NewsBB-0175

Agentic AI operates an atomic force microscope in live experiments

Zahra Ayar, Georg Fantner and colleagues published a paper on three MCP-connected agents that run an atomic force microscope. One converts plain-language requests to checked commands, one judges image quality and retunes imaging, and one diagnoses artifacts. In live experiments, the paper says, the tuning agent did as well as expert operators. No code is linked.

Open original source arxiv.org · Zahra Ayar, Marcos Penedo, Georg E. Fantner et al.

Claimed
Why this evidence label?

Abstract reports live experiments on different samples; no code or data release linked.

How we check entries
Preview image for Agentic AI operates an atomic force microscope in live experiments
The source's own preview image, from arxiv.org.

Briefing

The agents are named AFM Messenger, AFM Pilot and AFM Doctor, and all three connect through MCP. An ambiguity-check layer screens Messenger's commands before they execute. Doctor applies only post-processing steps from a pre-approved set.

The authors say this guarded layer, more than the model's capability, brought wrong-command execution to zero in their benchmarks. In the live experiments, Pilot showed no significant difference from expert operators in image quality, number of iterations and tuning time. The paper is arXiv 2608.26198, submitted on 25 August 2026 by six authors, with Georg E. Fantner listed last.

Tags

Interface

Report a problem with this entry