Federal, state and local government linked under one standard and one national AI — linked, not merged, with each level keeping custody of its own data.
The People’s Portal covers all three levels of government. This is how — and why it is a linking standard rather than one national database. The cascade — the Commonwealth publishing by direction, each state and territory tapping into the same format, and every council tapping in through its own state — is drawn on the DAA home page.
Australians deal with three governments and are made to learn the boundary between them before they are allowed to ask a question. Which level runs the hospital, which one owns the road, which one issued the licence, which one holds the record that is wrong. A person moving house tells three levels of government separately, in three different formats, and is treated as a new customer by each. A small business holds a federal ABN, a state licence and a council permit, and reports the same facts three times a year to three systems that cannot see each other.
Meanwhile the numbers that matter most are held at the levels that publish them least. Prisons, police, courts, hospitals, schools, land and planning are state responsibilities. Rates, development approvals, waste and local roads are council responsibilities. A national dataset that stops at the Commonwealth’s own agencies is a dataset that cannot answer the questions people actually have.
Linked, not merged. Each level keeps its own system and stays the custodian of its own data. The Commonwealth cannot lawfully seize a council’s records or a state’s health data, and pooling twenty-seven million Australians’ records into a single file would be a privacy failure waiting to happen. What links them is a standard: one verified identity, one publishing format and cadence, one open API, with records linked to the person rather than piled into a central database. Each level remains accountable for the accuracy of what it publishes, because it never stops owning it.
Offer, not mandate. A federal government can direct its own agencies and nothing more, so the twelve-month commitment applies to the federal core. States and councils are given the same platform free and open source, with adoption funded — and for a council of a few thousand people, that is a system it could never build for itself arriving at no cost. The incentive is the product, not a threat.
Who drives it. The Commonwealth builds it, funds it and runs the federal core — it is the only level that can move at national scale and the only one that can direct its own agencies from day one. But ownership of the thing itself should not sit in Canberra alone. The governing body for the standard is offered as a joint Commonwealth–state authority: the DAA builds and operates, the states and territories sit on the board that sets the standard, and local government is represented through the state members. A state signing up is joining a system it helps govern, not surrendering its data to the federal government — and that distinction is the difference between a platform that gets adopted and one that gets fought in the Federal Court for a decade.
AusLLM does the joining. The reason this is now possible when it was not a decade ago is the AI. A council running a twenty-year-old rates system does not have to replace it, re-key its records, or hire developers it cannot afford. AusLLM reads what that system already produces and maps it to the national standard — formats reconciled, fields matched, terminology translated, gaps flagged for a human to confirm. The integration work that used to be a multi-year IT project per agency becomes a mapping exercise the national AI performs once and maintains. That is what makes three levels of government, hundreds of agencies and five hundred and thirty-seven councils a solvable problem rather than a permanent excuse.
The fragmentation does not stop at the three levels. Inside each of them sit hundreds of agencies, each on a stack of its own. Defence and Veterans’ Affairs. Nine police forces that cannot read each other’s records. Ambulance, fire and emergency services organised state by state. Every hospital network on its own patient system, so a record does not follow a patient across a border. Courts, corrections and probation separated inside the same state. Every school system, every land titles office, every water authority, and 537 councils each running their own rates, planning and works software. Then the back office multiplied hundreds of times over — payroll, procurement, asset registers, HR, rostering, fleet, mapping, case management — bought separately, licensed separately, maintained separately, integrated with nothing.
Every one of those systems was rational when it was built, by an organisation that had no choice but to solve its own problem, at a time when the tools to join them did not exist. The failure is not that they exist. It is that nothing has changed since — and that the resulting waste is duplication rather than corruption. The same address captured hundreds of times. The same eligibility check rebuilt in every agency. The same dataset re-collected because two departments cannot read each other’s format. Three levels of government arguing over numbers they each calculated differently.
Joining these systems used to mean replacing them. That is a decade-long, billion-dollar programme per agency, and it is why every attempt failed and why governments stopped promising it. The national AI removes the requirement. Each system stays where it is and keeps running; AusLLM reads what it already produces and maps it into the single format — fields matched, terminology reconciled, gaps flagged for a human to confirm. Integration stops being a rebuild and becomes a translation layer that is written once and maintained.
So the commitment is deliberately narrower than the ones that have failed before. Not replacing government IT: joining what already exists, keeping it running, and ending the practice of paying hundreds of times over for the same work. And it is measurable — cost per transaction, forms eliminated, duplicate collections removed, time to decision, payment times, and the falling count of separate systems, all published on the Data & Metrics door. The DAA is the one department whose own performance is visible on the product it builds.
The Commonwealth gets a federal core it can direct into existence immediately, and national data that is finally complete rather than truncated at the border of its own responsibilities.
The states and territories get a platform they did not have to build, a seat on the body that governs the standard, and the ability to compare their own performance against every other jurisdiction on identical measures — which is uncomfortable for whoever is behind and useful for everyone.
Councils get the largest benefit for the smallest effort. There are 537 councils in Australia — 137 in Western Australia alone — and a shire of four thousand people cannot build a modern digital service and never will. Under this policy it receives one — free, open source, maintained, with the AI doing the integration work — and its residents get the same standard of service as someone in a capital city.
The citizen gets the only outcome that matters: one login, one search, one view, and no requirement to understand the federation in order to use it.
And it is not only for citizens. Business — from a sole trader to a miner — deals with government more than anyone: registrations, licences, permits, lodgements, tenders and contracts. Government itself is a user, since most of the duplication is agencies unable to read each other. And researchers, journalists and builders work on the same material through the open API, which is what makes the transparency verifiable rather than merely claimed.
It is not a national identity database. It is not the Commonwealth taking custody of state health records or council rate rolls. It is not a mandate, and it does not depend on constitutional change or on winning a fight with the states. The standard is published, the platform is given away, the AI does the hard part, and the incentive to join is that the thing works.