AI model helps network planners answer questions – fast 

AgenticGrid is a new modelling platform for Australia’s electricity networks, built around AI agents rather than a conventional modelling tool with a chatbot added. 

Network planners face a growing backlog of questions their existing tools were not built to answer at speed — connection assessments, hosting studies, community and big battery cases, and the cost-benefit tests behind major investment decisions. Traditional models take weeks to build and days to run, so many scenarios never get properly assessed. 

The AI agents decide what to build and run and call conventional tools to do it: power flow is solved by pandapower, an open-source solver validated against IEEE test systems. Every action is a logged, permissioned tool call, and results cite the data and assumptions behind them, so planners can see how an answer was reached. 

The platform was tested on the Pinnacles Place area of Broken Hill, in Essential Energy’s network: 9,556 addresses, 300 substations and six 11 kV feeders. Essential Energy supplied real network topology and a year of transformer load data, which allowed the model to be validated against measured demand. During the test, AgenticGrid: 

• built a bottom-up model of the area in around two hours, against an industry benchmark of roughly three weeks for conventional practice.
• ran twelve headline demand-management scenarios, decomposed into around 60 seasonal and probabilistic sub-scenarios, in about half an hour.
• produced an AER-aligned evidence pack that planners can validate against their own systems. 

The team’s next target is automating regulatory report production. RIT-D submissions are highly structured, and the cost-benefit module already produces output in that form — but generating and validating a full submission end to end is a goal for the next phase. 

Since the test, the platform has delivered live results on a study sizing solar-plus-battery microgrids for remote NSW communities. The team have presented to Ausgrid, Energy Queensland, Essential Energy and Transgrid. 

AgenticGrid was developed by Monash University with Agentic Grid, Transgrid and Essential Energy. 

For a demonstration or an introduction to members of the project team, contact Annie Carapetian.

Take a look at the layout of the platform or find out more on the project page.