A carpenter once told me this after I’d messed up a very basic renovation task in my kitchen: Measure twice, cut once — because wood doesn’t grow back. Once you make the cut, the piece is either right or wrong. You can’t undo it. You can only start again with a new piece of wood.
I think about that rule a lot right now as I watch leaders make decisions about AI. In my work as Chief Human Success Officer at Zensai, where we help organizations create better business outcomes by enabling people to thrive in the age of AI, I see the same conversation playing out everywhere. AI is here. It’s faster and cheaper. And, in the most common business logic out there, that means people become the variable to reduce. Cut first. Understand later. Announce the layoff on Monday, work out the operating model on Tuesday.
This isn’t transformation. It’s surgery with a lawnmower.
Cutting people before you understand the work is exactly the same as cutting the wood before you’ve measured. It looks decisive. It sounds bold on an earnings call. But once you’ve done it, you can’t get the piece back. You lose the knowledge, the relationships, the trust, and the customer intuition that were sitting inside those roles. And you rarely find out what you actually lost until months later — when numbers start to drop or execution starts to slip and no one can explain why.
I’m honestly not against efficiency. I’m not against reducing costs either. In most companies I’ve worked with every euro has had to work hard. What I am against is making irreversible decisions with almost no measurement.
So let me describe what measuring twice actually looks like in the age of AI.
- The first measurement is the workflow — not the org chart. The workflow is how work actually gets done. It’s where decisions are made, where value is created, where the bottlenecks are and where the handoffs happen. This sounds obvious, but most leadership teams can’t draw their own value flows on a whiteboard. They can draw the reporting structure, though. Which is a completely different thing.
- The second measurement is the human piece. What are people actually doing all day? What’s high value and what’s low value? Where is judgment needed? Where is empathy key? Where is pattern recognition essential? And where is it routine work that AI can genuinely support?
Only when you have both of these measurements can you start making sensible decisions about AI. This is the sequence I stand behind. Map workflows. Clarify who decides what. Define what stays human. Then, and only then, automate.
If you skip the measurements, it will come back and bite you. AI can do many things but it can’t own judgment and accountability. AI can’t rebuild the trust of a customer whose account manager was replaced by a chatbot that looped endlessly on useless replies. AI can’t coach a struggling employee back to performance. AI can’t walk into a boardroom and take responsibility for a bad decision. Those things are human.
This is where the carpenter rule matters most. You don’t measure twice because you’re afraid of cutting. You measure twice because you respect the cost of being wrong. AI decisions work the same way. Once you remove a role, collapse a team, automate a customer touchpoint, or replace a manager conversation with a generated message, you’ve changed more than a process. You’ve changed the trust, memory, and accountability inside the system.
The organizations that get this right aren’t the ones that move fastest. They’re the ones that measure before they cut.
Which brings me to the concept I care most about. Re-engineering work around value creation.
Every organization has activities that create value for the customer and activities that support them. AI can help with both. But the interesting question isn’t which people can we remove. The interesting question is which parts of the value chain should we redesign entirely — and how do humans and AI create more value together in that new design.
That’s a completely different conversation. It requires business, technology, and people leadership in the same room, looking at the same number. Not a headcount number — a value number. Revenue per employee. EBITDA per employee. Value per team. Time to customer outcome.
When those become the conversation, layoffs stop being the default lever. Because the goal is no longer subtraction. The goal is capacity. And capacity is what compounds.
So if you’re sitting with an AI transformation decision this quarter, ask yourself before you sign off on it: Have we measured? Or are we just cutting?
Because in AI, as in carpentry, the piece won’t grow back. It’s the conviction that led me to pioneer the Chief Human Success Officer role — and, together with Zensai CEO Rasmus Holst, to write The Human Success Playbook: A Leadership Tale for the Age of AI, a book about leading people, not just deploying technology, when the stakes can’t be undone. You can find it here.
