Advanced Machine Intelligence (AMI Labs), founded by Turing Award laureate Yann LeCun and Nabla founder Alex LeBrun, has closed a $1.03 billion seed round at a $3.5 billion pre-money valuation. The company is focused on building “world models,” a new class of AI systems designed to move beyond probabilistic large language models (LLMs) toward simulation-based reasoning and structured decision-making.
Nabla, a clinical AI company known for its ambient documentation assistant, holds an exclusive strategic partnership with AMI, securing first access to these models. As part of the transition, Alex LeBrun will serve as CEO of AMI Labs while continuing as Chairman and Chief AI Scientist at Nabla.
The shift reflects growing recognition of LLM limitations in clinical environments. While LLMs have demonstrated effectiveness in documentation and knowledge retrieval, they operate as probabilistic text generators. In high-stakes settings, this approach can fall short when deterministic reasoning, multimodal data processing and long-term planning are required.
AMI stated, “These systems predict how situations evolve, and how actions lead to consequences, so that they can plan sequences of actions under real-world constraints.” Rather than predicting the next word in a sequence, world models are designed to learn abstract representations of environments, enabling cause-and-effect reasoning and simulation-based “what-if” analyses before actions are executed.
This approach is intended to address challenges associated with continuous and complex clinical data streams, including physiological monitoring, imaging and audio inputs. By simulating potential outcomes in advance, world models aim to support auditable and safety-aligned decision-making frameworks suited for regulatory and hospital oversight requirements.
Nabla plans to integrate AMI’s technology into its next phase of product development, moving beyond ambient documentation toward Agentic AI. Such systems are envisioned to autonomously execute multistep workflows across fragmented electronic health record infrastructures, while maintaining persistent memory and defined safety guardrails.
Potential applications include coordinating referrals, navigating scheduling systems, assessing insurance constraints and initiating pre-visit laboratory orders. By combining embedded clinical presence with simulation-based reasoning, Nabla aims to develop AI systems capable of structured action rather than text generation alone, positioning world models as a foundational architecture for autonomous healthcare operations.
Click here for the original news story.