My position on AI, v1.

On a Mt Cook climb, the lead climber doesn't take the risk for the team. They take it before the team. The anchor placed on the next ledge, the line read, the rope tensioned, each is a position the next climber moves through. Without it, a slip becomes a fall. Each position is load-bearing.

I sit alongside leaders, students learning to build with AI, and institutions designing what comes next. From that seat, here's what I've watched happen to my own work.

In 2025 I taught tools, context, prompt engineering. Later I helped companies build with AI. In Q1 2026 I ran diagnostics. Today I do none of those things. In the same window I retired "calculated bets" as my headline, the OS diagnostic, the Tier 1/2/3 architecture, the workflow-redesign retainer. Demoted CMR to internal IP.

That's a lot of subtraction in a short window, and the pattern is hard to miss: everything I tried to teach, every framework, every product got passed by the field within months. The only thing that didn't expire was my position.

 AI can't do your position for you.

It can do almost everything else. It can plan, draft, analyse, review, iterate. It can absorb the work that doesn't compound. What it cannot absorb is what you stand for in an AI-shaped world, what you'll refuse to do, what you'll hold under pressure, what you're willing to bet on. That's the position.

Most leaders are treating AI as a productivity layer, copilots stapled onto the work that already exists. That's amplification, and it's the wrong move when the work itself is what needs to change. The right move is redistribution: let AI absorb the work that doesn't compound, so you can over-invest in the work that does. And the work that compounds is the work of holding a position.

So how do you hold one when the field moves faster than you can read?

You compound and subtract. Compound what's working. Subtract everything else, fast.

Two weeks ago an opportunity surfaced inside an existing partnership, a second programme, warm relationships, a fast yes available before the first value had even landed. The old me would have scoped it, sent an NDA, talked price. I said no.

A workshop last month was headlined "Calculated bets in the age of AI", language I used to love. By April it had stopped meaning what it meant in March. So I cut the headline, kept the substrate, and rebuilt the talk in an hour.

Same practice, two scales. Compound and subtract.

If a board or CEO asked for my position in three sentences: AI is here. To lead through it, you must have a position. My job is to help leaders make that position explicit and hold it.

This is v1. I expect the verbs to sharpen as more leaders take the move into their own rooms. The frame won't change; the teaching of it will. And here's the test I hold myself to: can my team repeat my stance without me in the room? If not, I've failed.

The lead climber doesn't take the risk so the team doesn't have to. They take it first, so the next climber can move through. AI is the next climber. Your job is the lead.

AI can't do your position for you.

What's your leadership team's position on AI?