I was having a red wine with a mate the other night — sharp operator who runs a mid-sized strategy consultancy here in Auckland. Good bloke, been in the game fifteen years. He was buzzing because they'd just done something that felt properly disruptive. They built this AI-powered tool that does the first-pass strategy audits for clients — pulls the data, spots the gaps, spits out a draft report. Instead of charging for it like every other firm, they gave the basic version away. Completely free. No strings.
Now their competitors are scrambling. Clients who used to pay big money for the whole package are using the free tool and only coming back when they need the real stuff — the judgement calls, the market context, the "here's what this actually means for your business" conversation. Revenue's up, margins are fatter, and they've turned a bunch of incumbents into roadkill without firing a shot.
He was grinning like a kid who'd just worked out how to pick the lock on the lolly jar. And honestly… I was grinning too. That's a real result.
Then I asked him the question that killed the vibe.
"Where does the money actually stick in five years?"
He went quiet. Properly quiet. The kind of quiet where you can hear the ice clinking in the glasses. He didn't have a clean answer. And the thing is, neither did I — not yet anyway.
The Conversation I Keep Having
That's the conversation I keep having with business owners. Not just consultants — accountants, lawyers, engineers, recruiters, marketers. Everyone's chasing AI adoption like it's the main game. And look, I get it. The productivity gains are real. I've seen them in my own work and across a dozen businesses I'm involved with.
But adoption was yesterday's question. The question that actually matters now is this: which part of the stack do you own, and does anyone else control what sits underneath you?
I've started calling it the Layer Cake. Four layers, stacked bottom to top. You can't just grab the cherry on top. The whole thing has to hold together. And if you're building on someone else's layers without realising it… well, you're renting, not owning. We've all seen how that ends.
Layer 1: The Tool Set
This is the bottom slab. The big models, the GPUs humming away in data centres on the other side of the world. ChatGPT, Claude, Gemini — the raw brainpower anyone can rent by the minute. It's commoditising faster than anything I've seen in business. Costs are heading to zero. Six months after someone drops a fancy new model, someone else makes it cheaper and better.
Think of it like electricity. Essential, yes. But no one's building a lasting business by being the one who generates the power. That game is already over.
Human parallel: You hire a great temp from an agency. Smart, capable, available tomorrow. But so can every other business in Auckland. That's not a competitive advantage — that's a commodity. Layer 1 is the agency.
Layer 2: The Organiser
(Geeks call it the 'orchestration layer')
This is where you wire things together — the AI acting like that brilliant middle manager who assigns the work, chases things up, keeps the whole operation humming without you having to be in every single loop. It spots the exception, routes it to the right person, updates the client, and moves on. No more bottlenecks. No more "I'll get back to you next week."
Right now, if you've got someone who can build this stuff, you've got a real edge. For the next year or so, anyway.
But I've been around long enough to recognise what's coming. Some slick platform will package all of this up and sell it for pocket change. I remember when having a website was a competitive advantage. Then having an app. Then having a CRM that actually worked. Every time, the plumbing becomes cheap and the edge moves up the stack.
Enjoy this layer while it lasts. But don't bet the farm on it.
Human parallel: You hire a brilliant ops manager who redesigns all your systems and processes. Game-changing — for a while. Then every business has one, and the templates are on YouTube. The edge doesn't last.
Layer 3: Your Private Data
Now we're getting somewhere.
This is the stuff OpenAI or Google Gemini can never get their hands on. Your client relationships. The weird institutional knowledge your team has built up over years — the stuff that lives in people's heads, not on a server. The judgement you've earned by being in your specific market, in your specific community, solving your specific problems.
If you're the only one who has it, no one else can ever be as good as you. Full stop.
And here's the bit that really changes how you think: once you start treating this private data as the crown jewels, it completely shifts how you capture information and how you use it. Imagine walking into every meeting with a full client brief already loaded — who they are, what they're like, what everyone else in your firm is doing with them, the insights you can share that they may not have access to. It's not just remembering things. It's weaponising the stuff that makes you you.
The ones that worry me are the businesses where the "private data" is actually just process knowledge — how to fill in a form, how to run a standard workflow. That stuff is getting eaten alive right now. The ones I'm excited about are where the knowledge is relational, contextual, earned through years of being in the room. That's a very different thing.
Lose this layer and the whole cake collapses.
Human parallel: It's the difference between a plumber who knows the trade, and a plumber who's worked on every house in your suburb for twenty years and knows exactly where the dodgy pipes are. Same job on paper. Very different value.
Layer 4: The Business Model
This is the top of the cake. And this is where it gets properly interesting.
Layer 4 isn't about tweaking what you do. It's about fundamentally rethinking how your business works — who you serve, what you actually sell them, and how you charge for it.
In the old world, most of us sold time. Hours. Deliverables. Bums on seats. In the new world, AI handles the grunt work below, so the question becomes: what are you actually being paid for?
My mate with the free strategy tool? He's living this right now. The AI does the first-pass stuff in minutes, so charging three hours for it is dead. Instead, he's moving to a flat annual advisory fee — "we will make sure your business doesn't get blindsided, full stop." Same client, completely different contract, significantly higher margin, and a relationship that's much harder to walk away from.
For others, it's full market disruption. New services that weren't possible before. New client types you couldn't afford to serve because the economics didn't work. Pricing models that would have been lunacy five years ago but now make perfect sense because your cost base has fundamentally changed.
The economics driving this are pretty straightforward. Two curves racing in opposite directions: the cost to do the work is dropping like a stone, but the cost to verify — to stand behind it with your reputation, take the liability, tell the client it's right — stays tied to human trust. That gap is where the money concentrates. This is the layer where you stop competing on doing the work and start owning the value that AI unlocks.
Human parallel: A travel agent who used to book flights and hotels is now useless. But a travel agent who designs entire life-changing experiences for high-net-worth clients — who handles the crisis when it all goes sideways at 2am — is more valuable than ever. Same industry. Completely different business model.
The Traps No One Talks About
Of course, there are traps. And the scariest ones are the ones that feel like progress.
The Missing Junior Loop. You automate the entry-level work because the AI does it faster and cheaper. Makes total sense. But then you stop hiring graduates. And five years later you look around and realise there's no one coming through who understands the business well enough to sit in Layer 4. You've hollowed out the pipeline. The top of the cake disappears because there's no one to stand on it.
I've seen a version of this before, long before AI. Firms that outsourced everything, ran lean, looked brilliant on paper — and then couldn't grow because they had no depth. Same mistake, fancier technology.
The Codifier's Curse. This one is sneaky. Your best people — the ones with all the judgement and hard-won knowledge — they generously train the AI. They codify everything they know. And in doing so, they hand their private data over to someone else's lower layers. The thing that made them irreplaceable becomes a training set. Their moat drains away and they don't even notice until it's gone.
The Window
If you want to survive — or even thrive — Layer 3 and/or Layer 4 is generally your best bet. The window to climb the cake and lock in the top two layers is open right now. It won't stay open forever.
So here's the straight talk. Stop asking whether you should adopt AI. Start asking which layer you want to own — and whether you're building the private data underneath it to make that ownership stick.
In the AI economy, it's not the fastest adopter who wins. It's the one who ends up sitting on top of the cake — and actually owns what's there.
Good luck out there. And don't forget to have fun!
