Top 10 AI Prompts and Use Cases and in the Financial Services Industry in New Zealand

New Zealand's financial sector is fast moving from experiments to real-world AI: regulators and industry alike are mapping the risks and the wins so firms can deploy use cases such as chatbots and virtual assistants, AI-driven fraud detection, risk modelling and KiwiSaver personalisation with confidence.

The Reserve Bank's “Rise of the machines” special topic flags financial‑stability implications and the need for oversight, while the FMA's research digs into current adoption, governance and where firms plan to apply AI next - from faster credit decisions (loan approvals that can move from days to minutes) to automated compliance and document processing.

Recent market studies show adoption surging (over 82% of organisations by 2025 and big‑firm use rising to the high‑60s in 2024), so this Top 10 collection pairs practical prompts with NZ use cases and governance guardrails.

For teams wanting hands‑on prompt skills and workplace AI fluency, consider practical training like Nucamp AI Essentials for Work bootcamp registration to learn promptcraft, tools and real business applications, and read the FMA research on artificial intelligence in New Zealand financial services for the regulatory view.

“We reviewed how AI is currently used in New Zealand financial services and firm's plans for future applications. We sought to understand both the benefits and the risks to inform more oversight.”

To select the Top 10 AI prompts and build realistic examples for New Zealand teams, the process was grounded in the FMA's 2024 sector study: the shortlist focused on use cases the report found in active deployment or imminent use - customer‑facing chatbots and draft communications, fraud and anomaly detection, risk modelling and operational automation - and on the controls the regulator flagged as essential, like data quality, vendor oversight and explainability.

The research drew on responses from 13 finance representatives across asset management, banking, financial advice and insurance (a c.40% response rate noted in commentary), and informed prompt design that tests both business value and governance readiness; prompts were therefore framed to surface outputs that are reliable, contestable and auditable.

Where possible, examples mirror real FMA findings (e.g. nine‑in‑ten users citing customer outcomes and firms planning future customer‑interaction use), and the methodology stayed transparent by aligning prompt scenarios with FMA analysis and legal insights such as those summarised by Lexology.

Read the FMA research and the legal commentary for full context.

“We reviewed how AI is currently used in New Zealand financial services and firm's plans for future applications. We sought to understand both the benefits and the risks to inform more oversight.”

Customer service chatbots and virtual assistants can be a practical win for New Zealand banks: when designed with a clear purpose they speed responses, handle volume spikes and let human agents focus on complex cases - exactly what Kiwi customers expect as branches shrink and digital channels grow (New Zealand in-branch banking trends and digital channel growth).

A bank such as ASB could deploy an omnichannel assistant to triage queries, verify customers, and escalate urgent fraud or mortgage cases to specialists, following the design with purpose playbook in Infosys's conversational‑banking guide (Infosys conversational banking guide for chatbot best practices).

Real-world vendor case studies show measurable outcomes - for example, early adopters have reported rapid adoption and high deflection rates (24% user uptake and up to 87% chat deflection in some deployments) - underscoring the so what: fewer routine calls, faster problem resolution and tangible cost savings when governance, data access and escalation paths are built in from day one (Emerj review of chatbots for banking customer service outcomes).

Regulatory compliance teams and legal drafters in Aotearoa are already testing supervised document generation to cut repetitive work while keeping humans firmly in the driver's seat: the Parliamentary Counsel Office's R&D with local partners produced clause‑by‑clause explanatory‑note and plain‑language prototypes that show AI can draft useful first passes if prompts are tightly scoped and outputs are reviewed, reworked and hosted with strong data controls (New Zealand Parliamentary Counsel Office AI drafting proof-of-concept).

At the same time, the Ministry for Regulation's work on the Regulatory Standards Bill illustrates how LLMs can rapidly classify very large volumes of feedback - 22,821 discussion‑document submissions were processed and then spot‑checked with a qualitative sample - to help officials meet tight timelines without losing the thread of public sentiment (Regulatory Standards Bill submissions analysis).

