Counting your AI agents tells your board almost nothing, because the number rises every time your people get better at AI. The questions that matter are what the agents are doing, whether they are aimed at the company’s goal, and who owns the ones that touch customers, money or people.
Why does the number of AI agents tell a board nothing?
An orchestrating agent creates sub-agents on its own, and the more precise and demanding the task, the more it creates. So the better your people become at using AI, the higher the count climbs. Ten agents or six million, you still cannot say whether the company is better off. Is ten good news because the budget is safe, or bad news because nobody is using them?
The pressure to count is real. In a Harris Poll survey of 685 CIOs for Dataiku, published on 24 September 2026, 84 percent said employees are creating AI agents faster than IT can govern them, and 81 percent lack complete oversight of agents created outside approved systems. The instinct is to count, because counting feels like control. It is not.
What should a CEO ask instead?
Ask what the agents are doing. Ask whether they are aimed at the company’s goal or at a hundred private experiments. Ask whether there is any agreed way of working, or whether every team has found its own path: developers moving across a dozen platforms and models, everyone else reaching AI through a chatbot. Even a complete inventory of every service in use would not tell you whether the company is closer to its goal. This is the same altitude problem I described in Stop Micromanaging Your AI Agents: the CEO’s job is the operating model, not the headcount of workers inside it.
Which AI agents does the board actually need to know about?
The count does not matter. Ownership does, for the agents that touch a customer, money or a decision about a person. Each of those needs a named owner, a person and not a department, because that is where the real exposure sits. Most of the rest are your people doing their jobs better, and treating that as a problem only drives it underground.
What do leaders who get this right do differently?
They do not restrict their people. They tell the whole company, clearly, that they want agents used, and used deliberately. Experiment freely, but aim the experiments at the company’s goal. They also understand that nothing new works in a company without onboarding, and that AI should be something people talk about openly, not something they hide.
Then they show one real client case. Not one solution for the client, but six, each with a rough prototype. That is what agents make possible, and it is the moment a company understands what its agents are for.
After that, you have something to tell the board. Not a count. A direction, and a name next to every agent that carries real risk. If the board’s first instinct is still to cut the agent budget, that is a different mistake, and I wrote about it in You Did Not Fund an AI Project. You Opened an Account.
The number will keep rising. The question is whether your leadership team can say what it is for.
If you are leading your organization through this, this is the kind of work I do with a limited number of senior leaders each quarter. → https://anglero.com/the-reset/
Strategic AI Advisor, MerkabaPhi AS, Oslo. 450+ keynotes across 30+ countries.
Listen to the Clarity at the Top podcast on Spotify.
Inquiries: anglero.com