Finance teams are no longer asking whether AI matters. The sharper question is where it matters, how quickly it's changing, and what a sensible finance leader should do next.
That was the focus of our most recent webinar, "How Automation and AI Are Transforming Modern Finance Functions", featuring Scott McLiver, Chief AI Officer at PwC New Zealand.
The discussion covered AI hype to practical finance use-cases, enterprise risk, human-led agents, and the decisions CFOs need to make now.
The current value of AI can be explained in three buckets: software development, customer service, and knowledge-worker productivity.
The first two are powerful, but not universal. Some organisations build software themselves or run large contact centres to service their customers. Many do not.
The third bucket matters to every finance team. It is the compounding value of smaller productivity gains. A finance model that adapts to evolve a 10-hour task into a two-hour task. Invoice and contract checks that move from four days each month to one day, with a human still reviewing the output.
These are not always headline-grabbing examples. They're often the work finance teams already know too well: month-end support, document review, reconciliations, analysis, modelling, and control checks.
AI can be a powerful tool for delivering value to finance teams, but it must be rolled out safely and appropriately.
One of the strongest themes from our discussion was the pace at which AI models are evolving. There has been a recent leap in frontier models where tasks that were unreliable only months earlier became genuinely useful. In finance terms, McLiver compared the improvement to moving from a "junior bookkeeper" to a "three-year experienced accountant".
That pace creates a problem for traditional project thinking. If a business spends heavily building a narrow AI tool today, there is a real chance that a frontier model or enterprise SaaS provider will make the same capability available cheaply within months.
For finance leaders, the lesson is not to avoid AI investment. Rather, it is to raise the bar for bespoke builds.
The most practical advice from the webinar was also the simplest: get enterprise-grade frontier AI into the hands of your people, then train them properly.
Not a lunch-and-learn. Not a passive webinar. Laptops-open training that helps knowledge workers become power users, then build human-led agents that support real work.
This is the step that improves day-to-day productivity and helps leaders make better decisions about where larger AI investments should go. Right now, too many organisations are trying to "pick the All Blacks" before they've learned the rules of rugby, as McLiver put it.
Traditional automation is useful, but brittle. If an invoice format changes, a supplier moves a field, or an exception appears, the process can break.
AI makes automation more flexible because it can interpret information more like a person. It can identify the invoice number, amount, GST, quantity, or date even when the format changes.
Agentic AI goes further. It's practically the same digital brain, but with tools, access, and the ability to complete a workflow under instruction. In finance, that could mean drafting reconciliations, checking documents, preparing analysis, or coordinating multi-step tasks with human oversight.
In all finance discussions, accuracy, auditability, security, and compliance still matter.
Consumer AI tools are not appropriate for business data. Vibe-coded tools may look impressive, but they need the same control environment as any enterprise system. For finance teams dealing with IFRS 16, AASB 16, FRS 102, month-end reporting, or audit evidence, that distinction is critical.
The safe paths are clearer: use enterprise-grade frontier tools, and rely on enterprise software providers who are embedding AI with appropriate guardrails.
Tool selection can become a reason to delay. Copilot, ChatGPT Enterprise, and Claude will all continue to evolve, and switching will become easier because many agents are built from instructions rather than traditional code.
The more important decision is to start using AI within your business safely, build internal fluency, and identify the finance processes where AI can reduce manual effort without weakening control. AI is not replacing the need for finance expertise, rather it's making that expertise more scalable.
For finance teams, the opportunity is not just to move faster but to build better systems, reduce manual pressure, and create more confidence in the numbers that guide the business.
Talk to the LOIS team to explore how automation, enterprise-grade controls, and smarter lease accounting workflows can support your finance function.