- One NZ skipped the easy AI wins and went after streamlining enterprise provisioning
- Working with UiPath, the operator automated processes across multiple disconnected systems
- It’s eyeing additional back-office tasks in its quest to be the most AI-enabled telco in the world
One New Zealand (One NZ) has set an ambitious goal for itself to become the most AI-enabled telco on the planet. But where does an operator even start a journey like this? The answer may vary depending on who you ask but in One NZ’s case, it wasn’t interested in starting small.
“We deliberately chose a highly complex pilot. We weren't interested in proving AI could automate a simple, isolated task,” One NZ’s GM for AI Cy Wright told Fierce. “To make the pilot worth the investment of time and money, we wanted to demonstrate that it could operate across multiple platforms, navigate real-world process complexity, and orchestrate work end-to-end.”
It’s target? Enterprise provisioning. One NZ teamed up with UiPath to automate the process across Salesforce, Oracle and its internal platforms using the UiPath’s Maestro orchestration platform.
Ashley Boag, COO for UiPath’s International Region, told Fierce that before the rollout, the main bottlenecks to provisioning were the disconnected systems One NZ uses and the resulting manual processing that that setup required. This left the operator with a 10-day provisioning timeline.
But with Maestro, One NZ was able to map out its processes and remove the manual chokepoints. “AI agents coordinate the work across systems and software robots execute within applications, so you get that end-to-end coordination as an orchestration layer on top of what’s already there,” Boag said. She added one of the biggest misconceptions is that telcos need to replace their existing systems to leverage AI.
Wright said the process cleanup was both aided by AI and required for the AI to work.
“The challenge is understanding the process, the dependencies, the decision points, and the downstream consequences well enough for AI to orchestrate and optimize it effectively,” he said. Wright added that what appeared at the surface level to be a single provisioning process was a “highly fragmented set of workflows."
“Without AI, it would have been extremely difficult to obtain the process knowledge required to clean up the fragmented handoffs and without the process cleanup, AI would not have had the context needed to orchestrate and validate the workflow effectively,” he explained.
All in all, One NZ was able to go from proof of concept to production in five weeks and saw ROI in less than six months, Boag said.
Next on One NZ’s AI roadmap
Many operators have targeted customer support use cases as the low-hanging fruit in their AI adoptions. But Wright said that One NZ believes some of the greatest value AI has to offer telcos will come from modernizing back office operations. This includes provisioning, of course, but also network planning, service and revenue assurance, billing and process orchestration.
He added One NZ is confident that the approach it applied in the enterprise provisioning pilot can be applied elsewhere in the business.
One NZ is “actively working on finance processes, data modernization, billing and payments, and we're even automating aspects of our retail digital platform prior to its replacement.” In the latter, AI is being used to help generate requirements, document downstream processes, identify dependencies and generally create a better understanding of how the existing platform operates – before One NZ begins to replace it.
The operator has also been working with Red Hat on cloud and automation, and - like countless other enterprises - it also adopted Microsoft CoPilot.
Balancing cost and innovation
Plenty of recent headlines have highlighted how businesses are facing rising AI costs. In the telco realm, Boag argued that a lot of the cost pressure operators feel stems from their decision to treat AI as a standalone expense layered on top of existing operations. When functioning like this, “every new model or tool adds integration overhead and compute cost without a clear return.”
For its part, One NZ is carefully monitoring utilization, optimizing agent deployments and measuring outcomes against each original business case.
“If an AI solution is reducing service calls, lowering handling times, improving employee engagement, accelerating delivery, or improving customer experience, then the investment is justified,” he explained. “If it isn't delivering measurable value, we need to challenge whether it should be scaled further.”
He concluded: “As the industry matures, I think the organizations that succeed won't necessarily be the ones spending the most on AI. They'll be the ones with the strongest governance, visibility, and value management disciplines.”