AI in healthcare - what's changed, and what leaders should do next.

Ray Delany
Dec 15
4 min read

When we first released our eBook, A Practical Guide to AI Adoption for Health Leaders in New Zealand, AI in healthcare was already gathering momentum, but it still felt a bit new and often misunderstood. Since then, a lot has happened.

Some AI releases have been genuinely impressive; others have fallen short. The hype has softened, scepticism has grown in some quarters, but beneath the noise one trend is undeniable: AI in healthcare is not slowing down. It's accelerating.

In fact, 2025 has seen healthcare firmly establish itself as an AI powerhouse. Globally, the sector is now deploying AI at 2.2x the rate of the broader economy, with AI spend reaching $1.4B, nearly triple that of 2024. This growth isn't driven by curiosity, it's driven by acute operational pressures, workforce strain, and the need for measurable return on investment (ROI).

This matches the themes we highlighted in our eBook: health organisations don't adopt AI because it's fashionable. They adopt it because the pain of not adopting it is becoming greater.

The strongest momentum is concentrated in practical, high-value use cases:

Ambient clinical documentation: Ai scribes continue to be the standout success story. In New Zealand and Australia, clinicians are freeing up to 1-2 hours per day by automating consultation notes. This aligns with our eBook examples of AI scribes reducing documentation burden and improving the quality of patient records. The uptake here is strong and growing.
Coding and billing automation: Administrative pressure is one of the biggest drivers of clinician burnout. Automating coding, billing and audit tasks is delivering measurable ROI within months, not years.
Clinical decision support advancing but still cautious: Predictive models for readmissions, deterioration and triage continue to mature. As we emphasised in the eBook, clinical AI is powerful, but requires strong human oversight, clear guidelines, and governance frameworks to ensure safe use.
Operational optimisation: Scheduling, rostering, supply chain and demand-forecasting systems are becoming standard in 2025. These systems save money, reduce waste, and relieve pressure on overstretched teams.

Less hype, more reality

Despite the surge in adoption, the mood in 2025 feels different. There is less breathless excitement and more grounded assessment.

Some recent AI releases have been underwhelming, exposing exaggerated vendor claims. Others have surprised even seasoned clinical leaders with their stability and accuracy.

This has contributed to:

Reduced hype
More measured investment
Greater demand for evidence
A focus on tools that deliver actual productivity rather than theoretical promise

We saw this same realism reflected in our eBook: early adopters succeeded by starting small, solving real problems, involving frontline teams, and monitoring outcomes.

Governance is catching up

One of the most important developments in 2025 is the maturing of global and local regulatory frameworks.

Global guidance is accelerating:

Australia produced the 2025 Final Report on Safe & Responsible AI in Heath Care
Specific requirements for explicit patient consent for high-risk clinical AI
Stronger clarity around data reuse for secondary purposes (e.g. model training)

New Zealand is moving too

Our eBook highlighted New Zealand's growing focus on safe AI evaluation and cultural appropriateness.

In 2025, that work has advanced further with the Medical Council signalling more detailed expectations around:

Clinician oversight
Explainability
The role of AI in clinical judgement
Transparency with patients

This mirrors the global trend: "human in the loop" is no longer a suggestion - it's a requirement.

AI update keeps growing

Even as scepticism appears there is no stopping the uptick in use. The simple reason is this: AI is now delivering real, measurable value, especially in medium-sized health organisations that can be more nimble in their approach - the precise audience our eBook was written for.

Across the region, AI is:

Improving documentation accuracy
Automating repetitive admin
Streamlining rostering and supply chains
Enabling earlier clinical interventions
Improving patient access and experience

These are exactly the areas where our eBook called for pragmatic, high-impact pilots - and that advice has only become more relevant.

So, what should organisations do next?

Here is simple, practical guidance, reinforced by what we've seen in 2025, and applicable to any New Zealand health organisation:

Start with one painful workflow

Pick a process that everyone complains about:

After-hours admin
Patient communication bottlenecks
Rostering
Clinical documentation
Backlog triage

Solve one problem first. Build trust. Then expand.

Keep clinicians and staff at the centre

Your people are your biggest asset. Success comes from:

Involving clinicians early
Piloting with small cohorts
Gathering feedback
Adjusting tools based on real-world usage

This human-centred approach sits at the core of CIO Studio's philosophy, and the early adopters in our eBook demonstrated its importance.

Demand transparency from vendors

Ask:

How does it work?
What data is it trained on?
How is bias mitigated?
How will clinicians review or override it?
What happens when it gets something wrong?

If a vendor can't answer clearly, that's a sign you should pay attention to.

Strengthen governance now

Create simple, practical guardrails:

Human oversight for all clinical decisions
Clear escalation pathways
Equity checkpoints for Māori and Pacific communities
Transparent patient communication
Ongoing monitoring and evaluation (not set-and-forget)

Strong governance is the backbone of safe adoption.

Measure outcomes, not features

Don't be distracted by shiny capabilities. Instead track:

Time saved
Cost avoided
Errors reduced
Throughput improved
Clinician satisfaction
Patient experience

AI is not about technology. It's about people and better care.

AI is becoming ordinary - in a good way

The story of 2025 is not about hype or grand predictions. It's about AI becoming a practical, everyday tool. One that helps clinicians focus on people, not on paperwork, and helps health organisations stretch limited resources further.

The organisations succeeding today are those treating AI as a tool, not a transformation. They are starting small, learning fast, involving their people, and grounding every decision in governance.

As we move through 2025 and beyond, the question for New Zealand health leaders is no longer: "Should we use AI?". The real question is: "Where will AI help us most - and how do we use it well?".