01 / Foundations
Understand agents, tools and power-system evidence
Build a common language for AI agents and workflows. Explore utility developments, power-system research, LangGraph and DAGs, with full source credits.
Read the full source guide →Agentic AI · Power systems · Operator decisions
Explore the concepts, follow power-system evidence and understand three operator cases. Try the optional notebooks afterward.
Choose a tutorialPresenters: Alaa Selim, MIT/Dartmouth · Mohamed Chehada, UT Austin.
No installation or model account is needed to view the presentations.
Start with the foundations, then work through the three operator cases.
01 / Foundations
Build a common language for AI agents and workflows. Explore utility developments, power-system research, LangGraph and DAGs, with full source credits.
Read the full source guide →
02 / Operator case tutorial
Follow three operator problems, the agent’s requests and the evidence returned by numerical tools. Discuss what changes when a limit or assumption changes. The slides use prepared results and need no software setup.
Case evidence and optional practice →Optional practice guide · six-page PDF ↓Use the arrow keys to move through a presentation. Each slide has its own shareable address, and the source links remain available online.
The slides need no software. Classroom Labs provides the three numerical exercises; the other notebooks are separate activities.
Run voltage L2–L5, restoration L6–L7 and cyber L8–L9. Change a parameter and compare results. L10B replays saved Llama choices and exact reasons; choose voltage, restoration or cyber.
Open in Google Colab View notebook on GitHub →Download notebookA separate graph example for understanding state, tool routing, branching and execution.
Open in Google Colab View notebook on GitHub →Download notebookInspect actual saved replies, returned tool evidence and the boundaries of the timing measurements.
Open in Google Colab View notebook on GitHub →Download notebookRUN_OPENDSS and RUN_LOCAL_LLAMA set to False.Classroom Labs performs fresh numerical calculations. L10B displays saved Llama replies and saved tool results, with no new model call or electrical solve. The default route needs no API key, language-model download or GPU.
Read the detailed Colab guide →Running the notebooks requires their supporting folders and the supplied setup instructions. Viewing saved outputs makes no new model call. Optional live experiments require their own tools and models.
Download the slides and lab materials, and read the original sources.