Where AI actually belongs in a service business (and where it doesn't)

The AI conversation right now is exhausting. On one side you have people telling you to replace half your team by Christmas. On the other you have people saying it's all hype and nothing has really changed. Neither is true, and neither is helpful if you're actually trying to make a decision about your business this quarter.

The useful question isn't whether to use AI. It's where to put it. Because in a service business, the wrong answer costs you the thing that made customers pick you in the first place.

Here is how I think about it after helping brands work through this a few dozen times now.

Start with what customers actually notice. Not what they say in surveys, what they notice. In every service business I've worked with, the moments that build loyalty are almost never the ones people expect. It's the follow-up call that wasn't required. The person who remembered your kid's name. The recovery when something went wrong. These are cheap to deliver and impossible to fake, and they are the reason customers stay.

Now look at what those moments have in common. They all require a human being with enough time, energy, and information to notice something and do something about it. Not a scripted response, not a policy, a decision made in the moment by someone who cares.

That's your answer for where AI belongs. It belongs everywhere that gets in the way of those moments. The inbox triage, the data entry, the meeting notes, the report generation, the first-line responses to the questions your team has answered a thousand times. Everything that drains attention without creating anything a customer would notice.

And it does not belong in the moments themselves. If a customer complains and gets a warmly worded email that clearly wasn't written by a person, you have not saved yourself time. You have taught that customer that you were never really there. That lesson is expensive and it sticks.

The trap most businesses fall into is the opposite of what they think. They assume the risk of AI is doing too much of it. The actual risk is doing it in the wrong places while ignoring the boring back-office wins that would free up their people to be more human, not less.

So a simple test before you automate anything: would a customer notice if this became a bit less warm, a bit more generic, a bit more predictable? If yes, leave it alone and pay a person to do it well. If no, automate it today and give your team back the hour.

That's it. That's the whole framework. Do the boring things with AI so your people have the capacity to do the memorable things with their full attention. Companies that get this right will feel faster on the inside and more alive to the customer at the same time. Companies that get it wrong will feel slick and forgettable, which is exactly the position you don't want to be in when a competitor with better judgment shows up.

If you want a hand working out which side of that line each part of your operation sits on, that's most of what we do. Get in touch and we'll talk it through.