A select group of Kiwis are taking on the world using AI. Image / Getty Creative
A select group of Kiwis are taking on the world using AI. Image / Getty Creative
Some Kiwis, and their organisations, are tentatively prodding around the edges of artificial intelligence. Our AI disruptors have grabbed it by the scruff of the neck and used it to create new opportunities for their business or help their organisations adapt to the new terrain.
PhilipWhite: Faster building consents
Building and Construction Minister Chris Penk’s mid-2024 move to require local authorities to offer remote building inspections – and adopt inspection-by-app as their default option – came with a hat tip to a local software firm.
Penk’s initial release came with a Ministry of Business, Innovation and Employment (MBIE) video detailing how Zyte has been used by the Mackenzie District council to do remote video inspections across its 7000 square kilometres, with only one building control officer. MBIE made a similar clip promoting Auckland Council’s use of Zyte.
Unlock all articles by subscribing to this international offer
All Access Weekly
Herald Premium, Viva Premium, The Listener & BusinessDesk
“The builder essentially becomes a robot,” Zyte executive director Philip White told the Herald.
A remote building inspector, using the Zyte AI, texts a link to the builder – who can then open it without the app installed on their phone and allow the inspector to control their camera as they move around the site.
Open up the latest news from the South
Get daily South Island headlines straight to your inbox.
“The builder essentially becomes a robot,” says Zyte executive director Philip White.
Beyond video Zyte AI can use the LiDAR sensor found on recent iPhones or NeRF (neural radiance field) technology used by various Android phone apps to turn a series of stills into a three-dimensional image using a neural network (a form of AI).
The idea is to create a 3D-model building site – the better for Zyte AI to spot potential safety or compliance risks or other issues that could hold up construction.
It lets a building inspector know if a site is ready for an in-person or remote inspection or re-inspection, White says, which he bills as a travel and time-saver for squeezed building and consent authorities.
Zyte AI can also act as a more conventional video chat platform.
White’s company’s software, which rose to popularity during the pandemic, is now used by 17 building consent authorities (BCAs) around the country, including Consentium, set up as an in-house unit by Kāinga Ora in 2021 as the only nationwide BCA.
White and his partner own most of the business. The largest outside investor is Parkable co-founder Warwick Beauchamp.
Late last month, Penk confirmed another major change. Trusted builders and tradespeople will be able to sign off on their own work under a new self-certification system he said would be introduced by the end of this year.
Time was money for builders and tradies, Penk said. An industry survey found that in 2024, two-thirds of inspections took longer than the statutory 20 days.
The self-certification would apply to low-risk work, freeing up BCAs for more complex, high-risk certification, Penk said.
“Wait-time targets and controlled self-certification point in the right direction – but only if supported by the right methods, controls, and technologies,” White said.
“Proven tools like Zyte’s AI, remote, and 3D systems ensure every inspection – whether council-led or self-certified – produces a full, auditable, and compliance-proving record while cutting delays, lowering costs and avoiding costly mistakes.”
Beyond the self-certification move, there was also plenty in Penk’s April 25 announcement to keep the professionals busy, with a new provision that BCAs will have to complete building inspections within three working days. For White, that means only one thing: fire up the AI.
John-Daniel Trask: A more powerful Raygun
If you’re a company leader, what did you do when ChatGPT first hit headlines, ushering in the current era of AI buzz?
It’s unlikely you took the bit between your teeth to the same degree as John-Daniel Trask, co-founder of Wellington’s Raygun - a tech firm whose software helps more than 100,000 developers worldwide all but eliminate errors.
“ChatGPT launched and in April 2023 we had an all-hands team meeting and felt we needed to go ‘all in’,” Trask tells the Herald.
“I described it as an ‘Internet Tidal Wave memo’ situation” when Bill Gates turned the Microsoft ship, “JD” said, referencing a famous Gates memo in 1995 predicting the huge impact the internet would have on business and society.
“In May 2023 we paused all company operations for a week – aside from answering customer support,” Trask said.
“Every person, every role, spent the week building AI agents and demonstrated them on Friday to each other.
“Every team member, even those who didn’t code, first passed our engineering coding challenge and then built agents that could speak, understand our company data, and demonstrated those agents.
“All in one week, with internal models hosted on our own GPU infrastructure (we use the big models externally more these days, but did invest in helping learn how to even host the models ourselves).”
“In May 2023 we paused all company operations for a week," Raygun founder John-Daniel Trask says.
“In 2024, we built AI agents that help us operate our marketing, evaluating SEO [search engine operation], adwords, our inbound trials, and helping to produce blog content. These were built by us and are still used internally.
“For example, if I do a podcast, we have an agent that takes the link, creates a blog post – using our own style and voice – and publishes the post with approval from a team member.
“We’re seeing many different multi-hour tasks that simple agents can take care of so our team can focus on strategic goals.”
Last year Raygun also launched a new tool called “AI Error Resolution” using large language models (LLMs) to help find the root cause of software bugs in minutes rather than hours.
It picked up the Innovation Gold at the 2024 Wellington Gold Awards, backed by Victoria University and others.
Today, every team member is using AI tools.
“We have an internal AI server for open-source models, and we’re only recruiting people who have demonstrated strong personal curiosity in learning about AI,” Trask says.
Ben Goodger. Images / LinkedIn, Open AI
Ben Goodger: OpenAI secret project
University of Auckland computer systems engineering grad Ben Goodger has been Johnny-on-the-spot for a couple of software revolutions now.
After creating several core features for Netscape (kids, ask your parents), he was the lead developer for the Firefox browser in the 2000s, from his new home base of San Francisco.
