AMI Labs (Advanced Machine Intelligence) is an AI research and product company building “world models” – AI systems that learn from video, sensors, and real‑world interaction, not just text. It targets the core limitation of today’s LLMs: they talk about the world but don’t actually understand physics, space, or causality, which caps their usefulness in robotics, embodied agents, and complex decision‑making. By training JEPA‑based world models that can reason about environments, objects, and plans, AMI aims to power the next generation of agentic systems, robotics, transport, and on‑device AI. This is a concentrated bet on the “post‑LLM” layer of AI infrastructure, led by one of the field’s most influential scientists.