Every company has a Sarah.

Sarah knows which customer orders are urgent even when nobody marked them urgent. Sarah remembers that one supplier always needs a follow-up call. Sarah knows which spreadsheet is “the real one,” which manager approves quickly, which warehouse person needs photos, and which exception can quietly become a disaster if nobody catches it.

Everyone trusts Sarah.

That is the problem.

A business should not collapse just because one capable person is unusually good at remembering things, chasing people, and translating chaos into action. When the best employee becomes the operating system of the company, the business is not well-run. It is being personally rescued.

“Ask Sarah” Is Not a Process

Many companies do not notice this problem because “Ask Sarah” works.

A sales order comes in with incomplete details. Ask Sarah.

A customer wants to know whether stock is really available. Ask Sarah.

A purchase request looks normal, but the supplier changed the packaging size last month. Ask Sarah.

This looks efficient from the outside. No formal process is needed because the answer appears quickly. The team moves. The customer gets served. The problem disappears.

But the process did not work. Sarah worked.

That distinction matters. A real process can be inspected, improved, taught, and automated. “Ask Sarah” cannot. It is not a workflow. It is institutional memory trapped inside one person’s head.

The company does not have a system. It has a person who has learned how to survive the absence of one.

Someone Is Quietly Absorbing the Chaos

The best employees in messy companies often look calm because they are constantly doing invisible work.

They clarify unclear instructions before they become mistakes. They notice when two teams are working from different versions of the truth. They follow up before a delay becomes visible. They remember that finance, operations, warehouse, and sales are using the same words to mean slightly different things.

This work is valuable, but it is dangerous when it stays invisible.

Management sees the outcome and assumes the process is healthy. Orders move, customers are served, and problems are “handled.” What they do not see is the hidden coordination effort underneath. Or worse, they see it and assume things will keep holding together because they always have.

Then Sarah goes on emergency leave.

Suddenly, nobody knows which orders are urgent, which stock numbers to trust, which supplier promises are reliable, or which exception has already been discussed.

This is not a heroic employee story. It is a system design problem.

The company has turned a capable person into human middleware. Sarah becomes the bridge between tools that do not talk, teams that do not align, and processes nobody fully owns. This is how businesses become fragile without looking broken. The fragility hides inside someone competent enough to keep rescuing the process.

AI Should Make Invisible Coordination Visible

Weak AI implementations aim to replace Sarah with a chatbot. That approach is naive and shallow, and it will fail because Sarah’s value was never just answering questions. Sarah’s value was knowing what mattered, noticing what was missing, and connecting context across the business.

Good AI should help make that coordination visible, then make it reusable by the rest of the organisation.

It should turn scattered messages, approvals, order notes, supplier updates, and customer commitments into a clearer operating picture. Not just a dashboard, but a shared layer of context that helps sales, operations, warehouse, and finance understand the same situation in the same way.

When an order is urgent, the system should show why. When a stock number is questionable, it should trace where the latest update came from. When the same supplier issue keeps appearing, it should stop treating each case as a fresh surprise. The point is not to automate every decision. It is to give the organisation the memory, visibility, and context that used to sit inside Sarah’s head.

That is the difference between using AI as a tool and using AI to augment the organisation. A tool waits for instructions. An augmented organisation captures context as work happens, highlights exceptions earlier, preserves reasoning, and helps people make better calls without depending on one person to hold the whole picture together.

Sarah should still be valuable. But she should be valuable because of her judgment, not because she is the only person who remembers how the business actually works.

The Goal Is Less Fragility, Not Fewer People

The lazy argument for AI is fewer people and cheaper operations. That view misses the deeper business case.

The stronger argument is less fragility: fewer dropped balls, fewer situations where one person’s memory is the control mechanism, and fewer decisions made from partial context. Some manual work will disappear, but the point is not to remove the people who understand the business. The point is to stop wasting them as glue.

A good AI system should help the business remember, coordinate, and reason more consistently. It should make the hidden work visible enough that the company can improve it. Because when everything depends on Sarah, the business has not found its hero. It has found its failure point.