Work Smarter, Not Harder: Top 5 AI Prompts Every Finance Professional in New Zealand Should Use in 2025

For New Zealand finance teams, mastering AI prompts is less about novelty and more about control: the FMA's national review shows AI adoption is growing across banking, asset management and insurance, but firms need stronger governance and risk checks to use it responsibly - so prompts must be precise, auditable and jurisdiction-aware (FMA national review on AI in New Zealand financial services).

Real-world use cases - from insurers combining satellite imagery and weather data to speed underwriting to platforms built for secure reporting - show the upside, yet professionals must guard against hallucinations and outdated datasets (Wolters Kluwer analysis of AI risks and professional obligations in New Zealand), and learn prompt patterns that preserve accuracy and audit trails while boosting productivity (Workiva finance-focused AI platform for financial reporting).

“AI is a transformative technology, and application is evolving at pace.”

Methodology blends local, hands-on testing with NZ-specific policy checks: start with practical prompting workshops and rapid pilots to convert prompts into repeatable workflows (Value 32 Practical Prompting Workshop Value 32 Practical Prompting Workshop), then validate outputs against the Government's Responsible AI Guidance - documenting purpose, training-data provenance, privacy-impact assessments, and IP/licensing checks so every prompt is auditable and jurisdiction-aware (New Zealand Responsible AI Guidance summary).

Practical pilots should assemble cross-disciplinary reviewers (data, legal, privacy, Māori-data advisors) and pair prompt-driven automation with tailored staff upskilling so accuracy, explainability and record-keeping are baked into production; the result is prompts that reduce manual toil while leaving a clear paper‑trail for compliance and stakeholder trust.

“Directors must treat AI strategy as both a technological and human-capital priority.”

Turn meeting audio into an auditable financial snapshot: use OtterPilot or Otter's real‑time transcription (it links with Zoom and can import recordings) to capture the discussion, then feed the transcript into ChatGPT with a tight prompt that extracts attendees, key numbers, decisions, and assigned action items with owners and deadlines - ending with a three‑line CFO executive summary and a one‑sentence risk note for compliance checks.

Practical steps: enable Otter's calendar/Zoom integration or import the MP3/MP4, train custom vocabulary for NZ terms and supplier names to improve accuracy, and for in‑room meetings place the mobile device next to the speaker as Otter suggests to boost clarity (Otter.ai transcription guide for meetings and webinars and the Beginner's guide to using Otter AI for transcription).

Sample ChatGPT prompt (adapt as needed): “From the following transcript, produce (1) a 3‑line CFO summary, (2) a table of decisions with financial impact, and (3) discrete action items with owner and deadline.” Last action items before publishing: confirm participant consent and data handling steps - Otter highlights user responsibility for notice and consent given legal scrutiny of transcription tools - so document approvals and retention rules in the meeting record (Legal scrutiny of transcription tools and consent issues).

“We are deeply committed to safeguarding our users' data and protecting privacy. Nobody should be recorded without their knowledge or permission, regardless of the recording device used.”

Turn regulatory drift into a fast compliance check by pairing a focused web‑search agent (Perplexity) with a powerful summariser (Gemini): prompt Perplexity to fetch the primary NZ sources (for example the DIA AML/CFT hub for accountants and the May 2025 customer risk‑rating guidance that takes effect 1 June 2025) and ask Gemini to produce a one‑page yes/no checklist that maps model outputs to on‑the‑ground obligations - e.g., does the onboarding record include a documented customer risk rating, verified beneficial‑owner fields, and OCDD triggers? Cross‑reference the Gemini summary against the RBNZ's AML/CFT guidance (updated 16 May 2025) and the FMA's customer‑due‑diligence guidance for companies so the checklist ties each AI finding to the exact rule or PDF; the practical payoff is a prompt that surfaces missing CDD items before funds move, not after a regulator's review.

Sample prompt seed:

Compare this client file to DIA, RBNZ and FMA CDD rules and list gaps, required documents and risk‑rating actions with source links and required deadlines.

(DIA AML‑CFT information for accountants (New Zealand), RBNZ AML/CFT guidance and resources (Reserve Bank of New Zealand), FMA AML/CFT customer due diligence: companies (Financial Markets Authority)).

Turn scenario work from spreadsheet busywork into a repeatable AI-driven routine by prompting ChatGPT or Gemini to ingest historical statements, select the handful of true drivers (revenue growth, DSO, gross margin, capex and staffing), and spit out baseline, best- and worst‑case cash‑flow projections plus a sensitivity matrix that shocks each driver by user‑defined percentages - then export each scenario as clean CSV/Excel sheets for modelling and board packs.

