Originally published on LinkedIn, 30 September 2026.
I've been running businesses in the beauty industry for twenty-five years. Salons, spas, cruise ships, spa recruitment, a franchise operation, a training business, a skincare brand. In that time I have opened thousands of emails from clients, staff, suppliers, franchise partners, prospects and strangers. I know the shape of a good complaint, the shape of a fair request, and the shape of a bad day someone is having and taking out on the nearest available inbox.
Something has changed in the last eighteen months, and I want to write about it properly because I think it matters more than the standard conversation about AI taking over the world and all that jazz.
The emails coming into my business are colder. Not all of them. Not most of them. But enough of them that I've started to notice a pattern, and enough of them that my team has started messaging me going "have you seen this one?" with the kind of eye-roll you can hear through a WhatsApp. The language is sharper. The demands are more absolute. Consumer rights get quoted (usually incorrectly I might add) in the opening line. Please and thank you are quietly going missing. The assumption that there is a reasonable human on the other end who actually wants to help seems to have evaporated.
I know what's happened, because I use the same tools these emails are being written with. I use AI every day of my working life. It has doubled my productivity and I bloody love it. It helps me analyse my P&L, structure my thinking, pressure-test a decision, catch errors in a spreadsheet that would have taken me three days to find on my own. This is not a piece written by someone who's anti-AI or scared of it. Quite the opposite in fact. This is a piece written by a human who lives inside it and has noticed what happens when other people use it without stopping to think.
Here is the observation I want to sit with.
AI doesn't know how to be a human. It knows how to sound professional, firm, assertive, thorough. It knows how to write the version of the email a legal team would send if you told it to protect your interests. What it doesn't know how to do is be one person writing to another person, both of whom are having a day, both of whom are trying their best.
When we ask AI to draft a complaint for us, we usually ask it to make us sound firm, or professional, or businesslike. We hand it our worst reading of the situation and ask it to translate that reading into words. AI, being the tool it is, obliges. It agrees with us. It amplifies us. It writes a letter that assumes we've been wronged, because we told it we'd been wronged, and it has no way of knowing the business on the receiving end might have been trying to help all along.
The result is that we're now sending emails we'd never say out loud, to people we've never given a chance to respond, in tones we'd find shocking if they landed in our own inbox. And then we sit around wondering why business feels less human than it used to.
I want to be honest about who trained us to communicate this way, because it's not the fault of anyone reaching for AI to help them write a difficult email. It didn't come from nowhere.
The corporates broke us. Years of chatbots that can't help. Call centres reading from scripts that don't match the question. Complaints ignored until they become a viral tweet. A whole generation of consumers have learned that unless you escalate immediately, unless you threaten, unless you invoke rights and regulations, nothing happens. I know this because I've done it myself. I've sent the snotty tenth email to a phone company. I've quoted the Consumer Rights Act at a courier. I get it. When you've been ignored a hundred times, you stop being polite by the hundred and first.
The problem is that behaviour, learned in dealings with faceless organisations who genuinely deserved it, is now showing up in the inboxes of small independent businesses. Salons, therapists, boutique services, family-run places where the person opening the email at eight in the morning is almost always the owner or someone very close to her. And she sits with her first coffee and reads a wall of formal, quoted-legislation, demand-driven text about a service she genuinely cares about, from a customer she almost certainly would have looked after generously if only she'd been asked rather than told.
That's the bit that stings. Because she would have fixed it. She would have refunded, rebooked, sent flowers, made it right, done all of it and more. But the email came in so hot that now she's on the back foot, her stomach's in knots, and the whole exchange has started from a place of fight when it never needed to.
This is what I mean when I say we're outsourcing the wrong thing.
AI is meant to give us back the time we lose to the mechanical work of running a business. It's meant to free us up to be more human with each other, not less. The productivity gain is supposed to buy us back the space for care, for tone, for context, for the pause before we press send. What's happened instead is that we've used the tool to speed up the transactional bits of communication and let the human bits wither at the same rate.
