Original title: Hierarchical automation of scanning probe microscopy through agentic orchestration and algorithmic control
Boris N. Slautin, Sheryl L. Sanchez, Aidan Swanger, Yu Liu +4 · September 2026
Running laboratory experiments with artificial intelligence is challenging. Smart AI agents excel at planning and thinking through open-ended goals. However, agents are less reliable at strict mathematical tasks, instrument control, and precise physical movements. To solve this, researchers built a two-part hierarchical system. The AI agent acts like a director. It understands the scientific question, weighs incoming evidence, and decides the next high-level move. Meanwhile, dependable non-agent computer algorithms handle the exact calculations, targeting, validation, and physical instrument controls. The team tested this setup on a high-resolution scanning technique called piezoresponse force microscopy. The system studied local material structures and electrical polarization switching. It built spatial maps from imaging, targeted measurements, adjusted testing signals, and stopped automatically when new tests ceased providing fresh evidence. It also successfully identified a hidden confounding factor and realized when the available imaging area could not answer a specific query.
Imagine cooking a complex meal where each dish has its own timing and temperature needs. If you try to handle every detail yourself, you might burn one dish while checking another. In this paper, the AI acts like a head chef who decides what to cook and in what order, while specialized kitchen tools handle the exact temperatures and timers. This way, the chef focuses on the overall plan, and the tools ensure each dish is prepared correctly without constant oversight.
This could let scientists focus on big-picture research questions while machines handle the tedious work of running experiments. It makes microscopy more reproducible by using reliable algorithms for precise tasks and AI for planning.
The abstract does not specify whether this approach generalizes beyond piezoresponse force microscopy or to other types of experiments. It notes that the system recognized limitations in the available field of view, but does not address scalability to different materials or experimental setups.
Ask anything about this paper. Answers come from its abstract, and say so when the abstract doesn't cover your question.
Hierarchical automation of scanning probe microscopy through agentic orchestration and algorithmic control — Boris N. Slautin, Sheryl L. Sanchez, Aidan Swanger et al.
Abstract licence: CC0
Explanation generated by Synthesis from the abstract — it is not the original paper. This is a preprint, so it hasn't necessarily been peer reviewed.