1-2-1: AI didn't level the field

25 Aug 2026 — 1 min read

Two leaders over the last few weeks used the same model for the same kind of task. One produced a working system. The other produced a polished version of what they were already doing.

Both of them thought they were "using AI". Only one was building with it.

1 Principle

The skill is no longer access to the tool. It's understanding what you're building with it.

Access used to mean "do you have the tool?". Now everyone has the tool. The new variable is comprehension. It's slower to compound, harder to fake, and the only thing that matters in twelve months.

2 Things Top of Mind

#1 The two leaders I keep comparing

I'll describe two leaders I've worked with this year. Both senior, both at companies around the same size, both using the same Claude subscription. Twelve months in, here's what each of them has. The first has a Personal OS: agents that run her briefings, draft her replies, surface what matters, prep her for meetings. She's compounded. The second has used Claude every day for the same period. He has a folder of impressive one-off outputs. No system. No compounding. Same tool. Different planet.

This week: Pick one workflow you've been using AI for. Write down what it looked like a year ago, and what it looks like now. If the only difference is speed, you're on the wrong side of the gap.

#2 Twelve months in, same work

Your team has been using AI tools for a year. They're more efficient now. Drafts come back faster. Meeting notes write themselves. Code reviews land in minutes. The throughput is up. Then someone asks what's actually different about the work. The roadmap looks the same. The decisions take just as long. The hard problems are still hard. Faster outputs have not produced different outcomes. The team learned to use the tool. They didn't learn to build with it.

This week: Find one thing your team does now that you couldn't do at all a year ago. Not "faster". Not "more polished". New. If you can't name one, the team has learned to use AI. They haven't learned to build with it yet.

1 Question

What's the one thing you can do with AI today that you couldn't do a year ago?

1-2-1

Liam Darmody

I’m a leadership coach and product advisor. I work with senior product and engineering leaders during periods of transition and growth, helping them regain clarity, align teams, and perform sustainably under pressure.

1-2-1: Time saved is the wrong number

A sentence I keep hearing from founders this year is "I need two more people and I can't afford them." You've probably tried to close a gap like that with AI. You may have a number for how much time it saves each week.

22 Sep 2026

Paid Members Public

1-2-1: Month six is the real test

I found out this week that my own AI team has been making recommendations off numbers that stopped being true in February. Nothing was broken. Everything was running. That was the problem. 1 Principle Systems don't fail at launch. They fail around month six, when the novelty ends

8 Sep 2026