The situation

Manawa Ora is a regional Primary Health Organisation serving 85,000 enrolled patients across 22 general practices. The network spans urban clinics and rural health centres staffed by 450 people — GPs, nurses, practice managers, administrators, and allied health professionals. Like many PHOs across Aotearoa New Zealand, Manawa Ora operates under sustained pressure from multiple directions at once.

Reporting requirements have expanded steadily over the past three years. Each practice submits data to the PHO, which consolidates and reports upward to Health New Zealand, the Ministry of Health, and multiple funding bodies. The volume and complexity of these reports continue to grow, while the staff capacity available to produce them has not kept pace. Practice managers and senior administrators carry the bulk of this workload alongside their operational responsibilities.

At the same time, patient complexity is increasing. More patients present with multiple long-term conditions, and the shift toward integrated care models means clinical and administrative teams are managing more coordination across services. Workforce shortages — particularly prevalent in provincial and rural practices — add further pressure, with fewer people absorbing a growing scope of work.

Against this backdrop, some clinicians and administrators have quietly begun experimenting with AI tools. GPs have begun using AI-assisted clinical notetaking supplied by third-party commercial app developers. A handful of practice managers are using AI to draft patient communications and summarise meeting notes. Not all of this activity is sanctioned or governed — organisational policies exist in some places and not others, there are no/few shared guidelines, and there is little visibility at leadership level into what tools are being used or how patient data is being handled.

This is not unusual. Across the public sector — in health, local government, and social services — staff are finding their own ways to manage increasing workloads with the tools available to them. The challenge for organisations like Manawa Ora is not that staff are exploring AI, but that the organisation has not yet established the governance and guidance to make that exploration targeted, safe and productive.

The specific challenges

• Reporting burden: Practice managers across 22 practices spend an estimated 12–15 hours per month on manual data consolidation and reporting. The process is repetitive, error-prone, and consumes time that could be directed toward patient services and practice operations.
• Shadow AI use with no oversight: Staff are using consumer AI tools for tasks involving patient communications, clinical notes, and administrative summaries. There are organisational policies governing AI use, but these were written over a year ago, and are not adequate to guide staff in today’s fast-changing AI environment. There are no approved tool lists, and no visibility into whether sensitive or patient-identifiable data is being entered into these tools.
• Change fatigue: Manawa Ora’s teams have navigated significant structural change over recent years — health system reforms, new reporting frameworks, and evolving funding models. Appetite for another large-scale initiative is low. Any new programme needs to be contained, practical, and clearly connected to the pressures staff are already managing.
• No governance confidence to approve pilots: Leadership recognises the potential of AI to support operations and clinical workflows, but lacks the governance frameworks, risk assessment processes, and policy foundations needed to approve even small-scale pilots with confidence. Without these foundations, the default position is inaction — while unsanctioned use continues.

Where we’d start

We’d start with something contained and low-risk: supporting practice managers and administrators with AI-assisted reporting and communication tasks, using approved organisational templates and structured prompts. This is a practical entry point because it addresses one of the most visible time pressures — the monthly reporting cycle — without touching clinical workflows or patient-facing systems.

Starting here achieves two things. First, it delivers an immediate, measurable reduction in administrative burden for the staff who need it most. Second, it gives the organisation a controlled environment to develop and test governance guidelines — approved tools, data-handling protocols, and usage boundaries — before extending AI use into more complex or sensitive areas. It’s a first step that builds both capability and confidence.

How we’d work with Manawa Ora

Discovery (weeks 1–3)

We’d start by understanding the specific challenges Manawa Ora is facing. Every engagement begins with listening, not assumptions.

We’d interview a cross-section of stakeholders — clinical leads, practice managers, senior administrators, and PHO leadership — to map the current reporting workflows, understand where AI is already being used informally, and identify the specific governance gaps that need to be addressed. This phase is focused on understanding the organisation’s situation, not confirming assumptions.

Duration: 2–3 weeks, depending on practice availability and scheduling across the network.

Design (weeks 3–6)

We’d design around Manawa Ora’s specific situation, drawing on what we learned in discovery, the regulatory environment, and the priorities the leadership team has identified. Working collaboratively with a small design group — including practice managers and at least one clinical lead — we’d develop:

• A tested prompt library for the most common reporting and communication tasks, built around Manawa Ora’s actual templates and workflows
• Clear, practical guidelines for safe AI use — which tools are approved, what data can and cannot be entered, and how to handle edge cases
• A draft governance framework covering AI use policy, risk assessment criteria, and an approval pathway for future pilots

Design is collaborative. The team is part of the process, so the outputs are practical, accepted, and owned by the people who will use them.

Duration: 2–3 weeks, with collaborative sessions scheduled around operational commitments.

Delivery (weeks 6–10)

We’d deliver a combination of training, workshops, and coaching designed for the people doing the work:

• Small-group workshops for practice managers and administrators — hands-on sessions working through the prompt library and guidelines using real (de-identified) reporting tasks. Focused on building confidence through practise, not theoretical AI knowledge.
• Governance briefing for PHO leadership and clinical governance leads — a structured session walking through the governance framework, risk assessment process, and pilot approval pathway. The leadership team leaves with the tools to make informed decisions about AI use across the network.
• One-to-one coaching for clinical leads at pilot practices — tailored sessions to support individual clinicians exploring AI-assisted workflows within the approved guidelines. Personalised, practical, and designed around each person’s role and comfort level.

Every prompt, guideline, and framework document is designed for Manawa Ora and provided in full. These are all included.

Duration: 2–4 weeks, with delivery scheduled across practices and sites.

Review (post-delivery)

We’d check in at 30 and 60 days post-delivery to ensure value is realised, provide ongoing support, and capture lessons for continuous improvement. These check-ins would assess:

• Whether the prompt library and guidelines are being used as intended, and what adjustments are needed
• Whether the governance framework is supporting confident decision-making at leadership level
• What additional use cases practice teams have identified, and whether the organisation is ready to extend the pilot

Duration: 30 and 60-day post-delivery check-ins, with email and asynchronous support throughout.

What the organisation would gain

• A tested prompt library for staff to use in reporting and communication tasks, built around Manawa Ora’s actual templates and workflows
• Practical AI use guidelines — approved tools, data-handling protocols, and clear boundaries for safe use across the network
• A governance framework including AI use policy, risk assessment criteria, and a pilot approval pathway
• Training materials for practice managers, administrators, and clinical leads — designed for ongoing use and internal delivery
• Governance confidence — the leadership team equipped with the frameworks and evidence to approve future AI pilots knowing risks are understood and managed
• Practical capability across pilot practices — staff with the skills, tools, and confidence to use AI safely and productively in their daily work
• A clear pathway forward — documented lessons, identified next use cases, and a framework for extending AI use across the network at a pace that works for Manawa Ora

This scenario is a composite example based on sector research and common organisational challenges. We maintain strict confidentiality for all client work.