IT as HR for AI agents? A good start. Now give every agent a manager.

Put to the Check takes a widely quoted idea about business and AI and runs it through the six questions of the ERMODA Check. We look at what it gets right, what it leaves open, and the steps you can take.

The claim

At CES in January 2025, NVIDIA’s Jensen Huang said: “In a lot of ways, the IT department of every company is going to be the HR department of AI agents in the future.” (Fortune)

What he got right

It’s a useful way to think about agents. They need onboarding, access, training data, updates and, eventually, retirement. IT already does most of that for software and people’s accounts. Treating an agent like a new member of staff, rather than a feature, is the right instinct.

Put to the Check

E · Expectation: Evidence needed. Treating agents as staff sets an expectation, but not a specific one. Each agent still needs a written answer to a plain question: what is it here to deliver?

R · Right people: Gap found. IT can decide what an agent can technically do. It can’t decide on its own what the agent may promise a customer, a supplier or a colleague. The people who own that part of the business have to agree.

M · Map: Met. IT is well placed to map which systems an agent touches and where a handoff could fail. This is the strongest part of the idea.

O · Owner: Gap found. HR hires people and supports them. It doesn’t manage their daily work. Every employee also has a manager, and every agent needs one too. Today that’s rare: in a 2026 survey of 750 senior technology leaders, 7.2% of organisations had a named person formally accountable for their agents’ behaviour.

D · Done: Evidence needed. What does good work from this agent look like, and how would you know? Without an agreed answer, you can’t tell a useful agent from a busy one.

A · Assured: Evidence needed. When an agent finishes something for a customer or a colleague, does that person get confirmation, and does anyone check?

The missing step

Keep IT as the HR department. Add a manager. That’s a named person in the business who agrees what the agent is for, reviews its work and answers for what it promised. It’s simple, and it doesn’t need a new platform.

Your move

If you run your own business

1. List every tool that acts or replies for you without a person reading it first. For most small businesses that’s two or three.
2. Put your name, or one team member’s, next to each one.
3. Once a week, spend fifteen minutes reading what each tool did and said. Keep what works. Turn off what doesn’t.

If you lead a larger firm

1. Give every agent that acts externally a business owner as well as a technical owner, and record both.
2. Write one line per agent: what it’s for, and what it may and may not commit to.
3. Use the systems you already run to prove it. Log each agent’s commitments where the team can see them, and review a sample each month against that one line.

AI agents can take real work off your team. Give them the same thing you’d give any good new hire: a clear job, a manager and a way to show the work was done.