The biggest constraint on what AI delivers inside a company is rarely the technology. It is the leader’s own thinking, and I learned that by watching it happen to me.
The realisation that would not go away
I run my own operation on AI agents every working day. More than half of my leadership team is AI, working around the clock, and somewhere in the first weeks of running it that way, a realisation landed that I have not been able to shake since: my thinking is the limiting factor in how I use AI agents.
I have never had access to resources this powerful, that do so much, so fast. And still the results were smaller than they should have been. Not because the AI underdelivered, but because my requests were too small. I was under-asking when the AI could over-deliver.
That is an uncomfortable sentence to write about yourself. It is also the most useful diagnostic I can offer any leader right now, because the same pattern sits inside almost every company I meet.
The habit that caps everything
Leaders scope AI work the way they scope human work: one task, one deliverable, one deadline. The habit is invisible because it has been correct for an entire career. With AI agents, it quietly caps everything. You get back exactly what you asked for, and you never discover what you could have asked for.
Scale that habit up and you get the pattern I have written about before: budgets released without anyone saying where to dig. The company version of under-asking is spending on AI without ever demanding what the technology is actually capable of returning. The constraint travels upward intact, from how one leader phrases one request to how a whole organisation frames its entire AI investment.
Meanwhile, somewhere below in the organisation, your best people are already asking more of AI than you are, usually without telling anyone. The gap between what leadership requests and what the technology can deliver is being discovered inside your company right now. The only question is whether the leadership is part of that discovery.
Becoming a different type of leader
Working with agents forced me to become a different type of leader. New framing of communication. New scoping of time and deliverables. A new sense of what one request is allowed to contain.
The nearest picture I can give you is this: imagine a hungry, brilliant graduate who can duplicate themselves in a moment whenever the work demands it. Now imagine briefing that graduate with the same care you brief a tired department. That is the gap, and it belongs to the leader, not the tool. Leading AI agents is a different job at a different altitude, and nobody arrives in it fluent.
Where the quieter work starts
So before anyone in your company announces that agents are running anything, there is quieter work to do first, and it starts at the top: each member of the leadership group takes on one small AI project of their own.
This is not something an online course installs in you. You grow into it. You mature into it, as a person and as a leader. The leaders who do this stop asking whether AI works. They start asking what they have been failing to ask of it. That is the moment the whole conversation inside a company changes.
Frequently asked questions
Why do AI projects underdeliver even when the technology is good?
Because the requests are too small. Leaders scope AI work the way they scope human work, one task and one deliverable at a time, so the output is capped by the ask rather than by the technology’s actual capability.
What does under-asking AI mean?
Under-asking is briefing AI with requests sized for human constraints. You get back exactly what you asked for and never discover what you could have asked for, which makes the results look like a technology limit when they are a leadership habit.
What new skills does leading AI agents require?
A new framing of communication, a new scoping of time and deliverables, and a new sense of what a single request is allowed to contain. It is a different job at a different altitude, and leaders grow into it rather than learn it from a course.
How should a leadership team start with AI agents?
Each member of the leadership group takes on one small AI project of their own before the company announces anything larger. The personal experience is what produces the shift in how each leader asks, decides and delegates.
If you are leading your organisation through this, this is the kind of work I do with a limited number of senior leaders each quarter: The Reset.
Thomas Anglero is a Strategic AI Advisor, keynote speaker and author of Intro to Artificial Intelligence. He has delivered over 450 keynotes across 30 countries for organisations including IBM, the WHO, the World Government Summit and the European Commission. He founded the IBM Watson AI Lab for Cancer at the Oslo Cancer Cluster and closed over $500 million in enterprise transformation deals as CTO and Chief Innovation Officer at Cognizant.
Listen to the Clarity at the Top podcast on Spotify. Enquiries: Anglero.com