A few weeks ago, I came across Tobias Zwingmann’s “AI Carpenters vs AI Gardeners” frameworks, based on developmental psychologist Alison Gopnik’s research on parenting. It finally gave me the language for something I’ve been thinking about and its application to NZ businesses.
I catch myself thinking this way too, and I heard the same thing at an AI leaders lunch last week, with many trying to get AI right at first bat. Pick the perfect tool. Choose the perfect use case. Lock in the perfect vendor. Avoid risk at all costs.
All valid concerns, but in reality they won't work in the AI age.
McKinsey’s 2024 numbers say 91% of employees already use AI tools. Only 1% of organisations consider themselves AI mature (that number is likely higher up now). This gap isn't because businesses can't follow a plan, it's because the plans being followed are the wrong ones.
Why Maturity Models Fall Over in AI
Bring in most AI consultants and they will show you a traditional maturity model.
Crawl: Identify use cases and run pilots.
Walk: Build governance and create a centre of excellence.
Run: Scale across departments and integrate into core processes.
The logic is tidy on paper. In practice, reportedly over 80% of AI projects stall in the prototype phase. MIT found that only 7% of organisations ever reach “maturity” even when they follow these frameworks.
AI value does not appear in neat stages. The legal team might quietly adopt ChatGPT to review contracts and save hundreds of expensive hours while the official AI programme is still polishing its 12-month roadmap. Some times want to sabotage those organic wins simply because they did not come from the top down.
This is the mismatch.
Maturity models expect orderly progression, milestones and central control.
AI success often comes from messy, unpredictable growth.
Carpenters vs Gardeners
Carpenters plan every cut before touching the wood. Measure twice, cut once. It works when mistakes waste expensive material and the environment stays still.
Gardeners think differently. They plant multiple seeds because they know not everything will take. They adapt to changing conditions. They watch for what grows naturally and then help it thrive.
Most NZ businesses are rolling out AI with a carpenter’s mindset. Six-figure roadmaps. Stage 1, Stage 2, Stage 3.
The gardener mindset says start with a theme, not a fixed plan, and nurture whatever is working.
The Three Jobs of an AI Gardener
1. Know and cultivate your ecosystem Your AI climate is shaped by regulation, systems, and culture.
For example: A Marlborough viticulture business experimenting with AI-driven harvest scheduling might quickly realise they need to build in local weather and soil data. That greenhouse was a simple internal data pipeline to feed the AI, not a massive new tech stack.
2. Plant and nurture selectively Some experiments are weeds. Remove them quickly. Others show promise and need help to grow from prototype to production. The most effective way is to find the people already using AI effectively, often without permission, and give them the resources to scale.
For example: A Canterbury manufacturing firm might notice one line manager has set up a Copilot script to cut the time needed for maintenance reporting in half. Instead of shutting it down because it bypassed IT, leadership should give them two developers for a month to turn it into a plant-wide tool.
3. Observe everything Look for the intended value areas, but also for the green shoots you never planted. Understand why they work before you scale them.
For example: A Queenstown tourism operator might discover their social media coordinator is using ChatGPT to translate promotional copy into multiple languages for niche travel markets. That small green shoot could become a multilingual marketing push that increases bookings from Asia and South America.
Two Levels of AI Adoption
Level one is individual productivity. People using ChatGPT or Claude for emails, summaries, quick research. This is where most companies are stuck.
Level two is system transformation. Redesigning processes, redefining roles, and reimagining how the work gets done. This is where the real competitive advantage sits.
Carpenters try to jump straight from one to two with a big plan. Gardeners use level one activity to decide where to replant, re-shape, and scale into level two.
The NZ Opportunity
We will not build the next GPT in Aotearoa. We do not have the capital or the data. That is not the opportunity.
Our strength is in the application layer. Industries like agriculture, tourism, logistics, health and manufacturing have specific needs, lean teams and unique processes. These are perfect conditions for AI gardeners.
The companies that start planting now will find the next Xero-scale successes first.
Measuring Gardening Success
Forget staged maturity. Watch for:
How many business domains you are re-imagining with AI?
How many experiments are sprouting naturally?
Competitive advantages that others cannot copy easily?
How fast you can move from assistant to autopilot?
The Best Way To Find Opportunity For NZ Leaders
Walk the floor and find the pockets where AI is already working. Provide support, and then scale.
AI in New Zealand will not be won by the perfect plan. It will be won by the organisations willing to get their hands dirty and grow what works. In this game, speed is the advantage. Gardeners move faster.
Written by Mike ✌
Passionate about all things AI, emerging tech and start-ups, Mike is the Founder of The AI Corner.
Subscribe to The AI Corner
