Traits Before Tools: What Barry O'Reilly Taught Us About Adopting AI in the Right Order
This week we sat down with Barry O'Reilly — author of Unlearn and Lean Enterprise, and most recently Artificial Organizations — for a live working session on a question we hear in almost every workshop we run: why does so much AI adoption produce so little change?
Barry's answer focused on the sequence most companies and leaders use when new tools and technology become available. Most leaders adopt AI in exactly the wrong order. They start with the tools — download the apps, run the training, mandate the usage and then wonder why nothing important is different. The problem isn't the tools. It's that when you start with tools before you understand how you do your best work, you end up automating the wrong things.
The right order: Traits, then Tasks, then Tools
Barry's model reverses the usual sequence.
Traits first. Start with how you do your best work. Identify the ways of working that make you effective as a leader. These are the things worth amplifying with AI, and they're also the things you can accidentally automate away if you don’t explicitly name them. Most leaders have never made this list.
Tasks second. With your traits named, look at where your time actually goes and separate the high-leverage work from the busywork. AI pointed at busywork buys you back time. AI pointed at your highest-leverage work will often cost you the thing that made you effective. Sorting through these tasks first is the crucial activity here.
Tools last, on purpose. Only now do you choose tools and not because they are the shiny new thing you read about in HBR. You select the right tools (AI included) because they amplify a named trait or absorb a named task from your lists.
If this argument sounds familiar, it should. It's the same finding we wrote about last week from Melissa Perri and Product Circle's survey data: 87.7% of product teams have adopted AI, and only 36% are better for it, because the tools were directly applied to tasks without considering which ones actually made sense to automate. Barry's model is the individual-leader version of the same correction. He suggests that you start with the work, not the tech.
Watch the full session
We recorded the whole conversation — the model, the examples, and the Q&A. Watch the recording here.
Try this before you open another AI tool
You don't need the recording to start. Take fifteen minutes this week and write two lists. First: the three ways of working that most make you effective. These should be the things colleagues actually come to you for. Second: the three tasks that ate most of your last week. Then ask, for each task: would automating this amplify anything on the first list, or just make me faster at things that don't matter?
That's the traits-before-tools sequence at individual scale.
Watch the full session with Barry O'Reilly here — and if this is the conversation your leadership team needs to have together, it's also what our work is built around.