These New Zealand projects underline practical guardrails lawyers and regulators should adopt - redact personal data, insist on human‑in‑the‑loop review, log prompt logic and model choices, and treat AI as a co‑pilot rather than a policy maker - to preserve transparency, mana whenua concerns and public trust as compliance drafting scales up (RNZ: How AI is being used to process public submissions in New Zealand); the result can be faster, more consistent drafting without losing the human judgement that gives law its legitimacy.

Metric	Value
Total submissions (discussion document)	22,821
Qualitative sample (share of submissions)	4.1%
Share of text covered by sample	34.4%
Staff‑reviewed sample size noted	939 submissions (+605 separately reviewed)

“As a rule of thumb, having humans in the loop will be the best practice - humans in charge and AI as a co‑pilot.”

For lenders such as Toyota Finance New Zealand, fraud detection and anomaly investigation now sit at the intersection of regulation and machine learning: a risk‑based AML/CFT programme - required by supervisors such as the FMA and RBNZ - starts with a thorough written risk assessment and a fit‑for‑purpose transaction‑monitoring system that flags true threats while keeping false positives manageable.

Modern systems use AI to spot patterns like smurfing (many small transfers) or rapid layering through newly opened accounts, and they feed alerts into human‑review workflows so suspicious activity reports (SARs) and prescribed transaction reports (PTRs - e.g.

NZ$1,000+ international wires, NZ$10,000+ cash) can be filed promptly and records retained for audit. Practical implementation draws on screening services for PEPs, sanctions and adverse media and on best practice for third‑party oversight; see the FMA guidance on AML/CFT obligations and examples of AML transaction‑monitoring systems for how teams can design, test and tune detection rules in line with supervisory expectations and operational realities.

Risk modelling, scenario analysis and stress testing are where New Zealand firms turn uncertainty into board-ready choices: by strengthening data infrastructure, aligning models with strategic goals and involving senior management, stress tests become forward‑looking tools that translate climate, cyber or geopolitical exposures into capital, liquidity and hedging actions.

PwC's guidance on stress testing stresses these exact levers and why institutions that act now build resilience and clearer decision paths, while PwC New Zealand's Risk Services brings together actuarial, data and model‑governance skills to build, validate and monitor models for banks, insurers and treasury teams exposed to FX, commodity and dairy risks.

The practical gain is memorable - a scenario that once lived in a slide deck can instead trigger a concrete hedging decision or capital buffer before stress hits the balance sheet - shortening the path from insight to action.

For practical frameworks and model‑governance detail see PwC's stress‑testing summary and PwC New Zealand Risk Services.

Scenario analysis metric	Share
Currently perform climate scenario analysis or stress testing	38%
Plan to perform soon	29%

“Stress testing's importance has surged due to APRA's emphasis on risk management amid global risks like climate change and cyber threats.” Nina Larkin, Partner, PwC Australia

Personalised robo‑advice is closing New Zealand's retirement‑planning gap by turning simple questionnaires into tailored KiwiSaver plans: local tools such as BetterSaver, kōura Wealth, GoalsGetter (Nikko am) and Milford ask a few focused questions to recommend portfolios, contribution rates and glide‑path adjustments that ordinary Kiwis can understand (see MoneyHub's KiwiSaver digital advice comparison).

That personalisation matters now more than ever because the FMA found the share of KiwiSaver assets in high‑volatility (risk category 5) funds quadrupled - from about 10% in 2021 to over 40% in 2024 - so automated advice that flags risk tolerance and long‑term impact can prevent “set and forget” choices from becoming costly.

Hybrid and AI solutions are scaling advice too: National Capital's pioneers and new ventures like Sevaka aim to automate large shares of routine servicing while embedding compliance checks to keep outputs auditable and safe (read the Investment News coverage).

The practical payoff is concrete - better fund matches, nudges to increase contributions, and cheaper access to advice for people who previously had none - all of which help translate policy shifts (like higher default contribution settings) into better retirement outcomes for most members.

