When One New Zealand set out to modernise its order provisioning process, the results exceeded expectations. A workflow that previously took around 10 days now completes in under 10 minutes. The project, built on UiPath’s Maestro orchestration platform, is part of a sweeping AI transformation program across the telco.
Cy Wright, One New Zealand’s General Manager, said the most significant outcome was visibility, not speed.
“The automation is almost the cream on top,” Wright said. “It’s actually the knowledge you get from performing the automation. Previously, if we had tried to put a process person out there to understand that process, it would have been incredibly difficult. It was tribe knowledge.”
The process in question handles roughly tens of thousands of orders per year across a complex ecosystem of digital, Salesforce, and Oracle platforms. Before automation, the work was performed manually by an offsite team, with no on-site staff Wright could consult and no end-to-end documentation to draw from.
Getting the system right took longer than expected. The team went through approximately 120 configuration changes during user acceptance testing, spending three weeks testing on live orders rather than the one week originally planned. The reason was fundamental: there were no test environments that replicated real-world process variation.
“We had to wait for live orders with real process variations to actually understand it,” Wright said. “What that now gives me is choice. I can see where somebody might need training on how they process an order, or if I do a configuration change in the source system, how much more efficient it will make the throughput.”
Why start with the hard problem
Conventional wisdom in enterprise technology favours low-hanging fruit as an entry point for transformation. Wright chose the opposite approach.
“Our logic was, if we could get automation to work here, we could get it to work anywhere,” he said. “When the guys produced the first POC, I said: this doesn’t prove anything. We need to be comfortable that it will deploy across our actual environments and work.”
Feiran Hao, UiPath’s VP, Pricing & Product Strategy, said the One New Zealand project reflected a broader shift in how his company approaches automation strategy.
“Low-hanging fruit was how we thought about tackling things in traditional automation,” Hao said. “Our language is shifting. We say, let’s take a look at some of your bigger problems. For the first time in technology, we have a platform that can orchestrate straight-through or dynamic workflows, automate every piece along the way, and with coding agent support, develop even faster. What used to be a way to get fast wins has become feasible to do at a much bigger scale.”
Human where it matters
One New Zealand operates a formal responsible AI policy and an internal training program called Elevate, which defines where AI can and cannot be applied. Every agent deployed must pass privacy and compliance review, regardless of the use case.
Wright said the governance framework was non-negotiable. “We need these agents to be highly governed. We can’t afford for a customer’s phone to be sent to the wrong location. So, we have small teams building and a lot of people validating. Anything that goes outside our guardrails triggers immediately back to a human in the loop.”
He was direct about accountability. “AI is a tool. Everything I deploy, my team and I are accountable for that deployment.”
Hao said UiPath’s orchestration philosophy was built around the same principle. “Certain decisions, especially when it comes to affecting customers, still need to be human made,” Hao said. “AI can analyse a situation and make recommendations, but the human should ultimately make the decision. And when that human makes a decision, they should have full traceability back into why the AI made its recommendation.”
What comes next
Wright said his team has approximately 30 more processes across the order to cash processes with new user cases added every day to the backlog.
The immediate priority alongside delivery is capability uplift. One New Zealand wants its own teams to own and operate these workflows rather than relying on external delivery resources.
The project has already attracted attention. Wright said One New Zealand has started advising other organisations navigating similar challenges, with the provisioning automation serving as a working demonstration of end-to-end agentic orchestration at production scale.
“When I presented it to our board, the fact that they could see an order orchestrating across these environments, every step in the process visible, that was incredibly powerful,” Wright said. “Now other customers are asking how we can help them on this journey.”
