DeviceAgent: An autonomous multimodal AI agent for flexible bioelectronics

Area of Science:

• Bioelectronics
• Artificial Intelligence
• Materials Science

Background:

• Flexible bioelectronics development is complex, requiring specialized expertise and limiting scalability.
• Current research faces challenges in automation, adaptability, and accessibility.
• Integrating AI tools can potentially overcome these limitations.

Purpose of the Study:

• To introduce DeviceAgent, an autonomous multimodal AI agent for bioelectronics research.
• To demonstrate DeviceAgent's capabilities in automating key experimental pipeline stages.
• To establish a scalable paradigm for AI-scientist collaboration in nanofabrication.

Main Methods:

• DeviceAgent integrates large language models (LLMs), vision-language models (VLMs), and computational tools.
• A multimodal context memory system supports end-to-end experimental pipeline tasks.
• The system allows human oversight at critical decision points.

Main Results:

• DeviceAgent autonomously generates bioelectronic layouts and fabrication protocols.
• It identifies microscopic defects using visual reasoning and analyzes electrophysiological recordings.
• The AI agent demonstrated capabilities in developing stretchable mesh electronics for cardiomyocytes.

Conclusions:

• DeviceAgent offers a scalable and accessible framework for AI-driven bioelectronics research.
• The integration of LLMs and VLMs enhances zero/few-shot generalization and contextual learning.
• This AI-scientist collaboration paradigm accelerates innovation in nanofabrication and bioelectronic device development.