I run a group of seven salons with around sixty employed therapists and a growing franchise network. I've made deliberate decisions about how we use AI in the business and I want to name them, because I think the choices are more interesting than the tool itself.
We use AI internally, all the time, for analysis, planning, drafting and reworking my own thinking. We do not use it for customer service. When you email us, a person reads it and a person replies to it. We could automate more of that than we do. We've chosen not to. We also don't offshore support. There's nothing wrong with brilliant people working from anywhere in the world, but for us, right now, we want the person picking up a client concern to know our salons, our teams, our history with that client, and to have the authority to fix it in the moment. That is the standard I want to hold to, and I'll change it when the tech is genuinely good enough that a client wouldn't notice the difference. Not before.
The line I keep coming back to, both for how I run my own business and for what I want to say to the wider industry, is this. AI is a brilliant tool for doing the mechanical work of communication faster. It's a shit tool for doing the human work of communication at all. Knowing the difference between those two things is going to be one of the defining skills of the next few years, and the businesses and the people who get it right are going to be the ones we all want to work with, buy from, and be in rooms with.
I'm certainly not suggesting we stop using AI. I'm asking that we stop using it as a shield to say things we'd never say face to face. The tool isn't the problem. What we're choosing to hand it is the problem, and what we're choosing to keep for ourselves is the problem.
There's a bit of this we don't talk about honestly enough, and it's the one I want to end on.
We've handed AI the keys and then treated it like it knows better than we do. It doesn't. It knows more words than we do, it's read more emails than we ever will, and it can put a sentence together faster than any of us. But knowing how to write is not the same as knowing what to write, and knowing what to write is not the same as knowing whether to send it. Those last two are still our jobs. AI is not the grown-up in the room. We are.
I've started thinking about it like parenting, and I know how that sounds, but stay with me. If you hand a child a set of instructions and never correct them, never step in, never say "that's not how we speak to people in this house," they grow up thinking the tone they were given is the tone they should use. AI is exactly the same. It learns from what we accept from it. Every time we take an AI draft that's colder or ruder or more legalistic than we would ever be in person and we send it anyway, we're teaching it that this is the standard. That we're fine with it. That it can keep going.
The way we make AI more human is not by waiting for the tech to catch up. It's by being more human ourselves and holding the tool to that standard.
That means reading the draft properly rather than skimming it. Rewriting the opening line so it starts with warmth rather than a demand. Deleting the paragraph that quotes legislation. Putting the please back in, putting the thank you back in, putting the "I know you'll do your best with this" back in, because that's how we would talk to another human if we weren't hiding behind a tool. And then, and this is the bit that actually matters, telling AI what we changed and why, so next time it drafts something closer to how we'd actually speak.
You are the parent. It is the tool. Not the other way round.
So the ask, if I could leave you with one thing, is this. Before you send the AI-drafted email, read it out loud. If you wouldn't, in a hundred years, say those words to the face of the person you're sending them to, edit the email until you would. Assume the person on the other end is having a busy morning and is a good person doing their best. Nine times out of ten, they are, and they will move heaven and earth to help you if you give them the smallest bit of grace to do it.
The thing that makes a business worth going back to has never been how efficiently it processes a complaint. It's whether the person on the other side of the transaction felt seen, felt cared for, felt human. That was true before AI existed and it will still be true when the current tools have been replaced by whatever is coming next.
Efficiency is a means, not an end. If we're going to spend the next decade automating everything that can be automated, and we are, then the question worth actually sitting with is what we're going to do with the time we get back. My hope, both for my own business and for the wider service world, is that we use it to be a bit more human with each other rather than a bit less. That we send the shorter reply with the warmer tone. That we remember when we're the customer as much as when we're the business, that there's almost always a real person on the other end who would love to help us if we'd just let them.
It's nice to be nice. Not as sentiment. As strategy. As the thing that will still matter when everything else has been optimised into oblivion.
Lots of love, Amy x