Test every plan on an open model before the public pays for it.

One model for systems that act on each other

A new rail line can change where people live, and with it the demand for water and power. Trees change how hot a street gets, and the US EPA says plants have a natural cooling effect. Planned apart, each department can meet its own target while the place pays more.

Data and software are no longer the barrier

Copernicus satellite data is free under an open data policy, a worldwide community of mappers maintains OpenStreetMap, and the US EPA says its free EPANET water model is used throughout the world. The scarce part is building a model, checking it against the past and running it for a real decision. In the UN's 2026 SDG report, UN-Habitat calls for data-informed urban planning, and World Bank authors call least-cost electrification planning essential for using scarce public money well.

AI runs the scenarios, and people decide

Every project would follow the same steps.

• Open data comes first, and a public body's own records come in only where they change the answer.
• A digital twin is a computer model of a real system, updated with its data, that predicts what will happen and informs decisions, following the US National Academies' definition.
• Simulation replays the system under a change, such as a closed road or a dry summer.
• Optimization solvers search many possible plans for the best one within set limits, such as a budget.
• Forecasting gives a range for future demand, not one number.
• AI agents are software that uses a language model to set up runs, check outputs against rules and draft summaries. They do not decide.
• A named official signs off every decision, and every run is logged.

Where the method applies

• Transport covers roads, railways and bus networks.
• Energy covers electricity generation, grids and demand.
• Water covers supply, distribution, leakage and reuse.
• Fuel and logistics covers fuel use and freight.
• Land and trees covers land use, green cover and urban heat.
• Food and agriculture covers harvests, food stocks and food loss.
• Pollution covers air, water and soil.
• Waste covers garbage collection, recycling and landfill.
• Disaster resilience covers hazards and the assets exposed to them.
• Public finance covers budgets, taxes and subsidies.
• Public procurement covers competition and prices in public buying.
• Governance and public services covers how services reach people.
• Healthcare capacity covers patient demand, beds, staff and operating rooms.
• Jobs and skills covers training for the work a region is likely to have.

Public money should buy open models

We plan to publish the shared parts of this work as an open-source project, with code under the Apache-2.0 license. Any public body could then inspect, rerun and extend a model without a license fee. It would not be another simulator. It would add loaders for open datasets, scenario files that a named person approves before a run, adapters for the models on each page, back-testing checks and a record of every run.

Start with one decision, not a platform

Pick one decision that has an owner and a date, such as which water pipes to renew first. We would build a first model from open data and back-test it on a past year to see whether it reproduces what happened. You get a scorecard of how well it matched, what it recommends and which data would change the answer. Then you decide whether to continue. Companies start the same way, through a Proof Sprint. Procurement routes differ by country, and this page is not legal advice.