Everyone has AI. Not everyone knows what to build.

My dad spent 50 years in construction, and he often said, "Anyone can bang two boards together." At first, that might have seemed dismissive. But what he meant was that the physical act of building was never the hardest part of the job. The real work happened long before the first hammer arrives on site.

The best builders knew how to spot the right opportunity. They understood how to structure a project, manage risk, sequence the work, and even decide what wasn’t worth building in the first place. These folks weren’t valuable because they could swing a hammer. They were valuable because they consistently made good decisions.

In contrast, think about AI’s effect on knowledge work today. Suddenly, almost anyone can generate a presentation, write a proposal, produce a product roadmap, or build a working application, in a fraction of the time it used to take. Looking at the finished product, you might assume that whatever made this work “hard” in the past has… poof!... disappeared.

It hasn't.

The visible work has become easier. The work that actually creates value remains the same.

Knowing what to build, what to ignore, which opportunities deserve investment, and whether something solves a real customer problem still requires experience, context, and judgment. Right now those decisions are becoming even more important specifically because AI has removed so much of the effort involved in execution.

The business implication: More output doesn't automatically create more value

You’ve dramatically increased your output. Now what? “Output” has never been a business objective.

• Your customers don't care how many features you released this quarter. They care whether the product you offer solves their problem.
• Your leadership team doesn't care how many story points were completed. They care whether your product releases increased adoption, improved retention, or generated new revenue.
• Your board certainly doesn't care how many lines of code your developers produced if the business isn't seeing a return on that investment.

The question organizations should ask has changed from "How long will it take to build this?" to "Should we build this at all?" That's a much harder question to answer, and it's where competitive advantage is shifting.

Our recent research on AI in product development found that 75% of product leaders say following through on product strategy is a major barrier to success. Faster execution isn't solving that problem. If anything, AI makes the gap between strategy and execution even more costly because teams can move quickly in the wrong direction. Organizations need operating models that reward learning as much as delivery. Teams need to validate assumptions, gather customer feedback, measure business outcomes, and adjust quickly when the data says an idea isn't working.

Without those feedback loops, AI doesn't make organizations smarter. It just helps them execute bad priorities faster.

Preserve the friction that improves decisions

One of AI's greatest strengths is that it removes friction. Routine tasks take less time. Teams can explore more ideas, build prototypes faster, and move from concept to execution in a fraction of the time. Usually, that's a good thing. The danger comes when organizations remove the wrong kind of friction.

Think about how an idea used to make its way through an organization.

• A product manager would challenge whether customers actually needed it.
• An architect would point out technical tradeoffs.
• An experienced engineer would say, "We tried something similar a few years ago, and here's what happened."
• Customer success would explain how clients were likely to react.
• Marketing might ask whether the value proposition was clear enough to resonate.

Those conversations might have felt like obstacles, but they were serving as an important form of quality control. They forced each team to defend its assumptions, consider alternatives, and improve ideas before investing time and money in building them.

Today, it's possible for a leader to sit down with an AI assistant and walk away a few hours later with a polished strategy, a business case, a product roadmap, and an implementation plan. Everything looks complete. But a lot is missing if no one challenged the assumptions, asked whether the problem was worth solving, pointed out the downstream consequences, or shared lessons from past attempts.

AI is designed to be helpful. Your most experienced people might not feel helpful in the same way. Sometimes they're skeptical. Sometimes they push back. Sometimes they tell you your favorite idea isn't the right one. It might not be what you want to hear, but often it’s what you need to hear. That’s expertise worth investing in.

Organizations should absolutely use AI to eliminate administrative work and accelerate execution. But they shouldn't eliminate the hard conversations that improve decisions.

Building an operating model for the AI era

As AI makes execution faster and less expensive, organizations need a different way of working. Instead of optimizing for output, they need to optimize for learning.

That starts by asking better questions before you start banging boards together.

• What customer problem are we trying to solve?
• How will we know if this initiative is successful?
• What assumptions are we making, and how can we test them quickly?
• What evidence would tell us we're on the wrong path?

Just as important, organizations need to build tighter feedback loops into the way they work. Every feature, experiment, or product release should be grounded in good data, and not just create new functionality for its own sake. Customer adoption, business impact, and real-world outcomes should shape the next decision instead of waiting until the end of a project to determine whether the investment paid off.

Your most experienced people need to be setting direction, challenging assumptions, identifying risks, interpreting results, and helping teams decide what to do next. Those are the decisions that determine whether faster execution translates into meaningful business outcomes.

So no, you don't need another AI proof-of-concept. You've already got plenty of hammers, and they've never swung faster. What you need is what the best builders always had: the judgment to pick the right projects, the experience to spot trouble at foundation level, and a crew that isn't afraid to tell you when the plan is wrong. That's the operating model worth building. We'd love to help you build it.

Benchmark your AI maturity

A better operating model starts with understanding where you are today. Complete our two-minute AI maturity check to benchmark your organization against more than 550 product and technology leaders.