Your AI Confidence Is Not a Plan
Here is the uncomfortable verdict for this week: enterprises are buying AI far faster than they are rebuilding the work around it, and that gap is where the money disappears. Confidence in the technology is not a strategy, and this week’s news shows how many leaders are confusing the two.
The evidence is not subtle. Gartner found that 63% of procurement leaders expect AI to significantly boost performance over the next three years, but only 25% have redesigned jobs and roles in response to AI implementation. That is expectation running three times ahead of action. McKinsey makes the same point from the other direction: agentic AI, systems that can act autonomously, requires a deeper shift than the first wave of tools, and the upside comes from redesigning how work flows through the organization, not from bolting AI onto today’s processes.
The alignment problem starts at the top
You cannot redesign work if the people who run it disagree about what AI does. Gartner found that 38% of chief human resources officers and chief information officers have a shared understanding of how AI will shape the future of work. HR owns the roles. IT owns the systems. When fewer than four in ten of them see the same picture, every downstream decision inherits the confusion, from role design to workforce readiness.
This is why McKinsey argues the CEO’s role in strategy work cannot be outsourced, and why the AI era calls for leaders to stay engaged with the work and make decisions quickly. Distant leadership slows execution. The alignment gap between functions is a leadership gap, and it does not close on its own.
Easy wins are setting a trap
The second trap is believing early success means you are ready. McKinsey warns that early successes with AI implementations can give leaders false confidence about their organization’s readiness for more complex deployments. Confusing easy wins with strategic capability can waste resources and delay transformation.
The cost of that confusion is now quantified. Gartner predicts that by 2028, 70% of enterprises will abandon agentic AI systems built by vendors through forward-deployed engineering, trapped by rising costs and the inability to adapt these systems independently. A quick win delivered by someone else is not a capability you own. It is a bill you have not read yet.
The stakes keep rising while you deliberate
The ground is moving faster than the typical planning cycle. Advertisements for AI hacking tools on underground forums surged from fewer than 50 per month in late 2025 to more than 1,400 by February, commoditizing attack skills. Anthropic has launched a Critical Infrastructure Defense Program to help protect power grids and water utilities as AI makes cyberattacks faster and cheaper. OpenAI released hundreds of new mathematical proofs from a powerful model, a sign that AI is penetrating expert-driven fields. Brian Chesky says launching a company now requires only a small team, something he says was not possible before.
Add the macro pressure. The San Francisco Fed president warns that AI-driven demand for chips and energy may create supply shocks lasting longer than typical cycles, which could keep interest rates higher for longer. The threat surface, the competitive clock, and the cost of capital are all tightening at once.
LATAM shows the alternative. The airline embedded digital and AI into its organizational structure, repositioned AI as central to strategy rather than a side project, and improved passenger experience, growth, and operational performance. That is the move. Not more pilots. Not more confidence. A redesign of the work, owned by the people at the top, before the easy wins trick you into thinking it is already done.