Metric	Value / example
Share in risk category 5 (2021 → 2024)	≈10% → >40% (FMA)
Main KiwiSaver digital advice tools	BetterSaver, kōura Wealth, GoalsGetter (Nikko am), Milford (MoneyHub)

“Our findings show that the increase of the default employee and employer contribution settings could result in retirement funds lasting on average approximately 30% longer than under the pre-Budget 2025 settings for median salary and wage earners who contribute without interruption over a 40-year working life.” - Jane Wrightson, Retirement Commissioner

Underwriting and claims automation is rapidly turning time‑consuming paperwork into board‑level outcomes for New Zealand insurers: intelligent document processing (OCR + NLP) and rules engines speed FNOL intake, triage claims by severity, and surface key underwriting signals so adjusters and underwriters can act sooner, not later.

Modern platforms extract messy PDFs and handwritten notes, run computer‑vision damage checks, flag fraud patterns and hand off only the complex cases to people - delivering concrete gains such as faster settlements, fewer manual touchpoints and clearer audit trails (see VCA's guide to claims automation).

A vivid, practical payoff is real: some systems can move from approval to payment in as little as 15 seconds after settlement, while enterprise deployments report straight‑through processing rates above 65%, slashing backlog and freeing staff for high‑value work (examples and metrics summarised by DeepOpinion).

For NZ teams, the lesson is pragmatic - pilot IDP on high‑volume claim types, keep humans in the loop for liability calls, and instrument every workflow for explainability and regulatory evidence so automation boosts customer trust as much as efficiency.

Metric	Example value (source)
Claims processing cost reduction	≈30% (VCA)
Faster claims resolution	50–70% faster (VCA)
Straight‑through processing (STP)	>65% (DeepOpinion)
Payment latency after approval	~15 seconds (VCA ClaimPay)

For New Zealand accountants and SME bookkeepers, pairing Aider with Xero turns routine bookkeeping into a proactive advisory engine: Aider syncs in real time with Xero (practices can connect in under two minutes) and pre‑analyses client data into a colour‑coded dashboard that acts as an early‑warning system - telling firms exactly which clients to call “ASAP” so work shifts from chasing receipts to delivering timely advice.

The platform automates period‑close checklists, flags data‑quality issues, and generates AI‑produced Performance Reports and meeting briefs that make numbers instantly actionable, helping practices scale compliance while offering broader advisory services; see the Aider Xero integration overview and Xero's roundup of AI apps for accountants for examples and setup notes.

Metric	Value
Rating (Aider)	5.0 out of 5 stars (11 reviews)
Free trial	14‑day
Countries listed	Australia, Canada, New Zealand, United States

“At client meetings I show them the Performance Report along with their annual accounts, the clients love the visuals, it makes more sense to them. It's all AI generated, and they're like, 'Wow!'” - Jessica Stebbings, Chartered Accountant

Document processing and OIA-response automation are fast becoming a pragmatic tool for New Zealand public agencies and the financial sector alike: Auckland Transport has applied AI to speed up search, pipeline automation and employee analysis, while the Ministry of Transport's published OIA correspondence shows officials may use AI to help analyse submissions - subject to rules about personal data and transparency - so teams can triage thousands of responses faster and surface themes for policy or compliance reviews.

Local case studies show concrete wins - semantic search and intelligent pipelines improve turnaround, and AT's analytics work has halved batch processing time - so legal teams can move from manual redaction and keyword searches to auditable, human-reviewed summaries that preserve privacy and mana whenua considerations.

For practical reading, see the Ministry's OIA thread on AI use and RNZ's roundup of public agencies adopting AI, and explore AT's Dataiku case study on semantic search and automation for document‑scale workflows.