After moving to Google, where he would become a vice-president, he co-founded the team that created the Chrome web browser, which would knock Microsoft’s Internet Explorer off its perch.
A few months ago, Goodger jumped ship to ChatGPT maker OpenAI, posting to social media: “I’m thrilled to be working at the frontier of technology, helping to develop products that benefit everyone!”
How exactly? It’s still under wraps. Goodger wouldn’t comment. On X, he posted a screengrab from Apple TV series Severance with the caption “The work is mysterious and important”.
But The Information – the clued-in news site founded by Wall Street Journal’s Jessica Lessin – has reported that Goodger is working on a new web browser for Open AI that has support for ChatGPT and ChatGPT search built-in.
If correct, it has the makings of a Google Chrome killer – and perhaps big implications for Google’s search ad revenue.
Natasha Crampton, Microsoft’s chief responsible AI officer, at the company’s headquarters in Redmond, Washington on June 20, 2022. Photo / Grant Hindsley, The New York Times
Natasha Crampton: Policing Microsoft’s AI
Microsoft is one of the biggest players in artificial intelligence, from its US$13 billion-plus ($22.08b-plus) investment in ChatGPT maker OpenAI to its own push with new technologies like Copilot.
And it’s a Kiwi expat who is wrangling whether the tech giant stays within its own rules and those of governments around the world.
The University of Auckland law grad joined Microsoft Australia-New Zealand as senior in-house counsel in 2018 before being promoted to senior attorney, artificial intelligence, research and human rights in 2018 at the company’s HQ in Seattle. She had a baptism of fire, with claims that the first-term Trump administration was using Microsoft facial recognition AI to separate children from their parents at the border.
Crampton helped prove the allegations incorrect, leading to a promotion to the newly-created position of chief responsible AI officer in 2019, with vice-president added to her title in 2023.
“I’ve always been interested in the intersection between technology, law and society, and that came pretty early on including what I studied at the University of Auckland,” she told the Herald.
The over-achiever landed an AJ Park Intellectual Property Law Prize, the Senior Prize in Law and a Russell McVeagh law scholarship in information systems.
Today, she couldn’t be more in the thick of it. Before they are released, Microsoft AI technologies that would be used to make important decisions about a person’s access to employment, education, healthcare, financial services or a life opportunity are subject to a review by a team led by Crampton.
Her more than six years of work around AI and ethics at Microsoft allowed the company to “move nimbly and thoughtfully”, Crampton said.
Andrew Chen. The topic of his PhD thesis, finished in 2019 was, “The Computers Have a Thousand Eyes: Towards a Practical and Ethical Video Analytics System for Person Tracking”.
Andrew Chen: Robocop
University of Auckland computer systems engineering PhD Andrew Chen mixed a career as a research fellow with tech venture capital work before taking on a less usual role last year: chief adviser, technology assurance – technically a new role, though more broadly he succeeded Carla Gilmore, who filled the role of emergent technologies manager, which was created in 2020.
The role includes assessing “Generative AI [GenAI] such as ChatGPT and other large language model and image generation AI technologies which are not currently approved for general use by Police”.
What keeps Chen up at night as the cops’ in-house tech ethicist?
“I think generative AI is creating new challenges for policing all over the world. These tools are so widely available and there are a million ways of using them that we haven’t imagined yet. We keep seeing new ways criminals are exploiting technology to harm innocent people and communities,” Chen told the Herald.
“While generative AI is not currently approved for any Police use, I’m starting to think about whether we might be able to use these tools in low-risk ways to help Police do their job better and more efficiently.
“With this I am also mindful of all the issues like reliability, bias [an area where early supermarket use of facial recognition has fallen down], information security, and data sovereignty,” Chen said.
“Police and the public sector generally are rightfully held to a higher standard in ensuring that what we do is safe and appropriate.”
Paul Copplestone cofounded Supabase.
Paul Copplestone: Enabling the vibe (and a US$200m raise)
One of the hottest trends of 2025 is so-called “vibe coding” – where you create anything from a website to a new app simply by describing it, in natural English, to a GenAI.
The process is so simple – but requires some more involved tools under the bonnet.
One of them is Supabase a start-up co-founded by Christchurch-raised Paul Copplestone and Liverpudlian Ant Wilson, who both ended up at the famed Y Combinator business incubator in Silicon Valley.
The start-up, founded in 2020, offers the Supabase AI Assistant and other tools to wrangle “Postgres”, an open-source database used by millions of software developers and vibe coders to quickly create a feature like real-time subscriptions. Its tagline is “build in a weekend: scale to millions”.
While big database players like Oracle are still firmly ensconced, Copplestone and Wilson’s thinking is that every software revolution – in this case AI and vibe coding – allows for disruption in the underlying database layer.
Venture capitalists agree. In April, Supabase revealed it had raised US$200m, led by Silicon Valley VC firm Accel, just months after it closed a US$80m round.
All told, the 5-year-old start-up has now raised about US$398m, according to a Fortune report.
The latest funds were raised at a US$2 billion valuation – which puts Copplestone in the same super expat class as Flower Labs’ Nic Lane (a former Western Heights High School, Rotorua pupil now developing a new way to train GenAIs in Cambridge); Christchurch-raised Alex Kendall, who co-founded London start-up Wayve (now valued at US$1.7b), which is using AI to develop self-driving technology for multiple brands of car makers; and Dave Ferguson, the Kiwi behind the US$14b Californian driverless delivery start-up Nuro.
Chris Keall is an Auckland-based member of the Herald’s business team. He joined the Herald in 2018 and is the technology editor and a senior business writer.
