Our executive coaching team has spent a combined 40+ years in the room with CEOs and their leadership teams—through setbacks, transitions, big wins, and hard calls. We’re starting a new column to pass along some of what we’ve learned, one question at a time. You send us what you’re wrestling with; we answer the ones we hear most frequently.
For BCG’s first In the Room column, I picked the question I get more often than any other right now. It comes out a little differently each time, but it’s always some version of this:
My board is pushing on AI. I know we need to move. But my team isn’t ready, and I’m worried we’ll rush into decisions that backfire. How do I balance urgency with responsibility?
If that situation sounds familiar, it’s because C-suites everywhere are grappling with it. It’s important to recognize that both you and your board are right.
Your board is right to push. With earlier tech shifts, like moving to the cloud, you could afford to be the 25th company to adopt and still be fine. AI doesn’t work that way. A wait-and-see approach isn’t the cautious choice; it’s the risky one. So trust the instinct to move.
Your worry is also right. Moving fast and moving recklessly aren’t the same thing, and the job for leadership today is to know the difference in real time. Here’s how I’d think about it, in three steps:
1. Don’t think of the issue as a technology problem. In any AI journey, technology is only 10% of the equation, and data is another 20%. But most of the work, 70%, is people work.
When people on your team say they “aren’t ready,” don’t hear that only as resistance. Often, what lies beneath it is fear. They read the same headlines you do. Some are wondering what AI means for their jobs, their judgment, and the value of the expertise they have built over years.
Start by saying the true thing out loud. The goal is not to replace their expertise. The goal is to extend it. But you also need to be honest about the fact that the people who learn to work with AI will have an advantage over those who don’t.
That means giving people practical, low-risk ways to build familiarity. Put AI into the flow of real work—preparing for a customer conversation, summarizing complex inputs, drafting a first version of a recommendation, pressure-testing a decision, or identifying patterns in feedback. The point is to help people experience AI as something they can use, not just something being done to them.
2. Don’t try to boil the ocean. The quickest way to fail and burn out your team is to implement AI in 50 places at once. To move swiftly but responsibly, pick two or three problems that matter most to your enterprise, and point AI straight at them. If your biggest issue is customer churn, start there.
And don’t make cost-cutting and headcount reduction the primary goal. The real prize AI can deliver is the work people can’t do at all: screening 100 million molecular combinations to find a new drug or material, answering a customer's hardest question in 30 minutes instead of a week, or settling the claim on a wrecked home or car in hours instead of weeks. That’s what I call aperture expansion. When you make that your aim, the efficiency and cost savings come along on their own.
3. Redefine what success looks like for your team. You can’t expect perfection. AI is statistical, not deterministic, and it will get things wrong. So do people, but we tend to accept a 7% human error rate while demanding 100% perfection from machines. Build a team culture that treats a 70% success rate as a win and mines the 30% failure rate for what it teaches. When an AI experiment fails, say so without hesitation, kill it, run a blameless post mortem, and move on to the next idea.
Just be deliberate about where you start. Don’t dive into regulatory reporting or any other place where a wrong answer could touch someone’s safety. Instead, begin with a process where being mostly right is genuinely useful, where a human can stay in the loop, and where the organization can learn without taking on unacceptable risk.
One last thing, since I know how lonely making these calls can feel: you are not going to get it exactly right, and you’re not supposed to. The job isn’t to have the perfect answer before you move. The job is to move, stay honest about what’s working, and keep your people with you while you figure it out.