Metric	Value (source)
Batch file processing time	4.5+ hrs → 2 hrs (OpenText / AT)
Video analytics running	500+ (OpenText / AT)
Cameras in use	~200 (OpenText / AT)
Data extracted	~1 TB/month (trains); ~8 PB/week (street cameras) (OpenText / AT)

“We may use an artificial intelligence tool to help us analyse submissions. We will take steps to avoid inputting personal information into any AI tool that is outside our network.”

Digital humans are becoming a pragmatic tool for New Zealand banks and insurers that need high‑quality, scalable roleplay and training: Soul Machines' Experiential AI powers lifelike “Digital Workers” that see, hear, remember and even empathise, so staff can practise negotiations, customer conversations and tricky onboarding scenarios with a responsive, consistent partner rather than a scripted bot - a format that also helps scale frontline coaching and reduce time spent on routine call handling.

The platform is built to sit inside existing workflows (so a simulated call can update CRM records or trigger an escalation), teams can trial Studio and the Digital Workforce quickly, and the company's New Zealand roots (co‑founders and R&D anchored to the University of Auckland and NZ tech leadership) make the tech especially resonant for local firms balancing adoption with trust.

For practical pilots, teams can try a short free trial or run training cohorts where learners spend real practice time - Soul Machines reports average conversations of about 20 minutes, a vivid sign that digital roleplay can feel more like coaching than a checklist.

Item	Detail
Platform	Soul Machines Digital Workforce official website
Company background	Soul Machines about page - NZ founders & University of Auckland R&D
Trial	7‑day free trial (Studio / Digital Workforce)
Workforce Connect price	$40,000 annually
Average conversation	~20 minutes (Soul Machines)

“The future of AI lies not in manipulating reality, but in enhancing our understanding of it. By focusing on Biological AI, we can create AI systems that are ethical, effective, and truly transformative.”

Management dashboards and decision‑support tools are the glue that turns messy data into board‑ready choices for New Zealand finance teams: consolidate CRM, ERP and forecasting feeds to present one clear story - so a CFO can see cash, liquidity and key risks in a single slide rather than wrestling with siloed spreadsheets.

Local planning and analytics vendors show this in practice - teams using AI‑enabled planning can cut the financial close from 10 to 3 days and run scenario models in real time (Fusion5 planning and analytics solutions for New Zealand) - while better board reporting practices mean dashboards must be visual, concise and linked to reliable source systems so directors get actionable insight fast (Board reporting best practices - insightsoftware).

The same data pipeline starts at the front line: CRM adoption (including mobile access for field teams) feeds richer customer and operational signals into decision models, so management can move from reactive firefighting to confident, timely strategy calls (Best CRM for New Zealand businesses - Capsule CRM).

A vivid payoff: what once took a week of spreadsheet stitching can now become a single, auditable dashboard that updates in seconds - helping boards act before a small variance becomes a big problem.

Getting started safely in Aotearoa means marrying ambition with discipline: run small, measurable pilots, embed humans in the loop, log model choices and data lineage, and treat governance as a product not an afterthought - a core takeaway from the FMA's sector review of AI use and risks in New Zealand financial services (FMA sector review of AI in New Zealand financial services).

The Government's July 2025 AI Strategy and its Responsible AI Guidance offer a light‑touch, risk‑based framework that encourages firms to adopt proven tools while safeguarding privacy, explainability and competition - and notes AI could add roughly NZD76 billion to GDP by 2038 if adopted responsibly, a vivid reminder of the upside for firms that get governance right (New Zealand AI Strategy and Responsible AI Guidance (July 2025)).

Practical next steps: pick a high‑value, low‑risk use case; document impact, controls and escalation; engage legal and risk early; and upskill teams with workplace courses such as the Nucamp AI Essentials for Work bootcamp - practical AI skills for the workplace so promptcraft and auditability become routine.

Course	Length	Early bird cost	Registration
AI Essentials for Work	15 Weeks	$3,582	Register for Nucamp AI Essentials for Work (15-week bootcamp)

“We reviewed how AI is currently used in New Zealand financial services and firm's plans for future applications. We sought to understand both the benefits and the risks to inform more oversight.”