Start 2027 by boosting your chances of success in the CEO role.
Learn more about our CEO Masterclass and apply to attend.
Taught in person by me at Texas CEO Ranch.
An executive told me yesterday that the hottest topic among CEOs right now is one question: What is your AI strategy?
It sounds like a smart question, but I answered it with another one. What is your people strategy? I wasn’t changing the subject. I was asking the question that I think comes before any answer to “AI strategy” is possible.
I have argued here before that AI is not a new technology in the sense that software was. Software is a tool. It is built to do a specific job. You buy it, install it and train people to use it.
AI is a different beast. It is a new way to get access to intelligence. Until now, the only way a company could get intelligence was to hire people.
That changes what the strategy question means. If AI were software, an “AI strategy” would be a technology plan: choose the vendors, fund the projects, roll it out. But because AI is intelligence, the real question is how your company puts intelligence to work.
You already have an answer to that. It’s the way you manage people today.
AI amplifies the management system that already exists
Before you draw up a plan to deploy hundreds more AI agents into your company, consider how well you’re managing the hundreds of intelligent human beings already working there. Do they understand what they’re supposed to accomplish? Are they getting the direction, information and coaching they need? If your organization struggles to make effective use of human intelligence, introducing artificial intelligence will hardly solve the problem.
AI amplifies whatever management system already exists. Without one, it accelerates work in the wrong direction.
There is some interesting research supporting this connection. In a 2025 Harvard study, effectiveness leading AI agents was closely correlated with effectiveness leading human teams (a correlation of 0.81). If you’re good at managing people, you’re likely to be good at managing agents as well.
AI amplifies whatever management system already exists. Without one, it accelerates work in the wrong direction.
I’ve written before about what AI needs to work well inside a company. Before you decide to push out the next level of your AI strategy, ask yourself and your executive team six questions. Then see whether the answers hold up two levels down:
Does everyone understand how we intend to win? Can they explain the strategy, quarterly priorities and tradeoffs? Would sales and product agree?
Do we have the right people? Are they in the right roles, meeting a high talent bar and motivated? AI magnifies their judgment.
Does everyone know what they own? Are responsibilities, decision rights and measures of success explicit? Do people know when to decide and when to escalate?
Are managers actually coaching? Do people receive frequent, candid feedback? Do they trust leadership enough to raise concerns? Who reviews AI-assisted work?
Is essential information available? Are goals, decisions and customer commitments recorded and current? Can AI access appropriate company context securely?
Can departments work as one company? Who resolves conflicting priorities? Do decisions reach the people who must carry them out?
These conditions determine whether intelligence can accomplish something worthwhile.
Jack’s AI strategy
Imagine Jack, CEO of Diamond Technologies, a 300-person software company. Jack has never established a coherent way to set priorities and hold his departments accountable to them. Sales promises features to close deals. Product has orders to reduce custom development. Customer success hears about it from angry customers. Jack keeps asking executives to communicate better.
Then the board begins asking about AI. Eager to show progress, Jack rolls out agents across the company. Sales generates proposals faster. Product automatically reprioritizes its roadmap. Customer success sends polished updates drawn from records nobody has reconciled.
The departments were already working at cross-purposes. Now they’re doing it at machine speed.
Coordination gets harder with each new participant. Ten people have 45 possible one-to-one relationships; a hundred have 4,950. Add agents across departments, and unresolved conflicts become an ever-larger network of expensive confusion.
The company needs an organizing principle
A company has to know what it stands for and where it is going, even when the CEO isn’t physically present. Otherwise every smart employee, and every AI agent, is left to invent those answers from whatever instructions happen to be closest.
This is the purpose of a companywide operating system. It makes the mission and strategy explicit, establishes values and decision rights, translates objectives into measurable goals, and creates a rhythm for reviewing results and coaching people. It provides boundaries without dictating every move. People will discover opportunities nobody anticipated. AI will too. Both need a common framework for deciding which opportunities are worth pursuing.
Over 30 years as a CEO and working alongside other CEOs, I’ve watched erratic performance become far more consistent once even a simple system was in place. Leaders could see problems developing, know who owned them and act while outcomes could still be changed.
So begin your AI strategy with the people strategy. Make sure you have excellent people who understand the mission, trust their leaders, know what’s expected and get the coaching they need. Make that operating context explicit enough for AI to work from it, too.
Then AI can advance the business instead of multiplying its contradictions.
