About LLM and Agentic AI in AEcroscoPy#
\(_{Yongtao}\) \(_{Liu,}\) \(_{youngtaoliu@gmail.com}\)
\(_{March}\) \(_{2026}\)
Large Language Models (LLMs) and agentic AI are transforming how researchers interact with scientific instruments. In AEcroscoPy, we integrate LLM-powered agentic AI to allow users to control the microscope and design experiment workflows using natural language — no Python programming experience required.
What is Agentic AI in AEcroscoPy?#
An AI agent in AEcroscoPy acts as an intelligent interface between the researcher and the microscope. Instead of writing Python scripts manually, users can describe what they want in plain English, and the agent translates those instructions into AEcroscoPy commands and executes them automatically.
For example, a user can say:
“Perform a raster scan with a 5 µm scan size, then apply a -5 V pulse at the center and image the result.”
The agent interprets the intent, constructs the corresponding AEcroscoPy workflow, and runs it — enabling full experiment control through conversation.
Key Capabilities#
Natural language microscope control — Directly operate the AFM (tip movement, scanning, pulsing) by describing actions in plain English.
Natural language workflow design — Compose multi-step automated and autonomous experiment workflows without writing code.
Accessibility — Lowers the barrier to entry for researchers who are not familiar with Python, making automated and autonomous microscopy accessible to a broader scientific community.
Interactive experiment guidance — The agent can suggest next steps, explain parameters, and help users make decisions during an ongoing experiment.
Why This Matters#
Automated and autonomous microscopy workflows have historically required programming expertise to set up and operate. By introducing agentic AI, AEcroscoPy enables domain scientists — materials researchers, physicists, chemists — to focus on the science rather than the code, while still benefiting from the full power of automated and AI-driven experimentation.