The Agentic Revolution: Lightning Sessions on Agentic Workflows
Talk details
Date
April 15, 2026
Time
11:00am PST / 2:00pm EST / 20:00 CEST
Overview
The (R)evolution of Scientific Workflows in the Agentic AI era: Towards Autonomous Science
Woong Shin (Oak Ridge National Laboratory)
Modern scientific discovery increasingly requires coordinating distributed facilities and heterogeneous resources, forcing researchers to act as manual workflow coordinators rather than scientists. Advances in AI leading to AI agents show exciting new opportunities that can accelerate scientific discovery by providing intelligence as a component in the ecosystem. However, it is unclear how this new capability would materialize and integrate in the real world. To address this, we propose a conceptual framework where workflows evolve along two dimensions, intelligence (from static to intelligent) and composition (from single to swarm), to chart an evolutionary path from current workflow management systems to fully autonomous, distributed scientific laboratories. By embedding reasoning and adaptation into workflows, these labs have the potential to accelerate discovery by factors of 10 to 100, transforming exploratory science into a continuous, machine-augmented process.
Title: TBD
Jan Janssen (Max Planck Institute for Sustainable Materials)
Agentic AI for experiments and data analyses at the APS
Du Ming, Xiangyu Yin (Argone National Laboratory)
We will introduce the current efforts of using vision language model (VLM) agents for automated and low-barrier beamline operations and data processing algorithm research. We first present Experiment Automation Agents (EAA), an agent capable of controlling beamline instruments and making decisions based on image semantics, with a few cases demonstrating how it automates and democratizes experimental operations at APS beamlines. We will then introduce works on agentic data processing, which includes PEAR, a domain-expert system that tunes ptychographic reconstruction hyperparameters using reconstructed image as feedback, and Pty-Chi-Evolve, an auto-research agent that autonomously searches for regularization operators during iterative reconstructions to enhance result quality.
Woong Shin (Oak Ridge National Laboratory)
Modern scientific discovery increasingly requires coordinating distributed facilities and heterogeneous resources, forcing researchers to act as manual workflow coordinators rather than scientists. Advances in AI leading to AI agents show exciting new opportunities that can accelerate scientific discovery by providing intelligence as a component in the ecosystem. However, it is unclear how this new capability would materialize and integrate in the real world. To address this, we propose a conceptual framework where workflows evolve along two dimensions, intelligence (from static to intelligent) and composition (from single to swarm), to chart an evolutionary path from current workflow management systems to fully autonomous, distributed scientific laboratories. By embedding reasoning and adaptation into workflows, these labs have the potential to accelerate discovery by factors of 10 to 100, transforming exploratory science into a continuous, machine-augmented process.
Title: TBD
Jan Janssen (Max Planck Institute for Sustainable Materials)
Agentic AI for experiments and data analyses at the APS
Du Ming, Xiangyu Yin (Argone National Laboratory)
We will introduce the current efforts of using vision language model (VLM) agents for automated and low-barrier beamline operations and data processing algorithm research. We first present Experiment Automation Agents (EAA), an agent capable of controlling beamline instruments and making decisions based on image semantics, with a few cases demonstrating how it automates and democratizes experimental operations at APS beamlines. We will then introduce works on agentic data processing, which includes PEAR, a domain-expert system that tunes ptychographic reconstruction hyperparameters using reconstructed image as feedback, and Pty-Chi-Evolve, an auto-research agent that autonomously searches for regularization operators during iterative reconstructions to enhance result quality.