AI Strategy in NZ Vocational Education: A CEO's Board Brief

When the board asks for an AI story, a tool list and a pilot count are the wrong answer. The credible version says which parts of the business AI will change, what evidence will show it is working, and how it sits inside your quality management system and annual self-review under NZQA's new framework.

Why this lands on your desk

You answer for growth, margin and regulatory standing. On 19 January 2026 NZQA's integrated Quality Assurance Framework (iQAF) took effect, with the Quality Assurance of Tertiary Education Providers Rules 2026 replacing the 2022 rules that included External Evaluation and Review. The new rules require an up-to-date quality management system and annual self-reviews. Anything you tell the board about AI will eventually be tested against evidence you can show.

The pipeline is the second pressure. Education Counts shows PTE domestic enrolments down 14% and international enrolments up 28% in the 12 months to December 2025, with overall enrolments down 8.5%. TEC's March 2026 monitoring report says international growth is supporting sector financial performance while pressuring TEC's funding baseline. A sharp director will ask how dependent you are on one segment, and whether AI adds cost or protects revenue.

Then there is the restructure. The VET redesign legislation passed in October 2025. Ten regional polytechnics and eight Industry Skills Boards became operational on 1 January 2026. Te Pūkenga (NZIST) is a transitional entity, to be disestablished on or before 31 March 2027. TEC says work-based learning is transitioning to new or existing PTEs, polytechnics and wānanga. That is opportunity and risk at once, and your AI story needs to say which one you are betting on.

What the public record does not say

This is where most AI stories overreach. The research behind this piece found no binding NZQA rule specific to PTE use of AI. NZQA's AI guidance that turned up is aimed at schools and NCEA. Its tertiary page on academic integrity and AI is thin and mostly points to other resources, including guidance from Quality and Qualifications Ireland.

The research also found no sector-wide data on AI adoption among PTEs. It found no NZQA or TEC statement on AI in back-office or compliance functions, and no official compliance-cost or cost-per-student benchmark for PTEs.

So don't tell the board that "the sector is moving fast" or that "regulators expect AI adoption". Neither claim is supported. Say what is known, say what is not, and make your own evidence the foundation. A CEO who admits the gaps sounds more credible than one who fills them with confidence.

A board-ready AI story in four parts

1. Name two or three specific problems. Pick problems already in your numbers: slow assessment turnaround, thin practice opportunities for learners, manual evidence-gathering for self-review. Leave out "innovation" as a category.
2. Say how use will be governed. Document where AI is used, who approves it and how it is reviewed within your quality management system. Make it part of annual self-review. It should not sit alongside it as a side project.
3. Set measures before you start. Decide what "working" means in advance, in terms the board already tracks.
4. State the limits. Note what you will not do yet, and what you are waiting on from NZQA, TEC or the new Industry Skills Boards.

Part two matters more than it looks. The rules have moved the burden towards self-review and monitoring, and the research did not find that it has clearly shrunk. If AI use is invisible to your quality management system, it becomes a liability you have to explain later.

Measures the board can track

Tie the story to the three things directors care about.

• Margin. Staff hours spent on specific tasks before and after, and delivery cost per learner for the programmes where AI is used. Be honest that you have no sector benchmark, so your own baseline is the comparator.
• Regulatory standing. Whether self-review evidence is more complete and easier to produce. Whether key-person dependency in compliance has fallen. Under the 2026 rules, subcontracting applications need evidence of internal quality assurance approval signed by the Chief Executive or a delegated senior manager (per SAARA's summary), and the PTE stays fully responsible for the subcontractor. Your name is on that approval.
• Pipeline. Learner completion and engagement in the programmes you change, and employer feedback. The Tertiary Education Strategy 2025–2030 expects stronger economic relevance and more employer, industry and iwi input into programmes. AI-supported delivery that makes practice more realistic for a specific industry is a credible link to that expectation. Treat it as a hypothesis you test, not a claim you assert.

Also show the board your concentration risk. If international students are carrying your margin, an AI story that only improves domestic delivery is a different conversation from one that supports both.

Key takeaways

• Answer the board with business problems, governance and measures, not a tool list or pilot count.
• Under iQAF and the 2026 NZQA rules, AI use should be visible in your quality management system and annual self-review.
• The public record shows no binding NZQA rule on PTE use of AI and no sector adoption data. Say so plainly.
• Anchor measures to margin, regulatory standing and pipeline. With domestic PTE enrolments down 14% and international up 28%, the board will probe segment dependence.
• Position the story against the VET redesign: Industry Skills Boards, regional polytechnics, and work-based learning moving towards PTEs.

Our take

The temptation is to arrive with the biggest story. We think the better move is the smallest one that is true. A board that hears "here are two problems, here is how we will govern the fix, here is how you will know" can hold you to it. A board that hears a vision cannot, and will eventually stop believing it.

This is also a position of strength in a regulatory moment. With external review replaced by self-review, a provider that can show its own evidence on how it uses new technology looks like a provider in control of its quality. Modest claims, backed by records, are the strongest thing you can bring into that room.

The first step is small. Before your next board meeting, write down the two operational problems you would most like to fix, and what number would tell you it had worked. If you can't do that in a page, the story isn't ready.

FAQ

Does NZQA have rules on how PTEs use AI?

The research behind this article found no binding NZQA rule specific to PTE use of AI. NZQA's AI guidance found is aimed at schools (NCEA), and its tertiary page on academic integrity and AI lists resources rather than setting requirements. Check NZQA directly for any update before you commit to a position.

How should AI connect to our annual self-review?

Treat AI use as part of your quality management system. Record where it is used, who approves it, what evidence of effectiveness you will collect, and how you will review it. The Quality Assurance of Tertiary Education Providers Rules 2026 require an up-to-date quality management system and annual self-reviews, so that is where the evidence belongs.

What should I tell the board about the risk of getting AI wrong?

Say that no sector adoption data or compliance-cost benchmark was found, so you will rely on your own baseline measures. Describe your governance, name the limits of what you will do, and commit to reporting results against margin, regulatory standing and pipeline.

Why does enrolment mix matter to an AI strategy?

Education Counts shows PTE domestic enrolments down 14% and international up 28% in the 12 months to December 2025. TEC's March 2026 monitoring report links international growth to sector financial performance. Directors will want to know whether your AI investment supports the segments your margin depends on.