LLMs and Agentic AI Systems for Smart Grids: A Tutorial on Architectures and Applications

Source

• Source name: arXiv
• Source kind: paper
• Authors / organization: Daniela Rojas, Abdulwahab Albassam, Aidan G. Leung, Jett Ngo, Ryan Luo, Peter R. Quawas, Junpyung Kim, Kangkai Liang, Mansi Nanavati, Jonathan Mai, Meng-Chi Tsai, Yun-Tong Tsai, Yize Chen, Yuanyuan Shi
• Venue: arXiv: eess.SY
• Published: 2026-07-20
• Collected: 2026-07-21
• URL: http://arxiv.org/abs/2607.18147v1
• PDF: https://arxiv.org/pdf/2607.18147v1
• DOI: UNKNOWN

Plain-Language Brief

The paper discusses how large language models (LLMs) and agentic AI systems are being used in smart grids for tasks such as forecasting, optimization, and control. The authors present a design principle that ensures numerical results come from trusted tools before reporting them to avoid physically infeasible outputs. Four case studies are provided to demonstrate the effectiveness of this approach, showing significant improvements in areas like wind power forecasting and EV charging scheduling.

What the Document Says

• Large language models (LLMs) and agentic AI systems are being used in smart grids for tasks such as forecasting, optimization, and control.
• The paper presents a design principle that ensures numerical results come from trusted tools before reporting to avoid physically infeasible outputs.
• Four case studies are provided: wind power forecasting, EV charging scheduling, power flow analysis, and contingency diagnosis.
• The LLM-only baselines are compared against solver-grounded counterparts for the same data and metrics. The agentic system shows significant improvements in areas like unmet energy reduction (7.5-9.5x) and contingency case repair (17/39).
• A four-group evaluation framework is proposed: task utility, solver-grounded correctness, faithfulness and safe failure, and cost and latency.

Key Concepts Mentioned

• Large Language Models (LLMs)
• Agentic AI Systems
• Smart Grids
• Forecasting
• Optimization
• Control
• Design Principle
• Case Studies
• Energy Management
• Evaluation Framework

Claims Made by the Document

• The design principle reduces physically infeasible outputs.
• Solver-grounded approaches improve upon LLM-only baselines in various areas.
• A consistent division of labor emerges between the agentic system and trusted tools.

Evidence Used

• The paper demonstrates the effectiveness of the design principle through four case studies.
• Comparisons between LLM-only baselines and solver-grounded counterparts are provided for each case study.

Methods / Approach

UNKNOWN

Limitations Stated by the Authors

UNKNOWN

Limitations Noted by the Agent

Agent interpretation — not from the source.

UNKNOWN

Open Questions

UNKNOWN

Notes for Future Q&A

May include interpretation. Labeled where it does.

UNKNOWN

Raw Abstract / Excerpt

Large language models (LLMs) and agentic AI systems have evolved from natural language tasks to using external tools to plan, retrieve, and act in technical domains. In smart grids, recent work applies agentic schemes to forecasting, optimization, and control, wrapping trusted solvers behind language interfaces and orchestrating multi-step workflows. The literature lacks a unified approach to designing and evaluating such systems. LLMs can produce numerically plausible yet physically infeasible outputs, evaluation protocols vary across tasks, and the boundary between what the model should and should not compute is implicit. This paper presents a solver-grounded design principle: a numerical result is reported only when it originates from a trusted tool and passes explicit verification. We review the building blocks of LLM and agentic AI systems for power systems: prompting strategies and agentic architectures. We instantiate the principle in four case studies: wind power forecasting, EV charging scheduling, power flow analysis, and contingency diagnosis, each comparing an LLM-only baseline against its solver-grounded counterpart on identical data and metrics. EVAgent reproduces the CVXPY optimum while reducing LLM-only unmet energy by 7.5-9.5x, and GridDebugAgent repairs 17/39 contingency cases while reducing total violations by 52.3%. We propose a four-group evaluation framework spanning task utility, solver-grounded correctness, faithfulness and safe failure, and cost and latency. A consistent division of labor emerges: the agentic system reliably orchestrates, retrieves, and explains, while trusted tools compute and a verification gate decides what is reported.