🤔 Is AI Coming For My Job? An Honest Gut-Check

You’ve typed it into a search box late at night and got a doom headline, plus a post from a man suspiciously thrilled about it. Neither helped.

Here’s the honest version, including the bits nobody knows, because a guide that pretends otherwise is just a nicer headline.

Working for yourself, the question is different. Nobody is restructuring you out. It’s quieter and worse: is what I sell still worth paying for, and what do I change now?

Cards on the table. For six years I sold the back end of other people’s businesses, then the tools got good at exactly that. I had a wobble, learned it properly, and the work moved rather than vanished. I sell courses now, so weigh this page accordingly.

What actually changes

1. The easy version of what you do gets cheap. Not your best work. The first draft, the tidy-up, the bit clients grudgingly paid for.
2. Speed stops being the thing you sell. If fast was your edge, everyone has that edge now, your client included.
3. Some clients try it themselves first. They arrive later with something half-built, asking a different question.

What genuinely does not

1. Somebody has to be accountable. A tool cannot be sorry, cannot be sued, cannot ring the supplier and sort it.
2. Being chosen. People buy from someone. That has survived every previous round of this.
3. Deciding what is worth doing at all. Clients rarely arrive knowing what they need.

What nobody knows

How fast. Which trades first. Whether your clients will care or carry on as before. Whether prices settle back up once the novelty wears off.

Anyone handing you a date and a percentage is guessing with confidence. That’s a style, not information.

Signal and noise

Noise: a headline with a big number and no method. A demo video. A prediction with a year in it. Anyone selling you “your trade is finished” or “you’re completely safe”, because both sell.

Signal: things that happened to you, or to somebody you could ring.

• A client asks for a lower price and mentions having a go at part of it themselves.
• A client sends you something they made and asks you to fix it, rather than to make it.
• An enquiry you always used to get stops arriving, for a season, not a fortnight.

One of those beats fifty headlines. None of those, plus a bad week of scrolling, is a mood rather than a market.

The prompt for the thing that rattled you

Nick this one:

You're helping me work out whether something I read about AI and my work is a real signal or just noise. I'm self-employed: a [WHAT YOU DO] selling mostly to [WHO, e.g. "independent letting agents and two small builders"], charging [HOW, e.g. "day rate plus two retainers"].

Here's what rattled me: [paste the headline or post, or write down exactly what a client said]

Give me:
1. What's actually being claimed, in plain words, and which parts are evidence versus somebody's opinion or forecast.
2. What would have to be true for this to matter to ME specifically, given who I sell to. Be concrete.
3. What I'd notice in my own work first if it were true, and roughly in what order.
4. What in this is genuinely unknowable. Say so plainly rather than filling the gap.
5. Finish with the single sentence I'd say to a friend who asked if I'm worried.

Rules: British English, under 400 words, no reassurance and no doom. You can't see the job market, my clients or my area, so don't pretend to. Make no confident forecasts you can't support, invent no statistics, studies or dates, and say "this is a guess" wherever you're guessing. Ask me up to three questions before you start.

Number 4 is the one to read twice. Most of the dread is your brain trying to finish an unfinished sentence. Naming the unknown bits as unknown lets you put them down.

What being early is worth

Not immunity. Anyone promising that is selling something.

Time, mostly. Six months of ordinary use means that when a client asks “could you just do this with AI?”, you answer instead of freezing. Daily users tend to be calmer AND more specific about what’s coming, because they’ve seen what these things are bad at.

Set your own early warning list

Stop refreshing the news. Decide now what would actually count.

Nick this one, straight after:

You're helping me build a short early warning list so I can stop checking the news about AI and my trade. I'm a [WHAT YOU DO] selling to [WHO], charging [HOW].

Give me:
1. Three things I'd notice FIRST in my own business if what people pay me for were really changing. Observable and specific, e.g. "a third enquiry in a row asking for a fixed price instead of a day rate", not "market disruption".
2. For each, what it would NOT mean, so I don't panic at one quiet month.
3. One sign pointing the other way: more work about, not less.
4. Then a short version I can save and re-run in six months, which asks me which of the three actually happened.

Rules: British English, plain, no pep talk. You can't see my market or the wider job market, so don't pretend to and don't forecast. No percentages, no timelines, no invented industry examples. Ask me anything you need to know first.

Save the short version somewhere you’ll find it.

This is the think, not the audit. It won’t tell you which parts of your income are exposed, or what to change by Friday. That’s a more uncomfortable twenty minutes: The 20-Minute Job Safety Check.

Tonight

1. Find the one thing you read this month that actually rattled you.
2. Run the first prompt on it.
3. Close the other tabs.

You end up holding one true sentence and a shorter list of worries.

P.s. if the honest answer is “I should just learn this properly”, that’s what The Claude Kickstart is for: five real jobs off your plate and a playbook for your trade. £47, one weekend.