The paper introduces SpatialAgent, an autonomous AI agent designed to streamline and enhance research in the field of spatial biology. By utilizing Large Language Models (LLMs) and a specialized toolkit, this system can independently plan, reason, and execute complex workflows such as gene panel design and tissue niche annotation. Benchmarking results demonstrate that the agent frequently outperforms human experts and traditional computational pipelines, especially when identifying biological patterns and predicting spatial coordinates. A key feature of the platform is its hybrid collaboration mode, which allows scientists to refine the agent’s output, leading to even more accurate and insightful results. Ultimately, SpatialAgent aims to democratize sophisticated spatial genomics analysis by reducing the need for extensive coding expertise while accelerating scientific discovery.

References:

• Wang H, He Y, Coelho P P, et al. SpatialAgent: An autonomous AI agent for spatial biology[J]. bioRxiv, 2025: 2025.04. 03.646459.