Conference

In this talk, I will introduce agentic and physical AI for soft bioelectronics fabrication, a new paradigm that integrates autonomous AI agents and augmented-reality (AR)-guided fabrication to enable intelligent, scalable, and reproducible soft bioelectronics manufacturing. I will first present our development of scientific AI agents capable of autonomously analyzing complex multimodal datasets with strong generalization beyond conventional machine learning algorithms. Building upon this foundation, we created DeviceAgent—an autonomous multimodal AI framework that integrates large language models (LLMs), vision–language models (VLMs), and domain-specific computational tools for bioelectronics design and research. DeviceAgent demonstrates five core capabilities: (1) generating device layouts optimized for functional and material constraints; (2) composing fabrication protocols that adapt to specific materials, tools, and cleanroom environments; (3) performing visual defect inspection through VLM-based reasoning on microscopic images; (4) analyzing electrophysiological recordings from soft electronic interfaces; and (5) coordinating these capabilities through a multimodal context–memory system that enables reasoning across tasks, contextual adaptation, and human-in-the-loop oversight. I will then showcase the autonomous development of stretchable mesh electronics for interfacing with human iPSC-derived cardiomyocytes—an end-to-end demonstration combining nanofabrication, materials science, and electrophysiology within a single AI-driven workflow. Finally, I will introduce the integration of AI agents with AR-guided lithography as a Physical AI system that enhances precision, scalability, and reproducibility in soft bioelectronics fabrication.