Two executives walk into two different AI strategy meetings on the same Monday morning.
The first asks her team how much headcount they can eliminate by deploying agents into customer support, finance ops, and procurement. The second asks his team which of the company's products would become genuinely differentiated if agents were embedded in the customer experience. Same week, same industry, same level of AI investment on paper. Three years from now, they are going to be in radically different competitive positions.
The companies winning the agentic AI era are the ones using it to create revenue. The companies losing it are the ones using it to cut costs. Both are real strategies. Both are being executed today. They look similar from the outside (same vendor evaluations, same pilot programs, same all-hands presentations) but they produce completely different outcomes over a 36-month horizon. The difference matters more than most boards realize.
Cost-cutting AI is the easier sell: boards understand it, CFOs love it. The math is legible: deploy agents, cut headcount, see savings in the next quarterly report. Vendors design their pitches around this story and most enterprise AI strategies, by default, are organized around it.
It is also the strategy that has the lowest ceiling and the shortest half-life.
Why cost-cutting AI is a defensive crouch
Cost reductions have a hard ceiling. You can save 10% on operations. Maybe 25%. Eventually 40%. You cannot save 200%. The math runs out.
Cost reductions also don't compound. A dollar saved this year is a dollar. A dollar of new revenue from a product agents made possible is a dollar this year, growing into more dollars next year, building data and customer relationships that compound into a moat.
And cost reductions are easy to copy. If you cut customer support headcount by deploying agents, your competitor can do the same next quarter. The savings show up on both balance sheets. You have not gained an advantage. You have updated the baseline.
The hardest part for executives running cost-cutting AI programs to see is that they are not winning the agentic AI race. They are participating in a race to maintain parity. The companies they are competing against are about to do the same thing. The relative competitive position does not change.
Or, let me summarize it this way… in the example at the start of the article, which executive is going to grow market share and which will lose market share?
What revenue-creating AI actually looks like
Product evolution at compressed timescales. Agents in the build-measure-learn loop let companies ship product iterations weekly instead of quarterly. The compounding advantage from being four times faster than competitors compounds for as long as the company stays four times faster.
Customer experience differentiation. Agents that make every customer feel like they have a senior account manager, an expert technician, or a personal financial advisor (depending on the business) create the kind of experience customers actively choose to pay more for. This is the opposite of automation. This is differentiation that did not exist before agents.
Time-to-decision compression. Agents that compress weeks of analysis into hours let leaders make more decisions per quarter. Companies that make ten well-informed decisions per quarter are positioned differently from companies that make three. Over five years, the gap is structural.
None of these patterns are cost-cutting in disguise. They are revenue creation, customer-experience moats, and decision-velocity compounding. They require different investments, different teams, and different boards to understand.
Why this is going to be a 36-month story
A widely cited McKinsey analysis of AI value creation has consistently found that the highest-value uses of AI in enterprises are revenue-generating, not cost-reducing. The companies capturing the most value from AI investments are concentrated in the revenue-creation cohort.
Three years from now, the cost-cutters will have hit their efficiency floor. Their P&Ls will look healthier on a margin basis. They will also be selling commodity products that look identical to their competitors, because nothing about how they spent the agentic AI window contributed to differentiation. The revenue-creators will have spent the same three years building moats that competitors cannot replicate by deploying the same vendors.
The boards that demand to know "what's our agentic AI cost savings story" right now are asking the wrong question. The boards asking "what's the customer experience our competitors cannot replicate" or "what new products become possible" are positioning their companies for the next decade.
If you are a data leader sitting between a board demanding the wrong question and an organization that could pursue the right one, you have one of the highest-leverage moments of your career to reframe the conversation. The CEOs who become household names in 2029 will be the ones who refused to play the cost-cutting game in 2026.
If this resonates, subscribe. And forward it to the CEO whose strategy is being defined by the wrong question right now.
— Kyle