I build AI into most of what I ship now. A home-services business on the Coast runs its whole operation through a system where photographed job sheets, notes and routine admin get structured automatically, so the owner isn't at a keyboard every evening. A triathlon club has an assistant grounded in its own information, so members get current answers about events and training without a volunteer fielding the same question forty times. A family app I run turns a photographed school note into a calendar entry.
That's the work. It's specific, it's narrow, and it pays.
There's a shorter list of things people ask me for that I steer them away from, and the reasons are consistent enough to write down.
A generic chatbot
Note the word generic, because this needs a distinction the sales pitches never make.
The off-the-shelf widget trained on nothing in particular, dropped into the corner of your site, capable of little beyond restating your homepage — that one usually exists because the website didn't answer the question in the first place. Someone lands wanting to know whether you cover their area and roughly what a job costs. If the site says so, they have it in four seconds and Google can read it too. If it doesn't, you're asking them to interrogate a machine for information you could have written down.
An assistant that actually knows your business is a different animal, and I build those. The difference is what it reads from: your own services, process, policies and stock, in an index that updates as that material changes, answering in your voice and saying it doesn't know rather than inventing. That's worth owning. It's also not what most people are quoted for.
**Instead:** answer the ten questions you genuinely get asked, properly, on the relevant pages. Then put an assistant on top of a clear site, rather than using one to compensate for an unclear one.
Generated website copy
This is the one I argue hardest about, and it costs me the occasional job.
Your site's job is to sound like the person the customer is going to deal with. That's most of what they're deciding — not whether you're competent, which they assume, but whether they want you in their house or handling their money. Generated copy is the statistical middle of everything already written on the subject, so it sounds like every competitor who did the same thing.
There's a practical problem too. Once a site is written that way, nobody in the business can maintain it, because there's no voice to match.
**Instead:** record yourself explaining what you do to a friend outside your industry. That transcript is better raw material than any prompt, and shaping it is a job AI is genuinely good at.
Automating something two people do differently
If two people do the same task in different ways and both think they're right, automating it doesn't settle the argument. It picks one at random, makes it permanent, and buries it where nobody can see a decision was made. I've been called in to fix the aftermath more than once. The tool works exactly as specified, and half the team quietly stopped using it in week three.
**Instead:** agree the process on paper first, with the people who do it. If you can't get agreement, that's the actual project.
Buying tools before you have somewhere to put the output
A tool produces summaries, drafts, transcriptions or leads, and there's no agreed place they go and nobody whose job it is to use them. So they accumulate, then stop being opened, and eleven months later it's still on the credit card.
**Instead:** before buying, answer three things. What does this produce, who looks at it, and what do they do differently as a result? If any of those is vague, wait.
Anything sold with a deadline
If the pitch involves a window closing or a discount expiring on Friday, the urgency is the product. The tools that genuinely help don't need it, because the thing they fix was already annoying you before anyone rang.
**Instead:** wait a fortnight. If it still seems like a good idea when nobody's pushing, it probably is.
The pattern
The things worth building tend to be boring and narrow. The things marketed hardest tend to be broad and exciting. That's not a coincidence — broad and exciting is easier to sell to someone who doesn't yet know what they need.
Stopping information being typed twice. Turning photographed paperwork into records. Making your own files and past jobs searchable in plain English. Being findable now that a large share of searches get answered without a click.
Unglamorous, specific, and they're most of what I do.