Insights · AI Consulting
AI consulting for NDIS and home care providers.
What a good AI consulting engagement looks like for a disability or home care provider: what to settle before you sign, what the first eight weeks should produce, which agents come next, and what you should own at the end.
Most NDIS and home care providers who ask about AI already know what they want it to do. Write the progress notes. Check the claims. Fill the roster. Pull the audit pack together. Our pieces on AI for NDIS providers and AI for home care providers go through those jobs one by one.
This piece is about the part that comes before any of them: hiring someone to build it. An AI consulting engagement for a care provider is not the same as one for a bank or a retailer, and the differences decide whether you end up with a working system or an expensive pilot.
Why care providers buy AI consulting differently.
Four things about the care sector change what a good engagement looks like.
- The price is set for you. NDIS price limits are now set by the Minister on advice from the National Disability Insurance Agency (NDIA), and Support at Home subsidy rates are set by government. You cannot price your way to margin, so the return on AI has to come from taking cost and risk out of administration. Any proposal that cannot say which admin hours it removes is not built for your sector.
- There are several rulebooks. NDIS supports answer to the NDIS Quality and Safeguards Commission, and agency-managed claims are paid by the NDIA. Home care answers to the Aged Care Quality and Safety Commission and claims through Services Australia. Many providers run both from one workforce. Our piece on AI and the NDIA explains which rules sit with which body.
- Privacy rules 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 takes in most care providers. From 10 December 2026, organisations that use a computer program to make decisions that could significantly affect a person must also say so in their privacy policy. Our piece on the 10 December privacy rule sets out what that means for AI agents.
- The IT team is small or does not exist. Most providers run on a care management system, a rostering tool, payroll, finance and a learning platform, looked after by a few people with other jobs. Whatever gets built has to be run by those people after the consultant leaves.
Before you sign: four things to settle.
A care provider should settle these four things with any AI consultant before a contract is signed. They take an afternoon, and they save months.
- Which rulebook each task follows. A progress note or an incident report answers to the NDIS Commission or the Aged Care Quality and Safety Commission. A claim or a price check answers to the NDIA or Services Australia. The consultant should map each task you want automated to the rules it must meet, before anyone designs anything.
- Where participant data will go. Every prompt, log and backup that holds a participant's record is health information. Ask where each one is stored, and ask to be shown rather than told. Our piece on where your AI data can live covers the usual leaks.
- Which decisions stay with a person. Eligibility, restrictive practices, a participant's goals and anything that changes the support someone receives should stay with a named human. Agree the list in writing.
- What is fixed. A fixed scope at a fixed price tells you the firm has done this before. Our guide to what AI consulting costs in Australia shows how to read the quote.
The first eight weeks: foundation before agents.
The first eight weeks of a good engagement for a care provider should produce a governed data foundation and one live view of the business, not an AI agent. That sounds slow. It is the quickest route to an agent that works, because an agent that reads six systems that disagree will produce six kinds of wrong answer.
- Weeks one and two: map it. Every system, every manual export between them, and the path a shift takes from roster to timesheet to payroll to claim. Most providers find at least one step that lives in one person's head.
- Weeks three to six: join it. Connect the core systems into one governed data foundation, with access set by role so a support worker, a coordinator and a director each see what they should and nothing more.
- Weeks six to eight: show it and rank it. A live dashboard leadership actually uses, covering cash flow, onboarding, compliance and capacity, and an automation roadmap that ranks each manual job by what it costs and what automating it would return.
That is the shape of the engagement in our NDIS provider case study: a provider of roughly 250 staff on six disconnected systems, joined into one governed foundation with a live role-based dashboard inside eight weeks, at a fixed price. The roadmap projected around $645K of conservative annual impact. That figure is a projection from the roadmap, not a realised saving.
Then the first agents.
With the foundation in place, the roadmap decides which AI agents come first, ranked by return rather than by what is interesting to build. For most care providers the early candidates are the same.
- Progress note drafting, with every note approved by the worker who delivered the support.
