OneAdvanced brings IQ to NZ: what to check before switching on AI

What OneAdvanced announced

UK business software vendor OneAdvanced launched IQ across Australia and New Zealand on 5 November, ChannelLife reports. OneAdvanced calls IQ a "system of work". IQ is a layer that sits across the company's sector software. The layer connects workflows, data and staff, and adds AI features such as recommendations, automated admin tasks and role-based guidance. The target sectors are local government, education, aged care, logistics, mining, manufacturing and commercial services.

OneAdvanced says it has more than 20,000 customers globally. Its named ANZ customers include TAFE Queensland, NSW Health, Super Retail Group and DP World. The New Zealand voice in the launch coverage came from Ryman Healthcare. Solutions architect Nicky Dawn told iTWire that the aged care operator expects staff to get quick answers from its clinical policy library. Neither report includes ANZ pricing, packaging or a general availability date.

IQ already existed before this launch. OneAdvanced released IQ in the UK in May 2026, and a release wave in September added IQ Chat, multi-model AI and early agentic workflows, according to TechIntelPro. The ANZ launch brings that same platform to local customers.

Why the governance claims need a closer look

OneAdvanced is leading with governance, and the wording varies between reports. ChannelLife says OneAdvanced's responsible AI policies align with ISO 42001. iTWire describes IQ as one of the first business platforms in Australia to integrate the standard and says IQ complies with its requirements for responsible AI. Buyers shouldn't treat those claims as equivalent. ISO/IEC 42001, published in December 2023, sets requirements for an AI management system inside an organisation. Certification against a management system standard applies to a named organisation and a defined scope. A buyer should ask whether OneAdvanced holds a certificate, which legal entity holds the certificate, and whether the IQ services delivered to ANZ customers fall inside its scope. Policy alignment without certification still has some value, as long as procurement records which of the two the vendor actually provided.

The determinism claim also needs testing. CEO Simon Walsh told ChannelLife that IQ "applies sector-specific context, corporate policies and deterministic rules" within workflows. Workflow rules produce the same result every time they run. The language models behind chat and recommendations can return different answers to the same question. A council approving a purchase order needs to know which approval steps run on fixed rules and which depend on model output. An aged care provider answering a medication policy question needs the same answer, plus a named person who carries responsibility when the model gets it wrong. OneAdvanced hasn't published that split in either launch report.

Data residency is the third gap. Both reports mention data residency commitments, but neither names a hosting region. TechIntelPro says IQ picks a suitable AI model for each task and doesn't name the model providers. For a New Zealand organisation, where records are stored and where model inference runs can be two separate questions. A policy query sent to an offshore model leaves the country even when the database stays in an Australian data centre.

What the launch means for NZ organisations

The pattern applies to NZ buyers of almost any ERP or line-of-business system, whether or not they use OneAdvanced. Vendors are adding AI layers to software you already license, and those layers arrive through scheduled release waves. When that happens, the decision to enable a feature can end up with an operations manager instead of whoever owns AI risk in the organisation. Vendor launch material now names specific standards and controls, so buyers can test those claims against contract terms before go-live.

Ryman's example points to the first use case most NZ firms will try: answering staff questions from a policy or procedure library. That carries less risk than letting AI create invoices or change stock records, so it's a sensible starting point. The quality of the answers depends on the library. An assistant drawing on a superseded procedure will give staff a confident wrong answer, and nothing in the interface will warn them.

Organisations in OneAdvanced's target sectors should also expect the feature set to change quickly. TechIntelPro describes the September agentic workflows as foundations for a fully agentic experience. A contract signed for policy search today may cover software that takes actions within a year.

Practical next steps

• Ask your vendor for its ISO/IEC 42001 certificate, the certifying body and the scope statement, and keep them with your contract. If no certificate exists, get the vendor's alignment claim in writing.
• Get the hosting region for stored data and for model inference in writing, along with the list of AI model providers acting as sub-processors.
• Map each AI feature to the process it touches. Mark which steps follow fixed rules and which use model output, and require human approval for model-driven steps that change financial, HR or clinical records.
• Before enabling policy Q&A, remove superseded documents from the source library and give each policy set a named owner.
• Find out whether new AI features in a release wave arrive switched on or off by default, and decide who in your organisation approves enabling them.
• Record baseline figures, such as time to find an answer or rework on transactions, before rollout so you can measure the effect afterwards.

If you're reviewing AI features across your finance and operations software, our ERP team can help you map what each feature touches. For setting approval rules and governance before vendors switch more features on, see our AI consultancy page.