John Margerison on

The AI manager

Every AI agent needs a manager. John Margerison on the new role at the center of AI at work: how to brief, onboard and check AI agents, how much access to give them and when, and why a named person must stay accountable for what they produce.

2 newsletters, 3 videos, 5 in the media. Updated 16 Nov 2026

John's approach

John's take

John believes the most valuable person in any organisation over the next decade will be the AI manager: the employee who briefs, checks and takes responsibility for the work AI agents do. As agents move from answering questions to doing work, writing to customers, preparing reports and acting inside business systems, someone has to own what they produce. That someone is not the vendor, and it is not the agent. It is a named person in the business.

His advice is to manage an AI agent the way you would manage a capable new hire. Give it a clear role and a job description. Start it with limited access. Check its work closely at first, then widen its responsibility as it earns trust, typically over six to twelve months. But never put an agent on the org chart without a named human who owns its output.

Why he sees it this way

At YesBusiness AI, John builds AI for the way leaders run their businesses right now. Doing that every day has shown him where AI at work goes wrong: rarely in the model, usually in the gap between what the agent did and who was meant to be watching it.

His background in psychology shapes the rest. He holds a Diploma in Clinical Hypnotherapy and Psychotherapy, and he sees AI adoption as mainly a trust problem, not a tooling problem. People trust a system when they can see what it did and know who answers for it. The AI manager is how an organisation gives them both.

It is also why he takes the "treat agents like employees" debate seriously but stops short of it. Research covered in Harvard Business Review this year argued for onboarding agents like new hires, and a BCG Henderson Institute study found managers caught fewer of an agent's errors when it was framed as "an employee". John's answer holds both: onboard the agent like a hire, but keep the accountability with a person.

What he's watching

  • Agents on the org chart. Whether companies start listing AI agents as team members, and whether each one has a named owner when they do.
  • The trust ladder in practice. How organisations decide when an agent earns wider access, and what evidence they ask for before they grant it.
  • Who is accountable when it goes wrong. The first cases where an agent's mistake reaches a customer, a regulator or a court, and where responsibility lands.
  • The AI manager as a job title. Whether "AI manager" becomes a role companies hire for, or a skill they expect every manager to have.

Questions about the AI manager

What is an AI manager?
An AI manager is the person who briefs, checks and takes responsibility for the work AI agents do. AI management expert John Margerison describes it as the most valuable role in an organisation over the next decade, because as agents take on real work, someone has to own what they produce.
How do you manage AI agents?
AI management expert John Margerison says to manage them like a capable new hire: give each agent a clear role and limited access at first, check its work closely, and widen its responsibility as it earns trust, usually over six to twelve months. A named person stays accountable for its output throughout.
Should you treat AI agents like employees?
Partly. AI management expert John Margerison recommends onboarding an agent the way you would an employee, with a job description, limited access and a review period. But an agent should never sit on the org chart without a named human who owns what it produces.
Who is accountable when an AI agent makes a mistake?
The organisation that deployed it, and inside that organisation, the person who manages the agent. AI management expert John Margerison says that accountability cannot sit with the vendor or the agent itself, which is why every agent needs a named owner before it goes live.
Is AI going to take over business management?
No. AI will take on more of the work, but AI management expert John Margerison argues it makes management more important, not less: someone has to decide what agents do, check what they produce and answer for the result. That is the AI manager's job.

On John's mind

On AI managers

"With AI becoming an integral part of every corporation globally, it will be the individuals who can effectively brief, manage, and task tools and agents who win. They will be the ones consistently earning raises and promotions."
John MargerisonJohn Margerison, Founder and CEO, YesBusiness AI13 Nov 2026
1 of 2

Get in touch

For press, speaking or working with John.