental and physical changes throughout the menstrual cycle. By tracking basal body temperature (manually or via API) and daily mood, the app predicts cycle phases and provides personalized, scientifically based recommendations for nutrition, supplements, activities, and mood management, accounting for interactions with medications or hormonal treatments. It consolidates multiple functionalities into a single platform, reducing the need for several apps while offering detailed explanations for each suggestion.

Exploration of predictive modeling approaches for customer lifetime value using the Buy ‘Til You Die concept.