- Claim checks before lodgement, comparing each line against the support, the participant's funding and the current price rules, and flagging exceptions to a person.
- Incident report drafting, with the reporting clock tracked and the decision to notify left with a person.
- Audit evidence, gathered continuously rather than in the fortnight before an audit.
Each agent needs a named human approver and a log of who approved what. If an agent helps decide who gets which support, it is likely to fall under the 10 December disclosure rule, and the consultant should tell you so before it is built, not after.
What you should own at the end.
At the end of the engagement you should own everything it produced: the data pipelines, the code, the prompts, the evaluation cases, the correction logs and the documentation, with your own people able to run it. Our piece on who owns the knowledge your AI creates explains why the correction logs matter most.
Agree how success will be measured before the work starts: admin hours removed, claim rejection rate, time to mobilise a new participant, audit findings. Our guide on how to measure an AI consulting engagement sets out how.
Questions to ask an AI consultant who wants your care business.
- Who regulates what we do, and who pays us? A consultant who cannot separate the NDIS Commission from the NDIA, or the Aged Care Quality and Safety Commission from Services Australia, will build one system for two rulebooks.
- What have you built for a care provider, and is it still running? Ask for the case study and the outcome, and ask what was projected and what was realised.
- How would you show this system to an auditor? If the answer is a slide, keep looking.
- What happens when the price rules change? They change every year, and in 2026 both the NDIS and Support at Home rules changed. A price change should be a rule update, not a rebuild.
- Who will do the work, by name? Ask for the people, their experience and how much of their time is yours.
Why RUBIX.
RUBIX is an independent, Australian-owned data and AI consultancy, founded in Melbourne in 2011. For 15 years our work has been the data underneath AI: platforms, governance, reporting and, more recently, the agents that run on top of them. We have delivered more than 450 projects for over 115 organisations.
Fifteen years in the Australian market matters in care more than most sectors, because the rules move every year. The NDIS pricing arrangements changed in September 2026, Support at Home started in November 2025, and the privacy disclosure rule arrives in December 2026. A firm that has delivered through many years of Australian regulatory change builds for the next change, rather than treating the current rules as fixed.
In care, the evidence we can show is the NDIS provider engagement above: six systems joined, a live view in under eight weeks, fixed scope, and a roadmap with the numbers attached. We will be plain about the limits of that record. We have not worked for the NDIA, and we have not yet delivered for a home care provider. The data problem in home care is the same one: rosters, notes, claims and compliance held in separate systems. That is the problem the NDIS engagement solved.
If you are not sure where to start, an AI readiness assessment tells you which of the jobs above your records can support today. Our about page sets out how RUBIX has grown since 2011.
Frequently asked questions.
What does an AI consultant do for an NDIS or home care provider?
A good one starts with the data, not the agent. In the first eight weeks they map your systems, join them into one governed data foundation with access set by role, give leadership one live view of the business, and rank the manual jobs worth automating by what each would return. Agents for progress notes, claim checks, incident reports and audit evidence come after that.
My care management software already has AI. Do I still need an AI consultant?
Sometimes not. If the job sits entirely inside one system, switching on the built-in feature may be enough. You need outside help when the job crosses systems, for example a claim check that needs the roster, the timesheet, the participant's funding and the current price rules, or when you have to show an auditor how the system reached its output.
Does the Privacy Act apply to small NDIS and home care providers?
Usually, yes. The Privacy Act's small business exemption does not cover organisations that provide a health service and hold health information, which takes in most care providers. From 10 December 2026, organisations that use a computer program to make decisions that could significantly affect a person must also say so in their privacy policy.
Has RUBIX worked with NDIS or home care providers?
RUBIX has delivered an NDIS engagement: six disconnected systems at a provider of roughly 250 staff joined into one governed data foundation, with a live role-based dashboard, inside eight weeks at a fixed price. RUBIX has not worked for the NDIA and has not yet delivered for a home care provider.
General information only, not legal, financial or regulatory advice. Current as at October 2026.
