The team has time now. The customer is still waiting.

What would your team do tomorrow if the paperwork finished before lunch?

Take a fictional distributor. Its customer service team has started using a system that prepares routine order updates. The information is checked and sent earlier. People have some room in the afternoon.

There is a list of unresolved customer problems they could work through. The customer who has asked twice about a partial delivery. The account waiting for a credit decision. A recurring complaint about orders arriving in separate boxes.

By Friday, the list has barely moved.

The team has produced more updates, though. Someone asked for a version for sales. A manager requested a daily digest. Both were easy enough to generate.

Imagine looking at that list after approving a project specifically to give your people time for customers.

“We've got the time now. Why are we still here?”

The customer team still has the same response targets. Resolving a credit question still requires an approval from finance. Nobody has agreed which routine work can wait while a difficult case gets proper attention.

Making the paperwork faster did exactly what it was supposed to do. The team is still working under yesterday's instructions.

Follow the demand as well as the hours

The Economist's September 2026 analysis describes several kinds of work growing around AI: building and maintaining data centres, implementing the technology, and serving demand that becomes affordable when a service costs less to deliver. It also reports declines in routine administrative and customer-service employment. These developments can happen at the same time.

The magazine estimates around one million American jobs created by AI so far. That is its estimate, drawing on several employment measures and comparisons with broader trends. It acknowledges that some of the growth has other causes. The figure cannot tell you how many people your business will need next year, and subtracting a separate layoffs count would not establish a causal net total.

For a distributor, the more useful question may concern customers. Could a technical team now answer smaller accounts whose requests previously cost too much to investigate? Could it prepare a maintenance quotation while the customer still has a shutdown window available?

There is also demand outside software. The infrastructure described in the article needs electrical equipment and cooling, with installation and maintenance behind them. A business already serving those trades can investigate a specific customer need. Entry still depends on qualifications, capacity and purchasing requirements. A headline about investment does not put you on an approved supplier list.

Choose one service to test with existing customers. Make a bounded offer, record whether anyone buys, and count the effort through delivery. A faster quote has commercial value when it helps win worthwhile work. Producing twice as many unwanted quotes gives your salespeople something else to chase.

Which queue gets the afternoon?

The debate about AI and jobs invites a very large prediction. Inside a company, there is a smaller decision you can actually make: where will the first usable hours go?

Leave that open and the easiest request can win. Another summary. A refreshed presentation. An answer to a question that already had a perfectly adequate answer yesterday.

Cheap output is tempting. It is also capable of generating work for whoever receives it. Someone has to read the new digest. If its numbers differ from the weekly report, someone will investigate. A manager can spend the morning reconciling two documents produced to save the manager time.

For an owner who wants the company to give customers more attention, this is a particularly irritating result. The equipment is working. The service still feels rushed.

In the distributor example, the customer service manager could reserve a defined afternoon block for unresolved delivery cases. Finance would need to agree how credit questions get answered within that block, and which decisions the customer team can make itself.

That turns an available hour into a usable hour.

The team also needs permission to leave something alone. If the new sales digest has no clear use, its requester should agree to drop it. Keeping every old obligation while announcing a new priority gives people a choice they cannot comfortably make.

Ask them to use their initiative, and the conscientious ones may simply finish both after hours.

You cannot put twelve people's spare minutes in one chair

There is a spreadsheet version of this problem too.

It adds small time savings across a department and converts the total into a person. The arithmetic can be correct while the staffing conclusion is impossible.

A few minutes between calls may help an employee write better notes or prepare the next conversation. Those minutes may never become a continuous block for investigating a complex complaint. If phone coverage must remain unchanged, the apparent spare capacity cannot all leave the queue at once.

Test the shape of the time you have recovered. Watch a normal week. Include the day someone is absent and the afternoon when orders bunch together.

Can the work be batched? Can one person cover routine enquiries while another finishes a difficult case? Does the person with time have the knowledge and authority required for the work you want done?

Without those answers, assigning a new project can just move the overload to someone else.

There are perfectly sound projects where the benefit is less overtime or less frantic cover during absence. There are also businesses with declining demand, where fewer paid hours or a smaller external contract is the intended outcome. Say so in the business case and plan it properly.

An automation result gives you evidence about work. Decisions affecting jobs require their own assessment and the applicable employment process. A tidy total of minutes does not settle them.

Make room for someone to learn the harder work

Recovered time can also let an experienced colleague teach someone how to handle exceptions. Reserve time for that person as well. Otherwise the junior gets a new responsibility while the only person who can help remains fully booked.

In the distributor example, start with one type of delivery problem. Let the junior prepare the case and propose a response; the experienced colleague checks it before anything is promised. Review the cases together, including the ones where the proposed answer was wrong. Expand responsibility only when the work supports it.

This gives the company a way to develop capability while improving service. It requires an actual place in the working week. Calling someone the AI champion without changing their workload can leave them running an unpaid internal help desk.

Give the gain somewhere to land

Return to the customer list.

Suppose the manager agrees that, for the next month, the recovered afternoon block goes to delivery cases that have remained unresolved beyond the promised response date. One person owns that queue. Finance has an agreed route for decisions the team cannot make.

The manager can now check something meaningful at the end of the month. Did older cases get resolved? Did fewer customers have to ask again? Did ordinary service remain acceptable while the team did this work?

The time estimate still matters. It helps explain whether the plan is feasible. The customer result tells you whether it was worth doing.

Avoid turning the review into another reporting project. Use the existing case records where they are reliable. A small sample of closed cases can show whether “resolved” means the customer received an answer or someone merely changed a status.

If the backlog stays where it was, inspect what interrupted the protected time before asking people to work faster. An approval that arrives after the block ends can ruin a sensible plan. So can a newly announced priority from someone who never heard about the first one.

This is the part of AI transformation I want a business owner to be able to see: the customer gets a proper answer while there is still time to do something with it. Your experienced person gets to follow the awkward case through. The weekly meeting can spend less time asking why it is still open.

You have room to improve the company you meant to build.

For a manufacturer or B2B distributor, bring one recurring workload and one result you have wanted the team to work on for months. We can discuss whether changing that workflow would create usable capacity for it, and what would need closer assessment.

Request a Strategy Session.

Bring the postponed work too. It belongs in the conversation.