The people in the middle: when coordination was the job

In most companies a small group of people do the judgment-heavy work and will do even better with AI. Another small group do routine tasks that software takes over first. The difficult question is the large group in between: the people who were very good at coordinating, following up and keeping processes running. That skill is exactly what AI does well.

Why this group needs a plan

These are often loyal, experienced, reliable employees. They know how things really work and who to call. Their value came from holding the process together. When a system routes the work, tracks the status and chases the approvals, much of what they did every day goes away, through no fault of theirs.

Saying "you will now handle exceptions" does not solve it. Handling exceptions well takes skills many of them were never asked to develop: deciding without a procedure, negotiating, owning an outcome.

What an honest plan contains

Realistic numbers. If a department of forty will need twelve people once the process is redesigned, say so to yourself first, and plan from that figure, not from a hopeful one.

Real time to learn. Moving from coordination to judgment takes months of practice on real cases, with support. A one-day training session changes nothing.

Several paths, not one: a new role in the same team, a role in another part of the company, a role outside with genuine help to get there.

A budget. Part of what the redesign saves should be set aside, before it starts, to pay for training, overlap periods and departures done properly.

Why it matters to the business

The way a company treats this group is watched by everyone else. Handled badly, it teaches the remaining staff to hide knowledge and resist every change. Handled well, it shows that redesign is something done with people, not to them, and the next change goes faster.

Before announcing any AI programme, count the people in the middle and write their plan. It is the hardest page of the project, and the one that decides whether the rest works.

← Blog