For Yu Jiang, the technical building blocks for an agentic grid increasingly exist. The harder question is whether grid organizations have the data, governance, control boundaries, and accountability needed to run them.
In his “Running the Agentic Grid: From Open Models to Operating Model” keynote at LF Energy Summit Europe 2026 on September 15 in Berlin, Yu Jiang, Managing Director and Data & AI Practice Lead at Sia, explored where AI agents could fit into grid operations and why the operating model around them matters as much as the technology itself.
He described how AI agents can help prepare a growing number of operational decisions while humans retain responsibility for outcomes.
The technical building blocks are increasingly available
Jiang opened with a vision of a future control room. During the night, one AI agent updates a forecast after wind generation comes in below plan. Another reoptimizes the grid around congestion points. A third prepares a plan for the operator to review and approve when critical decisions are involved.
His point was that the components required for that kind of workflow are not science fiction. He told the audience that the necessary building blocks already exist within the community under open licenses.
The urgency comes from the mismatch between how quickly electricity demand is changing and how quickly physical grid infrastructure can be expanded.
The official session description points to €584 billion in grid investment needs by 2030 and €1.2 trillion by 2040, while permitting can still take much of a decade. It also cites roughly 1,700 GW of renewable generation waiting in connection queues and €2.8 billion spent by Germany on congestion management in 2024.
For Jiang, this makes better use of existing capacity increasingly important. Building more infrastructure remains necessary, but construction alone cannot address every constraint at the speed required.
Using more grid capacity creates more operational decisions
Making more efficient use of existing infrastructure introduces a different kind of challenge.
Jiang contrasted putting a cable in the ground, which he said represents a decision that lasts for about 40 years, with flexibility contracts that generate repeated operational decisions.
Flexible grid operation can require decisions across contracts, regions, and substations every hour. That growing decision space is difficult for a small number of control-room operators to manage manually.
Jiang positioned AI agents as a way to help operators work through a volume of possible actions that becomes difficult to handle manually.
LF Energy projects provide open building blocks
Jiang connected that operating vision directly to the LF Energy community.
He pointed to the foundation’s open source project portfolio as a set of technical building blocks that can support different parts of an agentic grid workflow, from sensing and representing the grid to forecasting, simulation, scenario testing, decision-making, and action.
For forecasting, he specifically named OpenSTEF.
His broader point was that many of the underlying capabilities already exist across the open source portfolio.
That leads to the question at the center of Jiang’s keynote: if the tools and models increasingly exist, why are these workflows not already commonplace in control rooms?
The gap is operational readiness
Jiang’s answer was readiness.
He argued that the remaining challenge is not a lack of talent, budget, or models. Organizations need an operating model that allows AI agents to participate in grid operations in a controlled and accountable way.
He proposed four questions for organizations to ask:
- Data foundations: Can an agent see a reliable picture of the grid, supported by sufficient data quality?
- Trust and governance: What checks must an agent pass before a recommendation reaches an operator?
- Control: What limits can an agent operate within, and who enforces those limits?
- Accountability: Who approves and owns the outcome?
On the final question, Jiang was explicit that accountability remains with people, not the AI agent.
“In order to scale the autonomy through AI agents, we need to redesign how we operate.”
That shift from model capability to organizational readiness is the core of the operating-model argument in his keynote.
AI autonomy is not an on-or-off decision
Jiang also cautioned against treating autonomy as a binary choice.
He described four levels:
- Level one: Agents advise, providing recommendations to operators.
- Level two: Agents prepare and assemble actions, while humans approve each one.
- Level three: Agents act within defined limits and escalate cases that fall outside those limits.
- Level four: Agents execute actions, with humans auditing afterward.
Moving between those levels requires gates around data, trust, control, and accountability.
Just as importantly, Jiang said the objective should not automatically be to reach the highest possible level of autonomy. Organizations should operate at the highest level they can prove they are ready for.
From open models to an operating model
Jiang closed by returning to the distinction in the keynote title.
He expressed confidence in the technical foundations being developed across the open source community. But making the imagined 7:30 a.m. control-room scenario real requires more than those foundations.
Organizations also need reliable data, governance, trust, clearly defined ownership, and human accountability.
For Jiang, that is the remaining gap between having the technology for an agentic grid and being ready to operate one.
Watch the keynote
Watch Yu Jiang’s full keynote, “Running the Agentic Grid: From Open Models to Operating Model.”
Frequently asked questions
What is an agentic grid?
In the keynote, Jiang described a future control-room workflow in which AI agents update forecasts, reoptimize around congestion points, and prepare plans for operators to review and approve. He argued that the technical components for this kind of workflow increasingly exist, while operational readiness remains the larger challenge.
What does operational readiness for AI in grid operations require?
Jiang identified four areas: reliable data foundations, trust and governance, defined control limits, and human accountability for outcomes.
How can open source support agentic grid operations?
Jiang described LF Energy’s open source portfolio as providing building blocks across sensing, grid representation, forecasting, simulation, scenario testing, decision-making, and action. He specifically named OpenSTEF as an example for forecasting.
About OpenSTEF
OpenSTEF (Open Short-Term Energy Forecasting) is an open source machine learning pipeline for short-term energy load forecasting, developed by Alliander and hosted at LF Energy. It produces probabilistic 48-hour-ahead forecasts at individual substation and feeder locations for operational congestion management. Alliander uses OpenSTEF in production at hundreds of grid locations; Sigholm uses it as the forecasting engine for its Aurora platform, which optimizes approximately 40% of Sweden’s district heating production. RTE also contributes to the project.
About LF Energy
LF Energy is the Linux Foundation’s neutral home to build the digital foundation for energy together. We host more than three dozen open source projects used to plan, operate, and modernize energy systems. The portfolio spans from transmission system modeling to substation virtualization, smart meter integration to EV charging. We provide the trust layer: neutral governance, legal frameworks, security practices, and community of practitioners that turn open source software, standards, and data into infrastructure the industry can depend on for decades. Members include utilities, energy companies, vendors, researchers, and technology providers, aligned to deliver affordable, reliable, safe, and clean energy. For more information, please visit lfenergy.org.
About The Linux Foundation
The Linux Foundation is the world’s leading home for collaboration on open source software, hardware, standards, and data. Linux Foundation projects are critical to the world’s infrastructure including Linux, Kubernetes, Node.js, ONAP, PyTorch, RISC-V, SPDX, OpenChain, and more. The Linux Foundation focuses on leveraging best practices and addressing the needs of contributors, users, and solution providers to create sustainable models for open collaboration. For more information, please visit linuxfoundation.org.
AI Disclosure
This post used artificial intelligence tools for research, structural assistance, or grammatical refinement. The final content was reviewed, edited, and validated by human contributors to LF Energy to ensure accuracy and alignment with our community standards. We remain committed to transparency in the use of generative technologies within the open source ecosystem.
Last updated: October 9, 2026