AI just magnified the enterprise’s biggest problem

"We have the insight. We just can't get anyone to act on it." That's something I hear often from CX leaders. It shows up in different ways, across industries, but it always points to the same underlying issue.

Just last week, for example, I sat with a team at a global pharmaceutical company. They had everything: customer research, behavioral data, operational metrics. They even had access to every major AI tool on the market, with the budget to use them freely. Six months in, nothing was really moving. The insights were strong, the tools were powerful, but something essential was missing: what I keep calling the connective tissue, the layer that allows all of that information to actually work together.

That's a big problem, and it isn't new. But AI is making it even bigger.

The biggest problem in enterprise

Enterprise alignment has always depended on shared context. Without it, teams optimize locally, make decisions that contradict each other, and produce experiences that feel fragmented to the customer because, internally, they are. This is not a technology problem; it's a structural one.

Every enterprise already has the pieces of customer understanding. BI tells them what's happening. VoC explains why. Interaction logs — tickets, help center, digital analytics — track what people actually do. These are three views of the same customer, but they don't connect. You can group them under a label like "onboarding" and call it integrated, but that doesn't create understanding. It creates a bucket. The bucket doesn't know what happened before, during, or after any interaction. It doesn't connect a metric to a moment, or a moment to feedback, or feedback to a decision.

For a long time, organizations worked around this gap with effort: more workshops, alignment meetings, and cross-team translation. Senior leaders learned to read across silos and make judgment calls in the gaps. It was competent, but exhausting, and it never really scaled.

That was manageable when the pace of the business allowed it. But AI has drastically changed that pace, so the problem is magnified.

What AI exposed

What's happening now inside most enterprises follows a familiar pattern. Leadership commits to AI. Tools are rolled out. Teams are told to experiment. Within a few quarters, every function is building something: prototypes, pipelines, models, dashboards. It looks like progress.

But most of those outputs are built in isolation. They are useful within a team, but disconnected from the rest of the business. The underlying data remains fragmented, and the models are not grounded in a shared understanding of the customer or the business.

The result is what you would expect: inconsistent outputs, duplicated work, and a growing volume of insight that doesn't translate into action. The data naturally doesn't flow. It's in pockets of the organization. Crap in, crap out. And it's very easy to generate a lot of crap right now.

Grounded vs Ungrounded AI

There's a useful split here: grounded AI work and ungrounded AI work. Grounded work compounds; coding is the obvious example. The model writes a line, a compiler checks it, and the loop closes against something real. Ungrounded work just produces more output. Strategy decks, journey maps, marketing copy: AI can generate all of it endlessly, but "is this right?" has no answer. Two teams describe the same customer experience. Neither is wrong. There's nothing shared to verify against. In the enterprises I'm close to, this is where AI is hitting a wall. The model isn't the problem. What's missing is the referent: a shared layer humans and AI can both point at.

The honest diagnosis isn't that AI is underdelivering. It's that AI is exposing, at speed, something enterprises were already tolerating: no shared understanding across the business.

Insights without context don't drive decisions. They drive more meetings.

Same old problem. Bigger consequences.

This is why the framing inside CX teams is starting to shift. It used to be, “We have insight, we need to get the business to act on it.” Now it’s: “We have insight, we have data, we have AI, and we still can’t get anyone to act.”

And that exposes a new organizational gap.

The builder gap

Most enterprises are already structured around consuming insight. Product teams, analysts, domain experts, and executives all use information to shape roadmaps, priorities, and decisions.

But very few organizations are designed around building shared context.

What’s increasingly required is a different kind of role: the builder. The people who design that connective layer. The ones who think in plumbing. The ones who can see how something like the billing experience connects marketing, product, operations, support, and data into a single understanding of the customer.

So the real question becomes: “Who in my team is the builder?”

If the answer is no one, the rest doesn’t land.

The AI being rolled out across the business won’t connect. The insight already on the table won’t move. And whatever gets built next will become another local optimization: useful inside one silo, invisible to the rest of the organization.

Enterprises don’t need more data. They need the layer that makes their data make sense together.

That’s the work.

On May 27, I’m walking through the biggest piece of work we’ve ever shipped, and why I think it changes how enterprises make decisions alongside AI.

If you’ve been wrestling with this problem too, come.