I'm not writing this as a think piece. I'm not writing it because I read a headline or saw a viral thread and had a hot take. I'm writing this because it's happening to me, right now, in real time, and I think most people are still telling themselves a story about AI that isn't true anymore.

The story goes something like: "AI will take the boring jobs, the repetitive stuff, and free up humans for the creative, high-value work." I used to believe some version of that too. I don't anymore. Because the creative work is exactly what's getting hit — and it's happening quietly, one client conversation at a time, without a single headline attached to it.

Three clients, gone

In the last while, I've lost three clients to AI. Not to a competitor. Not to a cheaper freelancer overseas. Not to a bad economy or a client who just couldn't afford me anymore. To AI, full stop.

One was a web project. The client didn't have a bad experience with me, didn't shop around and find someone better — they ran the whole thing through an AI website builder instead of hiring a person at all. One was graphic design work that got handed to an image generator, because why pay a designer for a logo concept when a prompt gets you ten options in thirty seconds. And one was data entry — spreadsheets, Word documents, the kind of clean-up and formatting work I used to be able to count on as steady, unglamorous, reliable income. All three, gone to a tool instead of a person.

That's the part people miss. It's not just the entry-level stuff. It's not just call centers and data entry — though that's disappearing too, and fast. It's web design. It's graphic design. It's the kind of skilled, judgment-based creative work that people used to assume required a human brain, a trained eye, years of experience. It doesn't require one anymore, or at least a lot of clients no longer believe it does. And belief is doing a lot of the work here — because in plenty of cases, the AI output isn't actually as good as what a skilled human would produce. It's just good enough, and it's free or nearly free, and that math wins almost every time.

I want to be clear about something: I'm not bitter about this in the sense of thinking the clients did something wrong. If I were in their position, watching a tool spit out a passable website in twenty minutes for the cost of a subscription, I'd be tempted too. That's the trap. It's not that people are being reckless or cheap in some moral sense — it's that the economics have shifted so hard, so fast, that "good enough and nearly free" beats "genuinely good and costs real money" more often than it used to. I don't get to be mad at the market for behaving like a market. I just have to live inside what it's actually doing.

It's not just me

I've got a friend who's a compliance officer in the financial industry. Not exactly a "creative" field. Not exactly the kind of job you'd expect to be vulnerable to a chatbot. He spent years building the kind of expertise that isn't supposed to be replaceable — knowing regulations, knowing how to interpret them, knowing where the risk actually lives in a business. He's been looking for work for a while now, and he's finding the same thing I am: a huge chunk of what used to be his role is now something an AI system does faster and cheaper, and companies aren't necessarily replacing him with a person doing the job differently. They're just not hiring for the role at all, because the tool does enough of it to make a human hire feel unnecessary.

That's the detail that sticks with me. It's not "AI is doing his job worse but cheaper, and companies are cutting corners." In a lot of cases the tools do a defensible job — they scan documents, flag risk language, cross-reference regulatory changes, generate reports. They don't do it with judgment the way a twenty-year veteran does, but they do it well enough that a business can convince itself it doesn't need to pay someone six figures to catch what the software would've caught anyway. The bar for "good enough" keeps dropping to meet wherever the AI happens to land.

When a solo creative in Toronto and a compliance officer in the financial industry are both getting squeezed by the same underlying shift, that's not a coincidence. That's a pattern. Two completely different industries, completely different skill sets, completely different client bases — and the same wall showing up in both. And I don't think it's a temporary one. I don't think this is a blip we ride out until things "normalize." I think this is just what the labor market looks like now, and it's going to keep looking more like this, not less.

What I'm doing about it

I'm not interested in pretending this is going to reverse itself, or that AI is some passing fad that'll embarrass everyone who overreacted to it in five years. It's not a fad. So I'm adapting, because that's the only option that actually does anything. Complaining doesn't pay rent. Neither does nostalgia for how things used to work.

One thing I'm building right now is AI Web Mechanic — a service dedicated to AI content creation, built specifically for people who've already gone the AI route: they built a website with an AI tool, or had AI generate something for them — and now they're stuck. The site's live, or half-live, and nobody configured the hosting properly. Nobody set up DNS correctly, so half the time the domain doesn't even point where it's supposed to. Nobody touched SEO, so the site exists on the internet in the same way a tree falling in an empty forest exists — technically, but pointlessly. Nobody connected Google Search Console, so there's zero visibility into whether Google even knows the site is there. The AI built them a website. It didn't make them findable, functional, or secure.

