“We've had AI on the roadmap for a year. We're still not seen as an AI product.”

Most SaaS companies in 2025 shipped something with AI: a copilot, a summarization feature, or an intelligent recommendation. The demo improved, and the release notes became more interesting. Customer retention held, but win rates didn’t budge and the growth line didn’t move.

The instinct is to blame execution:

• The features weren't polished enough.

• Marketing didn't tell the story well.

• The timing was off.

But execution isn't the problem. The features shipped, but the story didn't change.

The harder question, which most leadership teams skip, is this: what does this actually make stronger?

Not what AI enables in the abstract. Not what competitors are doing. But specifically: which part of our product becomes harder to replace when we add this? Which customer problem becomes meaningfully more solved?

Without that answer, every AI feature is an isolated bet. Nothing compounds and nothing adds up to a position the market can recognize.

The innovation trap

Shipping AI and winning with AI are two different problems.

In their State of AI 2025, McKinsey found that 64% of companies reported AI is enabling innovation, yet only 39% reported achieving meaningful business impact from AI. That 25-point gap is not an execution problem. It is a strategy problem.

When there is no clear answer to what the AI is supposed to strengthen, teams default to what is visible: features that look like AI, demos that feel like AI, and positioning language that mentions AI. The product gets busier. The strategic position does not change. Innovation and impact are not the same thing.

What "slapping AI on your product" actually costs

There is a version of this that is easy to dismiss: the company that simply prefixes "AI-powered" to their marketing copy without changing anything underneath. The more common and more expensive version is subtler.

A team builds a real AI feature. It works. Users engage with it. But it was built in response to competitive pressure, or a board ask, or an engineer's enthusiasm. Not in response to a clear answer about how the product is trying to become harder to replace. The feature exists in isolation. It does not reinforce the reason customers stay. It does not make the product stickier in the workflows that matter. It adds complexity, expands the support surface, and because every competitor has access to the same underlying models, it commoditizes within a cycle or two.

The investment was real, but the return was not.

The question that changes the math

There is one question that reorients AI investment from isolated bets to something that actually compounds: where does our product create value that customers genuinely cannot recreate elsewhere?

That is a moat question, not a feature question. And most leadership teams have never answered it precisely. They have a sense of it. They can gesture at it. But when you ask Product, Sales, Engineering, and CS to each write down the two or three things that make the product genuinely hard to replace, you rarely get the same list.

That divergence is where AI investment goes to waste, in the absence of a shared, tested answer to what you are actually trying to strengthen.

AI applied to a real moat compounds. AI applied to a feature that customers like but do not depend on produces a better demo and a flat NRR.

What this looks like in practice

The companies building genuine AI advantage right now share one characteristic: they started with the moat question, not the feature question. They asked where their product is already embedded in something customers cannot afford to lose or reroute, and then integrated AI to embed it deeper, faster, and more difficult to replicate.

That is not a technology decision. It is a strategic decision that has technology implications.

The roadmap is the output of that thinking, not the starting point.