Start with the basics recommended by New Zealand government cash flow forecasting guidance (New Zealand cash flow forecasting guidance - Business.govt.nz), seed the prompt with the modelling steps from Phoenix Strategy's scenario guide (data collection, assumptions, validation, stress tests) (Phoenix Strategy scenario-based cash-flow modelling guide), and borrow tested phrasing for DCF, sensitivity and export requests from financial‑modeling prompt collections (Financial modeling prompt collection for DCF and sensitivity analysis).

The payoff is immediate: a single prompt that returns a 13‑week to 5‑year view, a sensitivity table that shows which lever kills runway, and downloadable sheets ready for CFO review - so decisions like hiring or delaying expansion become data‑backed, not gut calls.

"Cash flow, the lifeblood of any organization, flows through the organization thereby fueling every operation, investment, and opportunity." - Ryan McKinney, QueBIT

Turn fraud hunting from reactive firefighting into a repeatable, auditable routine by pairing Perplexity's fast web-and-document retrieval with n8n's automation: prompt Perplexity to scan transaction patterns and public guidance (for example, look up behavioural flags, payment anomalies and identity gaps against the DIA counter-fraud techniques) and then have n8n enrich each alert with Xero ledger trails and payment-provider signals before routing high-risk items for human review.

The workflow should map common NZ controls - segregation of duties and audit logs from Xero, identity hardening and de-duplication checks from DIA guidance, and payment protections such as 3DS and device-fingerprinting described in Airwallex payment-fraud guidance and Stripe fraud-detection resources - so every alert includes the raw evidence, the matched rule, and recommended next steps (hold payment, require secondary auth, or escalate to investigations).

This setup produces an evidence packet that's regulator-friendly and repeatable: Perplexity provides the context and source links, n8n creates the audit trail and ticket, and staff get a clear, fast triage list rather than noisy inboxes - so a

rapid burst of small transactions

or an unusual login pattern becomes a provable incident with a clear remediation path.

See the DIA counter-fraud techniques, Xero's fraud prevention tips and Airwallex's payment-fraud guidance for NZ-specific controls and integration points.

Turn the board pack into a one‑page, investor‑ready story: use ChatGPT to condense your financials into a tight “highlights” lede, a short cash‑runway table, the single KPI investors care about, and a one‑line “ask” - then move that output into a slide canvas for visual polish.

Seed the prompt with the investor‑update structure recommended by Carta (Carta guide to writing effective investor updates: Carta guide to writing effective investor updates), and the VCFO playbook for what investors expect from finance leadership in New Zealand (MoneyHub guide to VCFOs in New Zealand: MoneyHub guide to VCFOs in New Zealand) so the draft speaks the language of local backers.

For startups building a lean finance function, borrow the hiring and automation cues from Stripe's finance‑team guide when deciding who signs off on numbers (Stripe guide: building your startup finance team: Stripe guide to building your startup's finance team).

The practical payoff: a single slide investors can scan in 30 seconds that answers “how long will this last?” and “what do you need next?” - then use ChatGPT to draft spoken answers to likely follow‑ups before the meeting.

“We have 15 minutes left - let's do this: I want to hear you tell our story back to me.”

Putting prompts into production in New Zealand is as much about governance and training as it is about clever wording: map prompt workflows to the Public Service AI Framework and Privacy Act 2020, bake in datasheets and model‑card documentation, run ethical reviews and Treaty‑informed fairness checks, and keep auditable logs so Inland Revenue, auditors or an AI Forum NZ reviewer can trace who ran a prompt, why, and what data was used.

Start with low‑risk pilots, iterate with cross‑disciplinary reviewers, and translate successful pilots into repeatable playbooks that include plain‑language user notices and retention rules; think of governance like a receipt stapled to every decision - when regulators ask, the team can hand over the provenance.

Use official guidance when setting risk thresholds (see the national strategy and practical deployment toolkits) and close the skills gap with hands‑on training so staff learn both prompt craft and compliance workflows.

For concrete next steps, follow New Zealand's AI Strategy, adopt the AI Governance “Deploy” checklist, and consider team training like Nucamp AI Essentials for Work bootcamp to turn prompt experiments into safe, audit‑ready routines.

"The time has come for New Zealand to get moving on AI."