Two studies landed this week.

The first was from Workday. A quarter of UK workers are spending seven or more hours a week copying information between apps, reconciling conflicting AI outputs, and manually feeding context into prompts. One full working day. Gone. Not to bad AI. To disconnected AI.

The second was from KPMG and the University of Texas. One million AI interactions studied across thousands of workers. What they found: only five percent of workers consistently use AI as a genuine thinking partner. The rest use it for one-off tasks and stop there.

Read those together.

Most workers are spending a full day a week managing AI chaos. And only five percent have figured out how to work with AI in a way that actually functions.

That is not a technology problem. That is a governance problem.

The Workday finding has a name inside enterprise circles. They call it the middleware problem. Workers have become the connective tissue between disconnected systems. They hold the context. They reconcile the conflicts. They carry the thread from one tool to the next because nothing else does.

The KPMG finding has a different name. They call the five percent sophisticated users. Workers who set boundaries, iterate their prompts, ask the AI to explain its reasoning, and refuse to accept the first output as final.

What nobody is saying plainly is this.

The middleware problem and the five percent problem are the same problem from opposite ends.

When there is no governance layer above the tools, the human becomes the protocol. They hold the context because nothing else does. They reconcile the conflicts because no system is designed to. They manually feed the thread from one tool to the next because the tools were never built to talk to each other coherently. That is the ninety-five percent. Busy. Technically productive. Actually burning a day a week on work that should not exist.

When a worker builds their own working method — boundaries, iteration, friction, verification, a standing habit of not accepting the first output — they stop being the middleware and start being the operator. The AI works for them. They are not working for the AI. That is the five percent.

The difference is not intelligence. It is not access. Both groups have the same tools. The difference is structure. The five percent have a method. The ninety-five percent have a subscription.

And here is what that means at scale.

Every organization deploying AI right now is paying for two things simultaneously. They are paying for the productivity gains the five percent are generating. And they are paying for the one lost day per week the ninety-five percent are spending managing the chaos the tools created. Those two costs are sitting in the same budget line. Most organizations cannot see them separately because they have no framework for measuring governed AI use against ungoverned AI use.

That is the governance gap. It is not visible until you name it. Once you name it it is everywhere.

KPMG’s global head of AI concluded that you cannot simply hand workers AI tools and let them loose. You need deliberate routines. Effective behaviors made visible. Purposeful iteration taught as a standard.

That is a governance argument.

Workday’s VP said the companies seeing the most value from AI are building it directly into the systems where people, data, and work come together.

That is also a governance argument.

Neither company used that word. But that is what they found.

The Baseline has been making this argument for eighteen months.

You do not fix the middleware problem with better apps. You fix it with a coherent layer above the tools that carries context, maintains coherence, and stops the human from becoming the protocol.

You do not manufacture more five-percenters through talent acquisition. You make the five percent’s method visible, codified, and teachable. You give the ninety-five percent the structure the five percent built for themselves.

That is what a governance framework does.

The studies found the wound. The wound has a name.

“The Faust Baseline Codex 3.5”

”AI Baseline Governance”
Post Library – Intelligent People Assume Nothing

“Your Pathway to a Better AI Experence”

Purchasing Page – Intelligent People Assume Nothing

Unauthorized commercial use prohibited. © 2026 The Faust Baseline LLC

Similar Posts

Leave a Reply

Your email address will not be published. Required fields are marked *