Australia's largest health workforce is unpaid, uncounted and stretched close to breaking. Three million people, one in eight of us, provide unpaid care to a parent, a partner, a child or a neighbour, and 1.2 million of them are the primary carer who holds the whole arrangement together. If you had to replace that labour at market rates it would cost around $78 billion a year, according to Deloitte Access Economics' 2020 accounting for Carers Australia, which would make unpaid carers, were they ever to bill, one of the largest single items in the national health budget. We call it free only because no invoice is raised.
The cost does not vanish for want of an invoice; it is displaced, which is a different thing. It lands on the carer's own body first, with over half of primary carers reporting high or very high psychological distress in the 2024 National Carer Survey, and on their retirement next, with a primary carer forgoing roughly $567,000 in lifetime earnings and superannuation by the age of 67. Then on their employer, who loses a trained worker mid-career, and on the hospital, which readmits the person whose carer has buckled. Marilyn Waring made the point forty years ago in Counting for Nothing: a system that measures only what carries a price will, with perfect logic, grind down everything that doesn't. The carer holds up everyone downstream of them, and the national accounts cannot see them at all.
What we offer in return is thin and, worse, late. The Commonwealth's Carer Gateway has provided coaching, counselling and respite since 2019, which is real and welcome, but it is not wired to live, local data, so the help tends to arrive after the fall, the discharge or the diagnosis rather than before. Hilary Cottam, whose book Radical Help rebuilt frayed British welfare around relationships rather than transactions, calls this the central design flaw of twentieth-century welfare: it waits for the crisis it was meant to prevent, then manages the wreckage. A carer does not need a brochure about a waiting list. They need the waiting list jumped before the crisis hits.
Which points at the reframe. Carer strain is not a soft welfare problem to be met with sympathy and a pamphlet; it is a population-health problem with a measurable cost and, crucially, a payer who already bears that cost whether they act or not. The employer who keeps a carer in work spends far less than the employer who recruits and retrains their replacement, so the business case for stepping in early is already written, sitting unread. This is the instinct behind Juggle, an idea my colleague Annemarie first put on the table, and its logic is to give the carer the one thing the system withholds: help that moves first, with money already behind it.
It has three parts.
- An agent that does the juggling. A navigator wired to live, verified information on every local service, entitlement and waiting list, that sees the predictable shocks coming (the first fall, the Friday-afternoon discharge, the new diagnosis) and, within limits each family sets, books and pays for the respite or the home modification itself. The AI does the forty phone calls the carer cannot make from the hospital corridor.
- Money already set aside. Care shocks are predictable in kind and unpredictable in timing, which is the textbook definition of an insurable risk. A care buffer held in super or a bank account, topped up by employer and member and released the moment a care event is verified, paired with a carer-risk insurance product for emergencies and respite, modelled on the income-protection and critical-illness cover the industry already sells.
- A layer that keeps the peace with the services already there. A view, shared with councils, Primary Health Networks and charities, of where strain is building, so automated demand steers toward slack rather than swamping the one overstretched respite centre down the road. Pointed carelessly, an agent with a budget could flatten the very services it depends on.
Who pays is the question every one of these ideas lives or dies on, and here the answer is unusually clean. Employers pay first, as a benefit, because retention is cheaper than replacement. Super funds and banks run the care buffer and earn on the balances they hold. The carer-risk cover is bought by the person who might one day need it, the way they already buy cover against a disability. And the navigator plugs into Carer Gateway, which the Commonwealth already funds, so the public purse is asked to connect a service, not to stand up a new one. The care economist Nancy Folbre would recognise the shape instantly: not a subsidy for a market failure, but an instrument that gives care, the great unpriced input the national accounts still treat as free, an owner with a reason to invest in it.
None of it needs a new institution. The missing pieces are a capped compassionate-release rule so the buffer can be drawn when a care event is verified, a carer-risk product an insurer is willing to write, and one large employer, one super fund and one Primary Health Network prepared to run the first version between them. The harder question is the one every agentic system now raises, which is how much we trust software to spend money and book care on behalf of someone vulnerable, and the answer is that the family-set limits and the human in the loop are not a feature to add later but the thing you build first. Get that wrong and the tool becomes another source of the stress it was meant to lift.
There is a selfish case too, which is probably the one that moves things. Most of us will spend years as a carer and years being cared for, often without much warning, and the arrangement we build for the three million people doing it today is the one we are building for ourselves.