PetWise

For years, enterprise data strategy has focused on acquiring more data, integrating more systems, improving data quality and making information easier to access. Those investments remain important. But AI is exposing a different challenge.

The question is no longer whether a machine can access information. The question is whether it understands what that information means.

Modern AI can read millions of records and identify patterns faster than any analyst. Yet it can still produce a confident answer based on the wrong definition, hierarchy or business rule. The issue is rarely the data itself. The issue is meaning.

The Enterprise Meaning Problem

Consider a simple finance question: Why did contribution margin decline last quarter? The underlying data may be accurate, yet the answer still depends on the business context. Which sales definition applies? Which costs belong in the contribution margin? How are shared expenses allocated? Which fiscal calendar and customer hierarchy are authoritative? Which transactions should be excluded?

Experienced analysts often answer these questions almost instinctively because they understand the business. AI does not possess that intuition. It can only reason from the context it is given. If that context is inconsistent, undocumented or fragmented across spreadsheets, reports, departments and individual employees, AI inherits the ambiguity.

This is not fundamentally an AI problem. It is an enterprise knowledge problem.

Semantic Models as Meaning Infrastructure

Most organizations already possess the knowledge required to answer their most important business questions. Much of it is simply trapped in disconnected forms: spreadsheets, report logic, tribal knowledge, meeting discussions and undocumented processes.

I have increasingly come to think of that business knowledge as an enterprise asset. Definitions, hierarchies, calculations, relationships, allocation rules and terminology represent years of accumulated operating experience. Semantic models are one mechanism for making that knowledge durable.

  The question is no longer whether a machine can access information. The question is whether it understands what that information means.  

Historically, semantic models were often treated as reporting infrastructure. They gave dashboards, reports and analysts consistent definitions. In an AI-driven enterprise, their role becomes more strategic. They can become meaning infrastructure, which is a governed translation layer between enterprise data and the business context that intelligent systems need to reason effectively.

Data infrastructure tells AI where to find the facts. Meaning infrastructure tells AI how the business interprets them.

Trust Must Work Both Ways

Enterprise AI will only be as trustworthy as the context it is allowed to reason from. When finance, operations, supply chain, sales and executive reporting maintain competing definitions for the same metric, AI does not resolve the inconsistency. It amplifies it.

AI needs to know which business logic it can trust. Humans need to know which business logic the AI trusted.

That second half matters. When an executive reviews an AI-generated recommendation, they should be able to trace it back to the approved definitions, calculations, assumptions, hierarchies, allocation methods and source systems that informed it. Explainable AI should not stop at explaining how a model reasoned. We also need to understand what it reasoned from.

The Human-AI Feedback Loop

There is also an opportunity for AI to help strengthen the semantic foundation itself. AI can help document inherited logic, identify competing definitions, surface dependencies, clarify ambiguous rules, propose tests and help teams turn tribal knowledge into governed documentation.

But the feedback loop needs a human boundary. AI can help build the framework for trust; it should not be the final authority over the framework it operates within. Humans still need to decide which definitions are authoritative, set the guardrails, approve changes and manage the limits.

That loop creates a reinforcing cycle. People define and govern business meaning, semantic infrastructure makes that meaning reusable, AI reasons from it, humans trace and review the result, and what we learn improves the semantic foundation again.

The Real Enterprise Asset

Powerful AI will become increasingly accessible. Models will improve. Data volumes will grow. Technology will continue to change.

What may remain difficult to replicate is a trusted understanding of how a business actually operates. Organizations that capture that understanding in a governed semantic foundation create an enterprise asset that can outlive individual reports, applications, platforms and even ERP implementations.

The future winners may not be the organizations with the most data. They may be the organizations that have done the hard work of defining what their data means, governing which meaning is authoritative and making that meaning traceable.

Because AI does not need more data. It needs trusted meaning. And trusted meaning begins with a semantic foundation.