The Words a Business Runs On
The vocabulary we build a business on rarely means the same thing to any two people. For years that gap stayed invisible, patched over by human judgment. AI is what finally forces it into the open.
What is a client?
It should be an easy question. But every time I ask it, I get a different answer: it’s a person, it’s a company, it’s a group within a company. Everyone’s sure they’re right, but they’re all describing different things.
That gap matters more than it used to. The AI prototypes everyone is racing to build assume a level of underlying data coherence that doesn’t usually exist. It’s not just about “clean” data — it’s about what that data even represents. We’ve hand-waved around core business concepts that are deceptively simple. Until these definitional problems are solved, AI will only make them worse.
In the agency space, this “client” question is just one example of definitional gaps that can kill these efforts. A partner holds one definition, finance holds another, and nobody notices until they have to agree on something.
In the past, each group would bring along its own spreadsheet listing the “clients” that matter. The two sheets never quite agreed, but we used our human context to bridge the gap. Gifts to “clients” meant one list; overdue invoices, another. We held the extra context in our heads, handling the nuances and exceptions the sheets couldn’t capture.
AI supercharges these collisions. Point a model at a business where “client” means five different things, and it can’t produce useful answers. At best, it hits a dead end and asks for context. At worst, you get confident, wrong answers that bury those gaps. Even with added context, you’re manufacturing inconsistency and risk (not to mention extra work each time). Different people in the org no longer just have different spreadsheets; they each have their own LLM-backed conclusions, confidently built on different ground.
The work of defining the key concepts and objects a business runs on is unglamorous, but it isn’t especially hard. A “contact” is an individual; a “company” is an organization we work with; together they make a “client.” The hard part is getting everyone to use the same definitions. Do that, and “what is a client?” finally has one answer.
As we worked to unify vocabulary, data, and operations, I’ve watched it reduce confusion and open up real opportunities for automation. Each layer makes the next possible: clear definitions unlock trustworthy data, trustworthy data unlocks workflows, and those workflows let both people and AI do their best work. The less time we spend resolving spreadsheet gaps, the more we get to spend on the work that matters.
Newsletter
Liked this? I send new essays on data, software, and the business underneath them — occasionally, no spam.