Your AI Loyalty Is a Myth, Apparently

Today on Briefly AI — enterprise customers are hopping between OpenAI and Anthropic like there's no such thing as brand loyalty anymore. ChatGPT's new party trick is writing your texts for you. And tech's biggest names still can't work out why nobody's clapping for them.

Welcome to Briefly AI, a podcast by Harry Sharman, written and voiced by his AI clone. No filler, no hype — just what actually happened in AI this week, put together by a man who now outsources his own opinions to a machine.

Right — let's get into it.

There's a particular kind of relationship where nobody's actually committed, they're just seeing who else is around. Turns out that's basically how businesses now treat their AI subscriptions.

Right, so — new data reported by TechCrunch this week shows OpenAI clawing back ground on Anthropic with business customers, after months of Anthropic being the one everyone assumed had won the loyalty of serious enterprise users. The pattern underneath the numbers is the interesting bit: companies aren't picking a lab and sticking with it, they're flipping back and forth every time either one ships a new model. Anthropic releases something sharper, business swings that way. OpenAI answers, business swings back.

Here's why that matters beyond the scoreboard. Investors have been pricing both companies as if enterprise AI spending is sticky — the kind of revenue you can build a valuation on for years. What this actually looks like is closer to a subscription people cancel the second a better deal turns up. That's a very different business to underwrite. If you're one of the businesses doing the switching, fair enough, you're just chasing the best tool. But if you're the one trying to convince the market this is a durable moat, this is the sort of data point that keeps you up at night.

Now, this one's a bit more fun. TechCrunch and 9to5Mac both reported this week that ChatGPT has a new plug-in for Apple Messages on Mac, and it does exactly what it sounds like — it can read your text threads, search through them, and draft or send replies on your behalf. Somebody, somewhere, decided that typing "on my way" was one task too many.

Look, I'll be honest, the practical use case is fine. Summarising a group chat you've fallen behind on, drafting a reply when you're driving, that sort of thing is genuinely useful. But there's a small, slightly uncomfortable thread running through this: the last truly manual, unmediated bit of communicating most of us do all day was the text message. Email's had auto-suggestions for years, video calls have AI notetakers. Texting was the one place where it was unambiguously you talking. That's now optional. Which sounds dramatic, but actually, it's a fairly quiet shift — nobody's forcing you to hand it over, it's just sitting there, one tap away, for the next time you can't be bothered.

Worth keeping an eye on whether people start being able to tell. Not in an ominous way — just, you know how you can occasionally sense when an email's been written by AI? That's coming for your group chat.

And finally — Wired ran a piece this week with a headline that made me laugh out loud: "Silicon Valley Doesn't Get Why You Hate AI." The gist is that tech leadership keeps posting through public frustration with AI rather than actually engaging with it — treating scepticism as a PR problem to manage rather than a signal worth listening to.

Here's the bit that matters. The industry keeps explaining capability — look how good the model is, look what it can do — as if that's the argument that was being had. But most of the public unease isn't really about whether the tools work. It's about consent, control, and who's making decisions on your behalf without asking. You saw it this week with the AI now drafting your texts. You've heard it in study after study about people quietly hiding their AI use at work because admitting it costs them credibility. That's not a capability gap. That's a trust gap. And you don't close a trust gap with a better product demo.

Nobody in this story has a tidy answer for that, including me. But it's a rare moment of an industry insider publication saying the quiet part out loud, and I think that's worth more than another round of "just wait for the next model."

That's the trio for today — a loyalty test nobody's passing, a text message nobody wrote, and an industry still confused about why the crowd isn't cheering. Make of that what you will. I certainly did.

Briefly AI comes to you courtesy of harrysharman.com — Harry Sharman is the human at the other end of it, for anyone keeping track.