Session Chair: Mark Lauby – NERC

This industry session will bring together grid operators to discuss the pathway from emerging AI concepts to deployment in real-world power systems. The discussion will focus on high-value industrial use cases, practical requirements for adoption, and lessons from ongoing AI deployments. The session will also explore where Foundation Models can provide tangible advantages over application-specific AI models and conventional methods, helping identify the capabilities and evidence needed to move from research prototypes and pilots toward trustworthy, scalable, and operationally relevant solutions.

AI in Transmission Grids: From Use Cases to Business Value

Anna Wronka – PSE S.A.

The presentation will provide an overview of the most promising AI and GenAI applications for transmission system operators, including forecasting, system monitoring, decision support and asset management. It will also address organizational, regulatory and cybersecurity considerations that influence successful adoption.

AssistFlux: Toward Operational AI for Curative Topology Actions at RTE

Catalina Obando – RTE

As the power system grows more complex, with rising demand, evolving generation, and a new geography of flows, grid operation increasingly requires decision-support tools that are reliable, performant, and auditable. When an N-1 contingency creates a thermal constraint, operators must identify a curative topological action to relieve it, a demanding task under time pressure. AssistFlux, developed at RTE, aims to recommend such actions to dispatchers near real time. Trained on historical grid situations, a transformer-based model predicts the post-action flows on monitored lines for a given network state and candidate action space, enabling efficient ranking without exhaustive simulation. We present the current state of the project, our evaluation methodology, combining statistical metrics with assessment by operators, and the path toward operational validation.

Key considerations in taking Grid FMs from the lab to the control room

Herman Carstens – Elia group

Grid foundation models show great promise for a power system in transition and will find a place in the toolbox of grid operators. However, integrating new tools into critical infrastructure operations – especially ones such as foundation models – presents its own challenges. As complexity increases and reaction time decreases, our margin for error shrinks. The trustworthiness requirements that we place on our calculation tools and overall grid management systems also increase. There are also requirements on accuracy, reliability, data confidentiality, sovereignty, and their role among other specialised tools such as fast load flow solvers, forecasting tools, etc. Besides electrical and reliability engineering requirements, there are also the requirements placed on critical infrastructure IT systems. This talk will present a perspective on deploying intelligent systems at grid operators, applied to foundation models.

FM and non-FM AI at ELES

Speaker: Uroš Salobir – ELES, d.o.o.

ELES’s AI projects are closely related to the European AI-stream with use-cases specifically tailored to the hosting capacity, and management in multi-variable space. ELES also plays a vital role in the establishment of AI deployment Support Team, that will help ENTSO-E members safely and effectively deploy AI. On the non-FM front ELES tries to develop a niche application of AI on operator-training simulator tool.