That's the gap I'm stepping into. Not fighting the AI, not pretending it doesn't exist or trying to convince people to go back to hiring a human for the whole build — helping people deal with what the AI left behind. Because that's the honest reality of the situation: the tools are good at generating something that looks like a finished product, and they are not good at the unglamorous infrastructure work that makes that product actually function in the real world. Hosting. DNS. Security certificates. Search visibility. Analytics. The stuff that isn't visible on the surface but determines whether the website actually does anything for the business it belongs to.

Here's the honest problem with that plan, and I'm not going to pretend it isn't a problem: most people don't care about hosting, DNS, or Search Console. They don't know what those things are, and they don't want to learn. They just want the website to "work," and they assume it already does, because the AI said it was done and the page loads when they type in the address. Convincing someone they have a problem they can't see — a slow-leak problem, an invisible-to-them problem — is a much harder sell than convincing them they have a problem they can already feel. Nobody wakes up thinking "I wonder if my DNS records are configured correctly." They wake up thinking "why isn't anyone finding my business," and by the time they connect that feeling to the actual technical cause, they've usually already tried three other things first.

So that's the real work right now — not just building the service, but figuring out how to translate an invisible technical gap into something a business owner can feel is urgent. That's marketing, and it's also just honesty. I'm not going to oversell this as some magic fix. I'm going to tell people the truth: the AI got you 60% of the way there, and I do the other 40%, the part that actually determines whether any of this matters.

And for the people who'd rather skip the AI route entirely and just want a website built the old-fashioned way — by a person, from scratch, with someone actually thinking through the design — that's still what Jack of All Media is for. Classic web design, no AI shortcuts, just a human doing the work. I'm not trying to funnel everyone toward AI-assisted anything. I'm trying to meet people wherever they actually are, whether that's "I want a person to build this" or "I already let AI build it and now I need someone to fix what's underneath."

The part nobody wants to say out loud

I think my situation is going to keep being hard for a while. Not weeks. Not months, probably. Longer than that. I don't say that to be dramatic, and I don't say it looking for sympathy — I say it because I think a lot of people in creative and knowledge work are underestimating how fast this is moving, and I'd rather plan for a long stretch of adapting than get caught off guard again the way I've already been caught off guard three times this year.

There's a version of this essay that ends with some tidy line about resilience and reinvention, and maybe that's true, maybe I'll look back on this in two years and it'll read like the moment I pivoted into something that worked. I hope so. But I'm not going to pretend I know that right now. Right now it just feels like standing on ground that keeps shifting, and building whatever I can as fast as I can, because standing still isn't an option anymore.

And here's the bigger thing I keep coming back to, the thing that actually bothers me more than the lost income: AI is genuinely useful. It's exciting. I use it myself, constantly, in my own work — I'm not a Luddite standing outside the building yelling at the tools. But useful and exciting isn't the same as good for us. A tool can be impressive and still be a bad deal for the people whose livelihoods depended on doing the thing it now does instead, faster and cheaper. I don't think AI, as it's being deployed right now, is heading toward a good endgame for humans — not because it's evil, not because it's some villain in a movie, but because nobody in charge of deploying it at scale is seriously answering the question of what happens to the compliance officer, the graphic designer, the data entry person, and the web guy once the work dries up faster than new work replaces it.

That's the part that gets skipped in every optimistic AI conversation I see. Everyone talks about productivity gains, about how much faster things get done, about the incredible capability of the tools. Nobody's got a real answer for the people standing where the work used to be. "Learn to work with AI" is fine advice as far as it goes, but it doesn't change the math when there are fewer total jobs needing fewer total people, and the people who lose out first are often the ones with the least cushion to absorb the hit.

I don't have a tidy conclusion here. I don't have a five-step plan that fixes it, and I'm suspicious of anyone who claims they do. What I have is three lost clients, a friend who can't get hired despite doing everything right, and a new business idea that might work if I can figure out how to explain a problem to people who don't know they have it yet.

That's where I'm at. If you're in a similar spot — creative work, admin work, compliance, whatever corner of the economy you're standing in — I'd genuinely like to hear how you're adapting, because I think we're all figuring this out at the same time, with no map, and pretending otherwise doesn't help anyone.

What I do know is that the people who come out the other side of this okay probably won't be the ones who fought hardest to keep things exactly as they were. They'll be the ones who found the actual gap — the place where the AI does 60% of a job and leaves the other 40% sitting there unclaimed — and built something in that gap before everyone else noticed it was there. That's the bet I'm making with the AI Web Mechanic work. I don't know yet if it's the right bet. But it's a bet, and a bet is more than a lot of people in my position have right now, so I'll take it.