This journey began with robotic process automation (RPA). RPA is software that handles repetitive tasks that require human attention. The impact was immediate and quantifiable.
“If organizations didn't have RPA, I think they would need 20% more people, probably 500 or more people,” Paris says.
From there, the company implemented machine learning and generative AI in customer service, marketing, network management, procurement, and sales. This second phase focused on improving productivity, with the AI making recommendations to prompt humans to take action.
One NZ now claims to be among the world leaders in what Paris calls “agent AI” – systems that act autonomously without human supervision. The company has integrated 31 AI agents into its organizational structure to work alongside human employees.
“It took us eight hours to build our first agent AI and two weeks to deploy it, which is much faster than it takes to hire and train people for roles within our organization.”
Applications extend beyond back-office automation. AI Network Concierge can now triangulate data across multiple systems in seconds. This is a task that previously required multiple teams. The system proactively resolves customer issues and proactively optimizes network performance before problems occur.
One New Zealand achievement was astonishing. The AI system automatically powers the network infrastructure, saving 20% on energy bills.
During major events such as Eden Park rugby matches, One NZ's robots optimize network capacity without human supervision. During storms, AI manages the distribution of generators across the network in near real-time, working toward what Paris calls a “living, breathing, self-optimizing, self-healing network.”
The impact on customer experience is equally dramatic. “When I joined this organization, on average, a customer called us once every seven months,” Paris says. “Now it's once every two years.” Many network problems are now resolved before customers even know they exist.
One NZ's AI also handles cyber security at scale. The City of Paris cites an example where an AI tool identified and blocked 120,000 fraudulent messages targeted at a single customer. This network faces hundreds of attacks every day from foreign attackers seeking to compromise critical infrastructure. AI now detects attacks and automatically responds to them.
AI technology is currently available from major providers such as OpenAI, Google, and Microsoft. He says there is a recognition that data needs to be completely clean before it is used in AI systems. That's not true. AI can quickly structure unstructured data. This is what technology is good at.
Another common misconception is that legacy systems cannot be integrated with AI systems. Paris insists it's possible, although it takes effort.
“If you're not using AI, you should be. This is the business case Baidu. You just need to start. If you don't, you'll be left behind and destroyed by this technology.”
Telcos like One NZ have traditionally planned four to five years ahead, but when it comes to AI, that timeline no longer works. The time scale is compressed.
Having the right skills is important. A quarter of One NZ's AI budget goes towards employee training and tooling, helping teams understand that AI will enhance, not replace, their jobs.
Parris said the biggest hurdle to AI adoption is not technological. “The biggest hurdle is culture and getting organizations to think differently about how they need to operate.”
You just need to start. Otherwise, you will be left behind and confused by this technology.
In this case, the company would be, in his words, “fixing the plane while it's flying,” adopting an operating model that would require a complete transformation to keep pace with the rapid evolution of technology.
This tension between legacy structures and AI capabilities could be a defining challenge for New Zealand businesses in 2025. It's not about whether you adopt technology, it's about whether you can transform your organization fast enough to make the technology work.
The Paris experience at One NZ represents the cutting edge. It also highlights the widening gap in New Zealand's AI adoption story. While large companies with deep resources are moving forward, small businesses face a different reality.
Craig Young, CEO of technology user organization Tuanz, sees this difference every day. His organization created the AI community because its members felt “a little bit stuck” trying to understand what AI meant for their businesses.
“Big companies have the money to invest and can push forward. In the mid-sized space, they have some control over the tools they use, but they're very busy,” he says.
Economic pressures create further difficulties. Investing in new technology without a quick return is difficult when an organization is working hard just to survive.
What Tuanz sees across the membership is primarily ad hoc use of AI rather than strategic implementation. Employees use AI tools to assist in their work, but there is no organizational tracking or control. “There's real value there, but organizations aren't necessarily finding use cases that provide immediate benefits,” Young says.
One important lesson has become clear: keep humans under constant surveillance. “There are a lot of places where things can break down if you forget that there are people in the process,” Young says.
Young describes himself as “more of an AI realist than necessarily an optimist.” This technology delivers value once an organization establishes a use case. “But it always takes longer than you think. It's more work than you think. And it always costs more than you think.”
While businesses grapple with implementation challenges, some communities are taking matters into their own hands. Nelson high school teacher Richard Brudvig-Lindner saw the release of ChatGPT in November 2022 as a transformative moment comparable to the launch of Windows 95.
His answer was to create the Nelson AI Sandbox. It is a storefront operation on a main street where anyone can experiment with AI tools for free.
“We are not cheerleaders or evangelists for AI. We are just saying that AI is real. So what we can do is expand its use and make people aware of the concerns.”
The initiative works with dozens of nonprofits and small businesses and has a 97% satisfaction rate. Through funding from the Rada Foundation, an organization operating on a tight budget found a way to increase staff productivity. The sandbox has also spawned local startups.
This weekend, perhaps the world's first “Vibe-a-thon” will be held. It involves teams using AI-powered coding tools to collaboratively build games that teach the use of AI.
As adoption grows, so do concerns about risk management. Natassja Savidge, a Christchurch-based technology consultant at Inde, sees a pattern in how organizations approach AI adoption.
“People are asking for AI, often when what they actually need is an automation or integrated solution. The most concerning requests come from boards that mandate AI implementation without identifying a specific business problem. The first question should be, 'What's the pain point?'”
Risk extends beyond wasted investments. Free AI tools allow you to expose sensitive data to public training datasets. “Some people are uploading their company's Excel files to ChatGPT, which is free, and that information is being fed into publicly trained data,” Savidge said. In some cases, the use of chatbots has revealed private information about executives' salaries.
Perhaps most importantly, Savidge says that over-reliance on AI risks undermining fundamental skills. “If you were outsourcing everything to AI, and this big problem happens and the AI gets stuck in the loop, you’re going to have to pay a lot of money to have someone do it for you.”
She knows of an organization in New Zealand that rushed to automate its call center and laid off staff, only to have people rehired within weeks after the AI solution failed.
“Over time, we're losing the ability to constantly monitor humans.”
