What’s the biggest misconception firms have about agentic AI?

That it’s a smarter chatbot. It isn’t. A chatbot answers a question and stops. An agentic system reasons, plans and runs a workflow from start to finish through specialised agents that hand work to each other. In accounting, raw intelligence isn’t enough on its own. The system has to understand engagement context, your firm’s methodology, the regulatory obligations that apply and how one workpaper depends on another. Without governance and human oversight built into it, what you have is a fast tool, not a trusted one.

What separates a serious platform from a feature bolted onto existing software?

Architecture, not marketing. An assistant improves a task, but the task was never really the problem – the chain is. Production runs from onboarding through ingestion, extraction, trial balance mapping, preparation, review and sign-off, and most of the cost sits in the hand-offs rather than inside the steps. Bolt AI onto one stage and you simply move the bottleneck somewhere else. So, ask any vendor three things: show me the audit trail, show me where the human approves and show me what happens when the model gets it wrong. The third one is the tell.

If AI can prepare an entire workpaper, what’s left for the accountant?

Preparing workpapers was never the point; judgement was. Machines can handle reconciliations, documentation and a competent first pass. Materiality, unusual transactions, client-specific calls and professional scepticism cannot be outsourced. It does change how you train juniors, who can no longer learn the work by grinding through it, and firms need to think about that now rather than in five years. New Zealand is short of people and that isn’t ending. The real question isn’t whether to cut headcount, it’s whether scarce senior judgement goes to work that needs it, or to formatting a lead schedule.

Trust is everything in accounting. Why should a practitioner trust an AI-generated workpaper?

They shouldn’t, until they can see how it was built. Trust comes from transparency, not automation. Every conclusion should trace back to a source document, a calculation and a stated confidence level, with the reasoning there to inspect. Nothing reaches a client file without a human approving it. Sign-off stays exactly where it has always been, with the practitioner whose name is on the engagement.

What does the accounting firm of 2035 look like?

Organised around orchestration, rather than production. Agents run the routine work continuously, while accountants supervise, resolve exceptions and apply judgement where it counts. Capacity stops being a function of headcount. But the firms that pull ahead won’t be the ones that bought the most software. They’ll be the ones that redesigned their methodology, retrained their people and set clear rules about what AI may and may not decide. Start with one process, prove it against your own review standards, then scale.

Book a Zato workflow walkthrough

Zato is a CA ANZ Member Benefits partner. CA firms that enter into an Annual Ledger Subscription Agreement with Zato receive ledger credits valued at A$2500. To learn more, visit this link.