AI for home care providers.

A home care visit used to end when the support worker left the house. Under Support at Home it ends when the visit has been noted, itemised against the right service and funding source, priced, claimed, reconciled against the participant's quarterly budget and shown on a monthly statement the participant can understand. The care has not changed. The paperwork around it has.

That paperwork is exactly the kind of work AI agents now do well. It is also work done on health, financial and personal information about older people living alone, some of whom will never check a statement. Both facts decide how a home care provider should approach AI, and who it should trust to build it.

What changed for home care providers.

Five changes since late 2025 shape where AI is worth using, and where it has to be careful.

• Support at Home replaced Home Care Packages. From 1 November 2025, Support at Home replaced the Home Care Packages Program and Short-Term Restorative Care. Participants now sit in one of eight ongoing funding classifications, with quarterly budgets, plus short-term pathways for restorative care, end-of-life care and assistive technology and home modifications.
• Claims are itemised and paid in arrears. Providers claim through Services Australia after the service is delivered, itemised against the participant and the funding source, and must finalise claims within 60 days after the end of the quarter. Care management is funded by a 10 percent deduction from each quarterly budget and claimed in 15-minute increments against a named participant and day.
• A new Act and stronger standards. The Aged Care Act 2024 commenced on 1 November 2025, with registration into one or more of six categories and the strengthened Aged Care Quality Standards. Monthly statements to participants are a legislated obligation, not a courtesy.
• The rules keep moving. From 1 October 2026, personal care moves from the independence contribution category to clinical supports, so participants stop paying contributions for it. Subsidy rates rise on the same day. The Commonwealth Home Support Program has been extended to 30 June 2029. Any system that prices or claims has to absorb changes like these as a rule update, not a rebuild.
• Privacy obligations apply at any size. The Privacy Act's small business exemption does not cover organisations that provide a health service and hold health information, which includes most home care providers. From 10 December 2026, any organisation that uses a computer program to make decisions that could significantly affect a person's rights or interests must say so in its privacy policy.

Where AI agents earn their keep in a home care business.

The strongest uses share a shape: high volume, a published rule to check against, and a person who signs off. Five stand out.

• Visit notes. A support worker speaks a short summary in the car after a visit, and an agent drafts the note in your format, tagged to the participant's care plan and goals. The worker reviews and signs it. Notes get written on the day, which is also the day the claim evidence is freshest.
• Claims itemisation and reconciliation. An agent checks every delivered service against the service list, the participant's classification and quarterly budget, the agreed price and the contribution category before the claim goes in, and flags the mismatches. This is where cash leaks: rejected claims, services claimed against the wrong funding source, and care management time never recorded.
• Monthly statements. Drafting an itemised statement a participant or their family can actually read, from the claims data, is a well-bounded job for an agent. A person checks it before it goes out, particularly in the first months after a rule change like the 1 October contribution move.
• Rostering and travel. Matching worker skills, screening, participant preferences, continuity of carer and drive time across a region is a constraint problem, and agents are good at proposing a roster. A coordinator approves it, which also keeps it on the right side of the new automated-decision disclosure.
• Incident reports. Serious incidents in home services fall under the Serious Incident Response Scheme. Priority 1 incidents must be notified to the Aged Care Quality and Safety Commission within 24 hours of becoming aware, and Priority 2 incidents within 30 days. An agent that drafts the notification from the worker's account and tracks both deadlines turns a scramble into a checklist. A person still decides whether it is Priority 1 or Priority 2.

Where AI must not decide.

Some decisions stay with people, whatever the tool can do. Whether a participant is still safe to live at home. What changes in a care plan. Whether a clinical observation needs a nurse today. Whether a participant is in financial hardship. How an incident is classified. Whether a worker is the right person to send to a particular house.

AI can prepare the information for every one of those decisions. It should make none of them. An agent that is right 97 percent of the time is a gift on claims reconciliation and a serious failure on the safety of someone living alone. We have written more generally about where AI agents do not work yet, and home care sits at the careful end of that line.

If you deliver NDIS supports as well.

Many providers run home care and NDIS supports from the same workers, the same rosters and often the same spreadsheet. The work looks alike, but the rules around it do not. Aged care answers to the Aged Care Quality and Safety Commission and claims through Services Australia. NDIS supports answer to the NDIS Quality and Safeguards Commission, and agency-managed claims are paid by the National Disability Insurance Agency (NDIA) against the participant's plan. Two regulators, two payers, two sets of price rules, one workforce.

That is a data problem before it is an AI problem. An agent that reconciles claims needs to know which scheme a shift belongs to before it can check anything. The NDIS side is covered in our piece on AI for NDIS providers, and the NDIS provider case study shows what joining the systems first looks like in practice.

What to demand from the firm that builds it.

The model is rarely the hard part. The hard part is the data underneath it and the controls around it. Ask any AI partner these five questions before you sign.

• Where does participant data go? Every prompt, embedding, log and backup should stay in Australian regions, and the partner should be able to show you, not just tell you. Our piece on where your AI data can live covers the places it usually leaks.
• How does it handle a rule change? Ask how the system handles the 1 October 2026 contribution move. If it needs a developer, the next change will too.
• Who reviews what the agent writes? Every note, claim and statement should have a named human approver, and the system should log who approved what.
• Can it survive an audit? Ask how the partner would evidence the system against the strengthened Quality Standards. If the answer is a slide, keep looking.
• Will they start with an assessment? A partner who quotes a build before looking at your data is guessing. An AI readiness assessment tells you which of the uses above your records can support today.

Why RUBIX.

RUBIX has been an Australian data and AI consultancy since 2011. That is 15 years, 450+ projects and 115+ customers building governed data platforms for organisations where the data is sensitive and the regulator is watching. We have not yet published a Support at Home engagement, and we will not pretend otherwise. What we bring to a home care provider is the work closest to it.

• A care provider with the same shape of problem. For a leading NDIS provider we joined six disconnected systems, including rostering, finance and payroll, into one governed data foundation, a live role-based dashboard and an AI automation roadmap worth ~$645K a year, in under eight weeks. Capped prices, an award-wage workforce and rising compliance are the same pressures a home care provider carries.
• Sensitive health data is not new to us. Our work with Medibank covered AI readiness, responsible-AI guardrails and personalisation on a privacy-respecting data foundation.
• Data first, then agents. Most provider AI projects stall because notes live in one system, rosters in another and claims in a spreadsheet. We build the data foundation that joins them, then put AI agents to work on it.
• Australian, onshore and accountable. Our people are in Australia, data stays in Australian regions by default, and governance is designed in from the first week. Our AI consulting in Australia is built around those rules.
• Senior people do the work. Our forward deployed engineers sit with your team, so the people who scoped the work are the people who deliver it.

If you run a home care business and want to know what AI can safely take off your team, start with a conversation about your records, not a demo of a tool.

Frequently asked questions.

General information only, not legal, financial or regulatory advice. Current as at September 2026. Check the Support at Home program manual for the rules that apply to your organisation.