Leaders Ledger #1: Clear, and stuck

During April and May 2026 I sat alongside 53 New Zealand leaders. They represented charities, SMEs, and larger organisations from sectors including construction, accounting, manufacturing, aviation, healthcare, tourism, and the public service. Every one of them different. All of them saying a version of the same sentence: I know we should be doing something with AI. I have no idea what that actually means for us.

Fifty-three leaders, clear on direction but stuck on what to do. Are they behind? Not curious enough? Not brave enough? Not technical enough? None of that fits - these are capable people. They're reading the articles. They're experimenting with the tools. Many have done training, or are about to - internal, online, international. Most have pilots and projects already running.

Several are further ahead personally than their organisations are anywhere near. The capable individual, the system that hasn't caught up. The Microsoft 2026 Work Trend Index calls it the "Transformation Paradox" and puts numbers on it: organisational factors (culture, manager support, talent practices) account for roughly twice as much of AI’s real impact - 67% - as individual mindset and behaviour at 32%. 

This study names alignment and incentives as the culprits. My own two observations are structural too and both link back to incentives.

The increasing importance of language 

Now that we've got a house full of teenagers, we live in a tangle of group chats, WhatsApp, Instagram, the lot. In-jokes, nicknames, the same three memories, the things we each know the others find funny. A family's shared language.

NZ senior leadership teams mostly don't have one, not for AI. One GM is fluent, two or three are curious, the rest are quietly hoping it's a fad. Where there is a vocabulary, it's usually been set by a digital or IT leader: deployment, implementation, models. It leaves out the conversation that actually matters at that table, workforce change, engagement, where this sits against strategy.

Setting that language is the SLT's job, deliberately. Not the frontier labs'. Not so tightly defined that no one has room to be curious. And definitely not outsourced to consultants, AI educators, or a vendor selling the same glossary to everyone. You know your organisation better than any of them.

Incentives not fit for purpose

In most NZ organisations, KPIs, short-term incentives and share options still reward last year's job, done well. As Microsoft puts it: “the job of every leader now is to make change stick - which means making the metrics, incentives and expectations reward people for changing how they work, not for doing the old work efficiently.”

Get this wrong and you get KPMG, who set a 75% AI-usage target for their team. That counts activity, not judgement. Using AI 75% of the time tells you nothing about whether it's being used well.

People do what they're paid to do. So incentives have to move first, from the top. This is why in Ethan Mollick's leadership/lab/crowd model, nothing in the lab or the crowd shifts until leadership rewires what it rewards. Point the incentives where you want the company to go, and the rest follows.

This isn't only a NZ story. Microsoft’s WTI found just one in four people using AI at work say their leadership is clearly and consistently aligned on it.

The best essays on this, Jack Dorsey and Roelof Botha’s “From Hierarchy to Intelligence” are about rewiring the whole company around intelligence. But look at who they're written about: founder-led, board-aligned businesses with the authority to burn it down and rebuild.

The 53 leaders (give or take a couple) don't have that. They have a day job, a sceptical peer or two muttering this is just the internet and a board still circling privacy, risk and data security. So, what can you actually do? Today. 

In our family we talk about Peter Blake a lot. I was fifteen, let out of school for the day to watch the parade down Lambton Quay when we won the America's Cup. At the time I didn't appreciate the discipline of the man. The whole campaign came back to one relentless question: “will it make the boat go faster?” The genius was that everyone could ask it: the designer, the sailmaker, the bowman, the shore crew. One question, the entire team aligned behind it.

So, if you are a leader grappling with AI, what will make your waka go faster? Below are the four things I saw across all 53 that do.

Stop. 

Don’t start with tools. This isn’t a technology problem - not digital transformation, not a rollout, not a platform, not a vendor, not something to outsource.  Boston Consulting Group's 10/20/70 rule has the proportions about right. Long-term value is 10% algorithms, 20% technology and data, and 70% people and processes. The tools are the 10%.

So start with the 70%. What's your position? Your stance as a leader first, then as an organisation. (I wrote about stance here.) Use it to decide what comes next and, just as much, what you deliberately say no to. That's how to make the waka go faster before you've touched a tool.

Build. 

I grew up on a farm, and everyone I knew in the mighty Manawatu could build something out of nothing. It's in the DNA. The leaders pulling ahead now are the builders. It doesn't have to be grand: an AI agent at home, a side hustle, something to wrangle your personal finances, a wardrobe organiser (connect it to your shopping receipts and see what happens).

Building beats consuming, and there's evidence for it: in Mollick's Anthropic programmer study, the people who handed the whole task to the AI couldn't afterwards explain what it had done, while those who used it for part of the work, or made it show its reasoning, kept their own capability.

Build something small and real and it changes you. You learn iteration. You get a feel for what will actually make the waka go faster. And you might fail, which is the point; failing on something small is how the instinct gets built.

Own  

The nervous-lawyer version of this conversation imagines the machine making the call. The leaders making their waka go faster do the opposite. They use AI to augment themselves. They're better prepared. They're sharper in the room. They've got counter-arguments at their fingertips. They know exactly who's in charge, and what the AI is actually good for.

The risk when you don't is what Wharton researchers call "cognitive surrender." In another study by Mollick and colleagues at Harvard and MIT, elite consultants who outperformed on most tasks were more likely to be wrong when the AI was wrong. Handed an authoritative-looking answer that was false, they didn't catch it. They had the same impulse to defer.

Owning it means staying the one who decides. Let AI draft, research, and argue both sides. The judgement, and the accountability for it, stays with you. That's the line the leaders pulling ahead never cross.

Show 

Find something that genuinely works, that delivers real value and show it. Talking about strategy tends to harden the sceptics into their positions; a working demonstration does the opposite. It's hard to argue with something running in front of you. And showing needs no permission, no mandate, no sign-off, no business case. You just show.

There's half a century of social psychology under this. Fazio and Zanna found that attitudes formed through direct experience, seeing it, watching it work, are held more strongly and predict what people actually do far better than the same attitudes formed second-hand, from being told. A demonstration doesn't just win the moment. The conviction it builds is more durable, and more likely to change behaviour, than any argument about strategy.

Some people will never get on the waka. That's fine. Accept it and move on. Your job is the ones who'll paddle.

So: 53 leaders, and plenty of them told me a mandate would be nice. But I don’t believe it is permission that stands in the way.

You don't need a mandate to stop, to build something with AI, to own where you stepped in, or to show what you made.

A clear leader used to be the one with the answer, and the standing to act on it.

The ones I'd back now don't have the answer. They have better questions. They've given themselves permission to lead with curiosity instead of certainty. They've built something real, they keep a list of what they've refused to do, and they have something they can put on the table and point at.

So…  

What did you stop?

What did you build?

What did you own?

What did you show?