You have an AI mandate. The question is whether you have an AI return.
A board member asked me two questions last month that every executive running an AI budget should be able to answer.
First: are your customers actually asking for AI features? Is it monetizable?
Second: what are you doing to differentiate against competitors building the same things?
Both are exactly right. Most companies have no answer to either.
Here’s what I told them.
Our customers are not asking for AI. They’re asking for outcomes. Faster results. Cleaner data. Fewer hours turning feedback into something their leadership will actually read. The technology behind any of that is invisible to them — and should be. A handful of tech-savvy users ask whether a specific capability, like sentiment analysis, could do something useful. Most don’t care what’s under the hood as long as what comes out of it works.
That distinction is more important than most executives want to admit. If you’re building AI features to answer a customer demand that doesn’t yet exist, you’re not investing in customer value. You’re investing in your roadmap narrative.
When asked about signs of ROI on Meta’s $145 billion AI spend this year, Mark Zuckerberg said: “That’s a very technical question.” (Fortune, April 2026)
One hundred and forty-five billion dollars. And the answer to “what are you getting for it” is a deflection.
Zuckerberg can survive that answer. His balance sheet is a different category. Most companies cannot. When the board asks what you got for last year’s AI investment and the honest answer is “we’re still learning” — that’s not a strategy. That’s an expensive experiment with no defined end date.
Here’s what I keep watching: board mandates AI adoption. Product builds AI features. Someone starts tracking tokens consumed per employee, tools deployed, percentage of the org on Copilot. The metrics go up. Momentum feels real. Then someone asks what the customer actually got from any of it, and the room gets quiet.
My investors are private equity. Disciplined. When they asked about our AI roadmap I came with three buckets, not a deck.
Internal productivity: 10 to 25 percent gains across engineering, front-line teams, and operations. Input and output both measurable.
Engineering: 98 percent of our new code is now AI-generated, with senior engineers reviewing every line. Developers are one to two times more productive. We monitor token costs weekly — cloud spend is up 30 percent over last year and tracking toward 50 percent more this year. Every dollar of that increase has a number behind it.
Product: we build AI into the product only where customers see surplus value and we have the data access to make it work. Not what AI can theoretically do for them. What they will stay for, or pay more for, because of it.
Here’s what that looks like in practice. The AI capabilities we’re building are aimed at one outcome: helping our customers retain more of their own clients. Better accuracy reading feedback signals. Faster cycles from insight to action. A customer whose retention rate improves doesn’t need to know what model is running underneath. They stay and expand because the outcome is visible to them.
That’s the test we use before building anything. Can a customer describe what they got without using the word “AI”? If yes, build it. If the only way to explain the value is to explain the technology, it’s not ready.
Three buckets. No moonshots. No “exploring the space.”
They nodded. Not because the numbers were large. Because the framework was honest.
On the competitor question — we don’t differentiate on AI. We differentiate on outcomes.
One competitor in our space raised hundreds of millions of dollars. Pure AI CX platform, all verticals, pitched as transformative. What it actually is: a decade-old customer success platform with a new layer of AI on top and deeper pockets to market it. The customers it serves are still asking the same questions they asked ten years ago. The AI didn’t change the question. It just made the pitch louder.
Companies racing to announce AI features first are running a sprint inside a marathon. Customers will eventually notice when the feature doesn’t solve the problem they actually have. The companies building AI quietly behind outcomes customers already care about will win that race.
AI is not the moat. The outcome is.
So here’s the question worth asking before your next AI investment — before the next tool, the next initiative, the next quarter of spend:
Can I show this return to a disciplined investor?
Not a patient one. A disciplined one who wants to know what you got for what you spent.
If yes: invest more. The spend is justified.
If the answer is “it’s still early” or “the value is strategic”: stop calling it an investment. Call it a bet — and attach three things to it before you leave the room.
First: which bucket does this spend belong to? Productivity, engineering, or product. If it doesn’t fit a bucket, it’s not an investment. It’s a pilot with no graduation criteria.
Second: what’s the measurable output you’re committing to? Not tools deployed or employees onboarded. What changes in hours, dollars, retention, or revenue — and by when?
Third: what does a customer notice if this works? If the honest answer is “nothing yet,” that’s fine — write it down and set a date. The discipline isn’t in the answer. It’s in the question.
When the hype cycle ends, the companies that knew the difference will have something to show for it. Everyone else will have tokens consumed per employee.