Laboratory automation enhances experimental throughput and reproducibility in synthetic biology, yet widespread adoption remains limited by the programming expertise required to operate robotic platforms. Large Language Models (LLMs) offer promising solutions but lack domain-specific knowledge to program robotics for reliable synthetic biology workflows. We developed LabscriptAI, a multi-agent framework that enables LLMs to autonomously generate and validate executable Python scripts for customized hardware configurations through iterative self-correction. We demonstrate LabscriptAI capabilities through proof-of-concept adapting a community protocol for Plate Reader Fluorescence Calibration, followed by validation using cell-free protein synthesis (CFPS) workflows encompassing expression, purification, and quantification of green fluorescent protein (GFP). For real-world protein engineering, we first applied LabscriptAI to the Critical Assessment of Protein Engineering (CAPE) competition, automating fluorescence and thermostability characterization of 308 unique GFP variants designed by 53 student teams from 5 countries. Furthermore, LabscriptAI orchestrated distributed automation across a biofoundry work cell and fume hood-enclosed liquid handling robot to screen formaldehyde-converting enzyme variants. This integrated workflow successfully identified improved double mutants, including variants with more than 3-fold enhanced catalytic efficiency and improved substrate tolerance. The framework provides an intuitive natural language interface for generating robust automation protocols, facilitating human-robot collaboration by engaging research communities on the cloud and enhancing laboratory safety through flexible automation. Additionally, LabscriptAI autonomously interfaces with community databases following established standards to retrieve and deposit standardized data and protocols. Overall, our approach democratizes laboratory automation by eliminating programming barriers while ensuring reproducible, community-driven protocol development for synthetic biology research.
Gao et al. (Thu,) studied this question.