5 November 2026
The three levels
Over the last year I’ve sat across the table from a lot of business owners talking about AI. The conversations differ in the detail, but the businesses themselves land in one of three places. I described these levels to a potential partner to our business last week, and writing them down seemed the obvious next step, because almost every owner I meet recognises themselves in one of them straight away.
Level one is terrified. These owners feel that using AI would be cheating, or at least doing something untoward, so they have chosen not to engage with it at all. They’re in denial. Most have been in business a long time, have done things a particular way for all of it, and the change frightens them because of what it might do to the business they built. I understand the instinct. It’s still the wrong call, but at least it’s a consistent position.
Level two is one foot in. One staff member is using ChatGPT, another is on Claude, a third has Copilot because it arrived with Microsoft 365, and half the team is using nothing at all. There are no controls, no AI policy, no criteria for which tools are used or what they can be given, and no record of what client information has already been pasted into them. Level two feels like progress. It’s actually the riskiest place a business can be, because AI without controls can do real damage: to client confidentiality, to the quality of your work, and to your reputation when something goes wrong and you can’t explain how.
Level three is embedded. AI is built into the way the business operates. Staff receive ongoing training. There are criteria for which tools are used and how. Each user's set-up is managed. Someone on staff is responsible for deciding what the AI is connected to. Every licence is at enterprise level, and client security and protection are dealt with properly.
Blomfield is a level three business. We’ve been building AI into our operation for two years, and it took real work to get there. It was also the best operational decision we‘ve made in that time.
The decision
Every owner needs to answer one question, and needs to answer it now: is AI going to be part of how this business operates in the future? If the answer is genuinely no, then stop. Don’t invest, don’t run workshops, don’t buy licences. That’s a legitimate answer for a small number of businesses.
If the answer is yes, then level three is the only option. Level one means falling behind competitors who are already compounding the gains. Level two means carrying the risk of AI without the capability, which is the worst of both. There’s no stable middle position. One foot in is not a strategy, it’s an exposure.
What the big end of town is telling you
If you want an indicator of where this is heading, watch what the large corporates are doing with their money. Genesis Energy has put around $100 million into a digital transformation that its own technology leadership describes as building the base for AI, with the current financial year the peak of that investment. Air New Zealand signed a direct partnership with OpenAI, a New Zealand first, specifically to give its corporate staff secure, enterprise-grade AI tools, and has a five-year AI-led transformation partnership with TCS running alongside it. NZME has announced automation and AI initiatives expected to take roughly $4 million a year out of its cost base from late 2026, with more to follow.
These organisations didn’t arrive at this by accident. They engaged large consultancies and research teams, and every one of those advisers came back with the same answer: AI will be part of how business operates from here. The corporates are spending tens of millions acting on that advice. A small or medium business doesn’t need the same budget, but it does need the same answer to the question.
The expert problem
So why do so many businesses get stuck at level two? A big part of the answer is the quality of the help on offer. There’s currently a flood of people running AI readiness workshops and calling themselves AI consultants, and for a good number of them the entire qualification is six months of playing with ChatGPT. They can show you tricks. They can’t tell you what the tool should be connected to, what your policy needs to cover, or what happens to your client data, because they’ve never had to carry responsibility for any of it.
Real AI consulting is three skills, in a strict order. First, understanding business: how they run, where the money is made, where the risk sits. Second, understanding how technology interacts with a business: what happens when you put a new system into a real operation with real staff and real clients. Third, and only third, the AI itself. The AI is the smallest part.
Anyone who leads with the tool and not with your business has the order backwards.
What the real thing looks like
Kerstin Littin is our Principal Technology and AI Consultant, and she’s what the real thing looks like. She has spent twenty years building and shipping software that governments run their registries on, leading product teams and a portfolio that took a New Zealand software company to $40 million in annual revenue.
More than two million people use the products she launched, including systems for New Zealand's Ministry of Business, Innovation and Employment and the Australian Taxation Office. She set up and ran the product function for the company's Canadian operation in Toronto, moving government legacy systems onto cloud platforms, and she’s run workshops for audiences of more than 150 people. She has taken products from idea to launch, priced them, supported them, and answered for them when things went wrong. That’s business understanding, technology understanding and AI understanding, in the right order.
You can see the difference in how she works. When Yachting New Zealand asked us about AI earlier this year, Kerstin didn’t open with a tool. She spent the first conversation understanding how the organisation operates, who does what, and where the pressure sits. When they asked the question every business asks, Copilot or Claude or ChatGPT, she answered it in plain business terms rather than jargon, and the training she scoped was a measured four-week programme for the team that needed it, not a big-bang transformation with a big-bang invoice. That restraint is itself the tell. People who understand the work scope it properly.
"AI is such a daunting and fast-moving challenge for any organisation, let alone small ones like ours. It's hard to know where to start. Kerstin helped demystify things for us and really focussed on our specific needs (within budget). It's been great watching the team transform in how they use AI."
Raynor Haagh, General Manager Community and Development, Yachting New Zealand
Questions worth asking
If someone is pitching you an AI workshop, ask them three questions before you sign anything. What have you actually built and shipped? What business have you run, or helped run, with your own name attached to the result? And what went wrong on your last technology project? Things always go wrong on technology projects, and the answer tells you immediately whether they’ve ever done a real one. If the answers are thin, keep your money.
And if you want to start with someone who can answer all three, Kerstin leads our AI Ready programme and she’s great at meeting businesses exactly where they are, whether that’s level one, level two, or ready to commit to level three.
You can reach her at kerstin@blomfield.co.nz or on 027 574 1733.