NZ’s supply chains are highly exposed. Could new AI agents make them more resilient?

If 2026 has taught New Zealand businesses anything, it’s that the next supply chain shock is never far away.

Iran’s closure of the Strait of Hormuz sent oil prices above US$100 a barrel, added weeks to shipping routes and saw carriers slap surcharges of hundreds of US dollars per container on cargo bound for New Zealand.

DHL’s latest Export Barometer found 87% of New Zealand exporters faced higher shipping costs over the past year, while only 7% escaped supply chain disruption altogether.

From the COVID-19 pandemic, to the Ever Given’s temporary blockage of the Suez Canal in 2021, to the latest crisis in the Red Sea, recent years have brought other repeated reminders of this vulnerability.

New Zealand’s economy, with exports of around $80 billion a year, lives or dies on its supply chains. And each time there is an issue, attention lands on ships, ports and fuel reserves.

What gets far less attention is the function inside every enterprise that actually manages this exposure day to day: procurement.

Procurement decides which suppliers a firm depends on, on what terms, with what alternatives, and how quickly it can re-source when a route closes or a supplier fails.

In a small, distant, trade-dependent economy, procurement is not back-office administration. It is the corporate immune system.

Yet in most New Zealand organisations it remains under-resourced, manually operated, and largely untouched by the AI wave now transforming other business functions.

From chatbots to agents that act

The technology now arriving in procurement is different from the chatbots and drafting assistants most managers have already met.

New agentic AI systems can carry out entire multi-step tasks – deciding, acting and adjusting course along the way – with a human supervising the work rather than driving it.

In procurement, that means agents that discover and vet suppliers, prepare and assess tenders, analyse contracts and spending, track supplier performance and continuously scan for risk.

Internationally, this has moved well past the experimental stage. In Boston Consulting Group’s 2026 Tech Procurement Study, which I contributed to, we surveyed more than 200 procurement professionals and interviewed industry leaders and technology providers. Most enterprises surveyed are already piloting or deploying agentic AI in procurement.

Overseas, the question is no longer whether to adopt agentic AI, but how to redesign the procurement organisation to get the most from it.

How AI can help firms absorb shocks

For New Zealand firms, distance means longer lead times and fewer alternatives when things go wrong. The faster a business can identify a problem and find, vet and contract another supplier, the better placed it is to respond.

Manually, that cycle takes weeks. A risk-monitoring agent can scan shipping data, news feeds and supplier registers around the clock, compressing detection to hours. A supplier-discovery agent can assemble and pre-screen alternatives in days.

AI can also give smaller procurement teams capabilities previously available mainly to large companies. Most New Zealand companies can’t afford the large sourcing and supplier-management teams of the multinationals they buy from and sell to. A three-person team could potentially do work that once required many more staff.

The main barriers are within businesses themselves. Our respondents identified fragmented data, outdated software and governance as key problems. Some 71% cited trust in autonomous decision-making as a barrier, alongside security, regulatory uncertainty, accountability and auditability.

One finding is particularly relevant for New Zealand firms, most of which will buy this capability rather than build it. Performance outcomes were similar whether agentic AI was developed in-house, co-developed with a partner or bought as a managed service.

Buying a capable solution is a perfectly good way to start. But a large gap emerged in companies’ own AI expertise: 46% of in-house builders rate their internal capability as fully sufficient, against 7% of co-developers and 6% of outsourcers – and in-house organisations run the most autonomous agents.

That suggests firms can buy the technology, but still need people who understand how it works, the data it relies on and the decisions it makes. Building that expertise will allow them to give AI agents more responsibility safely.

From risk to resilience

For New Zealand procurement leaders, the lesson is to look at the whole procurement process, rather than automate individual tasks.

Procurement and IT need to work together, staff need the skills to manage the technology, and firms should plan from the start for successful trials to expand into everyday use.

For boards and executives, the message is simpler. Every scenario exercise run since COVID-19 has concluded that New Zealand needs more resilient supply chains.

Resilience in procurement is built transaction by transaction. The technology to transform that function is ready, and international organisations already using it are gaining experience that latecomers cannot simply buy.

In a country whose prosperity depends on goods moving across the world’s longest and most exposed supply lines, giving procurement the intelligent agents it needs is a question of national economic resilience.