Griffin on Tech: The AI agent revolution goes local, but can NZ keep up?

It’s been a blockbuster week for New Zealand’s AI scene. Microsoft dropped a new report suggesting our economy could benefit to the tune of $76 billion a year by 2038 if we embrace generative AI at pace. 

It arrived as Autohive, a spin-off from successful Wellington software testing company Raygun emerged from stealth, pitching itself as the infrastructure backbone for the coming wave of AI agents.

I’ve had more pitches in the last two weeks from emerging New Zealand AI startups than in the last three months. A lot of momentum is building, wrapped in the accompanying hype. But it is good to see our companies embracing the agentic AI movement in everything from financial services to land information management.

Microsoft suggests that if New Zealand leans into AI adoption, particularly in applications and data centres, we could see our GDP per capita jump by 14% and unlock $76 billion in annual economic value by 2038. 

The Redmond, Washington-based software giant claims that a 6-month trial of Copilot with a public sector agency “delivered an estimated 287% return on investment” and saved workers an average of 10 hours per month on routine tasks.

That’s before Microsoft gets its own AI agents into action in the public sector. But a Bloomberg report also suggests that large enterprises are struggling to get staff to use Copilot because their people prefer OpenAI’s ChatGPT, despite Copilot being based on some of OpenAI’s models.  

Microsoft’s report warns that if we treat AI as a bolt-on automation tool and neglect workforce upskilling, digital inclusion, and infrastructure, we risk a “conservative” scenario: half the economic benefit, higher unemployment, and a place in the global slow lane. The call to action? Double down on data centres (leveraging our renewable energy edge), foster a vibrant AI startup ecosystem, and, crucially, get serious about digital skills and regulatory agility.

Autohive and the AI agent tipping point

While the big players talk infrastructure, Wellington’s Autohive is betting on the next phase of the AI revolution: agents. These aren’t your garden-variety chatbots. As McKinsey’s latest report lays out, AI agents are proactive, goal-driven virtual collaborators that can automate complex workflows, adapt in real-time, and even make decisions on your behalf. 

Think squads of digital workers handling everything from customer service to supply chain orchestration—freeing up humans for higher-value tasks. Autohive’s pitch is to make this transition accessible for Kiwi SMEs, providing the infrastructure to build, deploy, and manage fleets of AI agents without the resources of a Fortune 500. It’s an ambitious play—and one that could help New Zealand punch above its weight if it can scale, and if Raygun founder and Autohive CEO John-Daniel Trask can carve out a place for Autohive in the increasingly crowded AI agent space.

Here’s the rub: most companies are stuck in what McKinsey calls the “gen AI paradox”—widespread deployment, minimal impact. Nearly 80% of businesses have tried generative AI, but just as many see no real bottom-line gains. 

The reason? Too many horizontal pilots (copilots, chatbots), not enough deep integration into core business processes. The real value, McKinsey argues, will come when we redesign workflows from the ground up, putting agents—not humans—at the centre.

The roadblocks: Skills, trust, and governance

The AI opportunity is undeniably huge, but so are the challenges. Microsoft and McKinsey both highlight the need for a massive upskilling effort, new governance frameworks to manage agent autonomy and risk, and a cultural shift to build trust in AI-driven processes. New Zealand lags its peers in digital skills and AI research, and our VC investment in AI startups is still fairly limited.

Investments in local data centres is valuable, but unless we tackle the skills gap and foster a robust local AI ecosystem, we risk becoming just another market for global tech giants, not a creator of value in our own right.

Chorus: Fibre as national infrastructure

The other part of the infrastructure puzzle that was in the spotlight this week was broadband, with Chorus seeing its plan to extend fibre from 87% to 95% of the country given a glowing value-for-money rating by the Infrastructure Commission. 

 The Commission projected $17 billion in benefits over the next decade if Chorus was given the green light to extend ultrafast broadband to 95% of the population, an additional 160,000 homes and premises.

As Chorus CEO Mark Aue puts it, “Reliable, scalable, high-performing connectivity is no longer a luxury; it’s essential”. The catch, of course, is the sheer cost of it - estimated at $2.5 billion - $3 billion. 

This is more than just faster Netflix in the wop-wops. If AI is the engine of future growth, fibre is the fuel. Without world-class broadband, rural New Zealand risks missing out on the productivity and innovation gains AI promises. Yet, as Chorus points out, we still lack a coherent rural communications strategy—something the government needs to fix, pronto.

Fibre is the gold-plated standard for internet access, but with satellite services, wireless providers and the mobile operators all serving the rural sector, I can’t see the Government writing a check to support the extensive fibre build that Chorus recommends, even if the RoI really does stack up. The easier, if technically inferior option is likely to win out in the current environment.

Startups going gangbusters

Finally, as I wrote in my BusinessDesk column this week (paywalled), what a stellar run for our tech startups. In the last two months, three of them alone (Halter, Tracksuit, and Projectworks) have raised $217 million in new capital, much of it from major US VCs who have had a hand in the success of a host of well-known Silicon Valley companies.

I don’t think this is an anomaly either, with more deals in the pipeline. It’s a sign of the maturing startup ecosystem and the growing international interest in what we have to offer. It was great to see Xero, our homegrown cloud accounting juggernaut, make headlines with its proposed $4.1 billion acquisition of American payments platform operator Melio. That’s a hugely ambitious move by Xero - combining the best accounting software with a high-quality payments provider.

As startup investor Rowan Simpson quipped on LinkedIn: “The US startup ecosystem loses another promising high-growth company to a foreign buyer. 🤨“, a wry reference to the complaints that usually accompany the sale of a Kiwi startup to offshore buyers that its not pretty much dead to us.

The opposite is increasingly true, with development teams being retained in New Zealand, some of the sales proceeds being recycled into funding the next generation of startups. That’s the cycle that helped make Silicon Valley great, and it’s increasingly happening here.