AI and the NDIA.

An NDIS provider answers to two national NDIS bodies, and most conversations about AI only mention one. The NDIS Quality and Safeguards Commission gets the attention, because it registers providers, audits them and receives incident reports. The National Disability Insurance Agency (NDIA) gets less, but it runs the part of the scheme where AI saves the most time: plans, budgets, price limits and payments.

If you are thinking about AI for claims, rostering against plan budgets or keeping up with pricing, the NDIA's rules are the ones your system has to encode. They changed again in September 2026, and more changes are already scheduled. This piece sets out what the agency's rules ask of a provider, where AI helps against them, and where it has no business deciding.

Two bodies, two jobs.

The NDIA is the Commonwealth agency that runs the National Disability Insurance Scheme. It works out participants' plans and funding and pays for supports. The NDIS Commission is the regulator of providers: registration, the Practice Standards, complaints, reportable incidents and behaviour support. The Commission's own summary is blunt about the line between them: it does not regulate the NDIA.

For an AI system, that split decides which rulebook each task follows. A progress note, an incident notification or an audit trail answers to the Commission, and we covered that side in AI for NDIS providers. A claim, a price check or a budget forecast answers to the NDIA's rules, and to the Minister's pricing determination. That is the side this piece is about.

What the NDIA's rules ask of a provider in 2026.

Five sets of rules shape any system that prices, claims or plans against NDIS funding.

• Price limits now come from the Minister, on NDIA advice. Under the National Disability Insurance Scheme Amendments (Securing the NDIS for Future Generations) Act 2026, the Minister became the decision maker for NDIS price limits. The first determination, made in September 2026, sets the 2026-27 limits after taking advice from the NDIA and adopts the NDIS Pricing Schedule 2026-27. The NDIA still runs the Annual Pricing Review and publishes the schedule providers work from.
• Only listed supports can be funded. Since 3 October 2024, NDIS supports are defined by lists in the transitional rules. A support outside the lists cannot be funded, apart from limited replacement supports the NDIA approves. Every line on a claim has to map to a support that is allowed.
• Claims have clocks. A payment request must be made within two years of the support being delivered, and the NDIA may hold claims lodged more than six months after the service for review. Participants can be NDIA-managed, plan-managed or self-managed, and the path a claim takes depends on which.
• The claiming system is changing under providers' feet. For plans in the NDIA's new computer system, service bookings are gone. A provider has to be recorded as the participant's "my provider", and claims go through a bulk payment request.
• Planning changes next. The NDIA's new way of planning, built around support needs assessments, rolls out from April 2027. The agency has said NDIS plans will continue to be approved by trained NDIS staff, not automated systems.

And the rules keep moving. The NDIA's pricing review for 2026-27 recommended differentiated pricing for registered and unregistered providers from 1 January 2027, which the government is still consulting on. The Fair Work Commission's decision on Schedule E of the SCHADS Award takes effect on 1 December 2026, and the government has said the Pricing Schedule will be updated later this year to reflect any required changes. A pricing or claims tool that needs a developer for each of those changes is already behind.

Where AI agents earn their keep against the agency's rules.

The good uses look the same as elsewhere in the NDIS: high volume, a published rule to check against, and a person who signs off. Against the NDIA's rules, four stand out.

• Pre-claim checks. Before a payment request goes in, an agent checks each line against the current price limit, the support lists, the service agreement and what is left in the participant's budget, and flags every mismatch for a person to clear. Rejected and under-claimed lines are where provider cash leaks.
• Price changes as rule updates. When the Pricing Schedule changes, an agent can compare the old and new schedule line by line, show which services and participants each change touches, and draft the updated rates for a person to approve. That turns a mid-year award change into a review rather than a rebuild.
• Claim clocks and plan-type routing. An agent that knows whether each participant is NDIA-managed, plan-managed or self-managed can route every claim the right way and warn well before a delivered support gets close to the six-month review point or the two-year limit.
• Budget forecasting. Projecting each participant's spend against their plan, so a coordinator sees an overspend months ahead rather than in the last claim, is a well-bounded job for an agent. The coordinator decides what to do about it with the participant.

Where AI must not decide.

The NDIA's rules draw some of these lines themselves. Whether a support is reasonable and necessary, and what goes into a plan, are the agency's decisions, and the NDIA has said its own plans stay with trained staff. A provider's AI should not try to predict or pre-empt them either. Inside the provider, a person decides what to tell a participant about their funding, whether to dispute a rejected claim, and whether a price change means renegotiating a service agreement.

The Commission has taken the same line on the safeguarding side. Its February 2026 position statement on AI in behaviour support plans does not endorse AI tools, does not stop providers using them, and warns that putting participant information into a public tool without safeguards may breach the NDIS Code of Conduct. We have written more generally about where AI agents do not work yet.

Two privacy rules apply here too. The Privacy Act's small business exemption does not cover an organisation that provides a health service and holds health information, which captures most NDIS providers at any size. And from 10 December 2026, an 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. Keeping a named person on every funding decision is good practice, and it also keeps the system on the right side of that rule.

What to demand from the firm that builds it.

Ask any AI partner these five questions before you sign.

• How does it handle a price change? Ask them to walk you through the 1 December 2026 SCHADS change. If the answer involves a developer and a release, the next change will too.
• Can it tell the two rulebooks apart? A claims check that confuses the NDIA's pricing rules with the Commission's Practice Standards will flag the wrong things.
• Where does participant data go? Every prompt, log and backup should stay in Australian regions. Our piece on where your AI data can live covers the places it usually leaks.
• Who approves what the agent produces? Every claim, rate change and forecast should have a named human approver, and the system should log it.
• Will they start with an assessment? Claims checks only work if rosters, agreements and plan budgets can be joined up. An AI readiness assessment tells you whether your records can support them 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. To be plain about it: we have not worked for the NDIA, and we do not build the agency's systems. We work for providers, on the provider's side of the rules.

• We have done this for an NDIS provider. For a leading NDIS provider we joined six disconnected systems into one governed data foundation, a live role-based dashboard and an AI automation roadmap worth ~$645K a year, in under eight weeks. Read the NDIS provider case study.
• 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. A pre-claim check needs rosters, agreements and plan budgets in one place. 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 and data stays in Australian regions by default. 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.

Many providers deliver in-home aged care as well, where the rules differ again. Our piece on AI for home care providers covers Support at Home. If you run an NDIS business and want to know what AI can safely take off your claims and pricing work, 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 NDIS Pricing Schedule and the NDIA's current guidance for the rules that apply to your organisation.