AI agent helps prepare synchrotron X-ray experimental measurements, paving the way for autonomous operation
Artificial intelligence (AI) models are now used daily by many people worldwide, both for professional and personal purposes. Over the past decades, scientists specialized in various disciplines have ...
Artificial intelligence (AI) models are now used daily by many people worldwide, both for professional and personal purposes. Over the past decades, scientists specialized in various disciplines have also started using these models to conduct research or simplify their experimental practices.
Researchers at Stanford University and SLAC National Accelerator Laboratory recently explored the possibility of using an AI-powered agent to prepare a synchrotron-based X-ray experiment. Synchrotrons are large research facilities at which electrons are accelerated to produce very bright X-rays, which can then be used to study the atomic structure of materials, molecules and biological samples.
In a paper published in Nature Machine Intelligence, the team at Stanford and SLAC proposed using an AI-based agent to prepare a real synchrotron X-ray experiment. They showed that this agent could autonomously plan actions, interpret observations and generate instrument-control commands to complete sample alignment.
"Our work demonstrates an agentic AI X-ray scientist that can autonomously align single-crystal samples at synchrotron X-ray beamlines," Zhantao Chen, first author of the paper and now an assistant professor at The University of Texas at Austin, who led this work while at SLAC National Accelerator Laboratory and Stanford University, told Phys.org.
"This agent can query experimental status, reason about what's going on and what needs to be done, and then carry out the experiment toward successful sample alignment. The original inspiration came from the fact that sample alignment is an important first step in almost every single-crystal synchrotron X-ray experiment, yet it can be tedious."
The quest to automate X-ray sample alignment
Chen and his colleagues initially set out to automate a specific process that scientists conducting synchrotron experiments often find tedious. This is the alignment of a crystal sample to enable the collection of quality X-ray data.
"We initially wanted to develop AI to automate this process," explained Chen. "As the project evolved, however, we became interested in a broader question: Can AI reason through and perform experimental tasks much like a human scientist at a real-world synchrotron beamline? These two motivations—the practical need for automation and the more ambitious scientific question—together inspired this work."
While this recent study primarily focused on the autonomous alignment of samples, the researchers wanted to use this specific task as an example of what AI agents, particularly large language models (LLMs), could achieve in experimental settings. Their demonstration shows that these agents could perform experimental tasks autonomously, adapting to unexpected or changing conditions.
"Simply put, there is a collection of experimental tools that the AI agent can use, including those needed to perform the experiment, such as reading experimental logs, taking detector images and doing motor scans," said Chen.
"AI can use the tools to understand what's going on and what needs to be done next toward the eventual goal. Think of it this way: If we want an AI agent to tighten a screw, we need to give it both a screwdriver and instructions on how to use it. The tools let the AI interact with the real world, which here is our experiment, while the prompt teaches it how to use those tools effectively."
In their paper, Chen and his colleagues suggest that this adaptability might be a key advantage of LLM-based AI agents. In changing conditions or when new information is fed to them, these models could rapidly adjust their responses to account for this additional information.
"This adaptability is especially important in real experiments, where unexpected situations are common," said Chen. "Without it, traditional automation often requires countless hand-written if-else rules to cover different scenarios."
The path toward AI-enhanced scientific facilities
The recent work by Chen and his colleagues highlights the potential of LLM-based agents for planning and preparing large-scale scientific experiments. A key innovation was the integration of the AI agent with a virtual beamline developed by the team, which allowed researchers to develop and refine the AI X-ray scientist in simulation before deploying it on a real synchrotron beamline.
"Readers might know that large-scale scientific facilities are often very oversubscribed, meaning that developing such an agentic AI workflow using real instruments from scratch would be impractical since the instruments are always very busy," said Chen.
"We developed a virtual beamline (simulator of the experimental environment) and then developed our entire agentic AI workflow there, which turned out to work at the real-world beamline with only minimal modifications to accommodate the input/output format used there. This opens many opportunities—imaging lab developers can prepare digital twins for their lab instruments, then AI can practice experimental workflows and skills extensively in simulation before being deployed on real instruments and eventually contribute to daily experimental work."
In the future, this paper could inspire the development of new AI-based platforms and tools to enable the automation of routine practices or processes carried out at large research facilities. Eventually, the team's proposed AI agent could be used to autonomously run new types of synchrotron experiments.
"We believe the workflow we demonstrated can be extended to many other experimental tasks, paving the way toward AI-assisted or even AI-driven scientific experimentation," added Chen.
"We now plan to continue pushing both the depth and the breadth of this work. On one hand, we want to develop more capable AI X-ray scientists that can help scientists make new discoveries. On the other hand, we're actively developing more general agentic AI frameworks that can be applied beyond X-ray beamlines to materials science and physics laboratories more broadly."
Written for you by our author Ingrid Fadelli, edited by Sadie Harley, and fact-checked and reviewed by Robert Egan—this article is the result of careful human work. We rely on readers like you to keep independent science journalism alive. If this reporting matters to you, please consider a donation (especially monthly). You'll get an ad-free account as a thank-you.
Publication details
Zhantao Chen et al, An agentic artificially intelligent X-ray scientist, Nature Machine Intelligence (2026). DOI: 10.1038/s42256-026-01261-5.
Who's behind this story?
Freelance journalist with BSc Psychology and MA International Journalism. Covers AI, robotics, neuroscience, and astrophysics since 2018. Full profile →
BSc Life Sciences & Ecology. Microbiology lab background with pharmaceutical news experience in oil, gas, and renewable industries. Full profile →
Bachelor's in mathematical biology, Master's in creative writing. Well-traveled with unique perspectives on science and language. Full profile →
© 2026 Science X Network
Citation: AI agent helps prepare synchrotron X-ray experimental measurements, paving the way for autonomous operation (2026, July 21) retrieved 21 July 2026 from https://phys.org/news/2026-07-ai-agent-synchrotron-ray-experimental.html
This document is subject to copyright. Apart from any fair dealing for the purpose of private study or research, no part may be reproduced without the written permission. The content is provided for information purposes only.