A family buys an Oxford English Dictionary for their children. The book is thick, the cover handsome, and it looks wonderful on the coffee table. Year after year, no one opens it.

In reality, an Oxford English Dictionary is more than a list of words. We all remember being stopped by a word we did not know and reaching for it. Say the word was bourgeois. The dictionary lists several meanings for that one word, shows where it came from, and dates its earliest recorded use to 1604. It shows the compounds and derived words built from it, gives the pronunciation in both British and American English, and points you to the nearby entries on the page. The students who reached for the dictionary often built larger vocabularies and a surer instinct for expression.

The Oxford dictionary is a learning tool whereas the business glossary we build in the catalog is a documentation tool. It records a definition and stops there.

The Comparison We Often Miss

The Oxford English Dictionary is the principal record of the English language, built over more than a century to trace how words live and change. It does more than define. It shows where a word came from and how it has been used, each sense backed by real evidence. A reader opens it for one word and leaves with a little more of the language. It taught you not just what a word meant, but how to use it.

A business glossary is the enterprise’s attempt at the same idea, a single place where the terms a business uses are defined and agreed, so everyone means the same thing by the same word. It lives inside the data catalog, maintained by stewards, approved by councils, and meant to give the organization one shared language. The intent mirrors the dictionary. In practice, it usually stops at the definition. It records that a term exists and what it means, and rarely connects that meaning to the data beneath it or the business decision in front of the reader. The Oxford English Dictionary taught you to use a word. The glossary only tells you it exists.

How We Got Here

Walk through almost any business glossary and you will see governance teams collecting terms that are already universally understood: Customer, Product, Employee, Address, Phone Number, Country, Currency. Few people open a data catalog to ask what a customer is. The real value sits in the terms that carry weight in a decision, the ones like Active Customer, Net Revenue, Risk Exposure, or Customer Lifetime Value. Those are the ones worth settling.

Most of the time, we are simply translating a column description from the physical data dictionary into a business term, dropping it into the catalog, and calling it progress. We follow the process, move each term through intake, review, and approval, and once a definition is recorded we call it done. The questions that actually matter are the ones the glossary never reaches, who relies on the term and what business decision turns on it, where it is used and what breaks when it is wrong. A definition alone answers none of that. How does that help anyone?

Glossary for External Reporting

I was asked to look into a reconciliation problem at a telecom company, where the external reporting team struggled every quarter to make its numbers line up. Those figures rarely come from one place. Before a metric is trusted enough for regulators, investors, or the annual report, it has to be reconciled across several functions. Consider a metric like revenue. Finance, Billing, Sales, Product, and Analytics each calculate it their own way, and external reporting depends on it.

So we built a glossary around those metrics in the Collibra catalog. For each one we captured its definition, its purpose, how it was calculated, and who owned and approved it. The users themselves added the business decision it supported. Once it was done, another department came asking for the same. We never had to sell the glossary. The need did that for us.

The terms worth capturing are the concepts that create confusion, produce inconsistent reporting, drive important business decisions, or need explaining again and again.

The Metadata of a Business Glossary

The definition is not enough. Like the OED, a business glossary must give the reader more than meaning. It must connect each term to the data it describes, the rules that govern it, and the work it serves.

A term carries its data classification, so the reader knows at once whether the information is sensitive or regulated. It sits inside the business process it serves, grounded in the work it supports. Its own entry holds the calculation, an example, and its place among broader and narrower terms. Behind it sits the reference data, the code values and code lists that set what it is allowed to hold. From there it reaches the physical data dictionary, the real tables and columns where the concept lives. The last link is the business data quality rules, which tell the reader whether the data behind the term can be trusted today.

Six links, and the term stops being a dictionary entry. It becomes a place a business user can stand and see the whole picture: the meaning, the data, the rules, and the trust behind them.

[Six-link diagram lands here. Business glossary term at the center, connected to data classification, business process, definition with calculation and hierarchy, reference data, physical data dictionary, and business data quality rules.]

AI-Ready Business Glossary

Agentic AI changes everything. We can no longer afford to build a static glossary, a set of definitions kept behind glass. What the enterprise needs now is a business glossary that serves as the business decision layer for people and machines together. A person opens the glossary for the meaning behind a term. An AI agent needs the same term in a form it can act on, with the calculation, the source, and the limits spelled out. The same term, serving both humans and machines.

A static business glossary inside the catalog helps no one now. In the modern era, the glossary has to be embedded where the work happens, inside the reports leaders read, the dashboards they trust, the AI agents that reconcile the numbers and check for compliance. When the meaning is clear, an agent can be trusted with the business decision behind it. What the next generation needs is a glossary that stays current on its own, carries its own history, and gives people and machines one agreed answer to what a term means today. Not a definition kept behind glass at a metadata museum.

This is not a distant idea for me. For my capstone project at the close of my Berkeley certificate in AI: Business Strategies & Applications, I proposed to build an Intelligent Glossary Curator, an AI agent that contextualizes data assets in business terms. Designing it convinced me the glossary’s next chapter is already within reach.

So here is the question worth carrying. When your AI agents come looking for business context, will your glossary be AI-ready?