The Boring Four

The reason AI feels like hype is that the good bits are boring.

Nobody makes an advert about answers the same question four hundred times without once getting fed up. So the adverts are about robots and strategy and the future of work — and the actual useful thing, the small dull repetitive thing, never gets a mention.

Which leaves operators with a wildly inflated idea of what it does, and no idea at all what it's for.

It isn't a brain. It's a very specific kind of good.

Strip the marketing off and what's left is narrow. It isn't thinking. It isn't deciding. It doesn't know your business, your customers or what you charge unless somebody has told it — and it will never volunteer that nobody has.

But inside that narrow band it's genuinely, usefully good. And the band maps almost exactly onto the four leaks we spent last season pulling apart, which is the only reason we're bothering with it at all.

One: the same thing, with small variations

The email you've written two hundred times, adjusted for who it's going to. The quote that's the same shape every time with different numbers in it. The follow-up. The booking confirmation. The “sorry, we're full that week — could we do the Monday after?” message.

This is the single biggest win available to most small businesses and it is the least exciting sentence in this newsletter. It's the Twice note from Issue #2, handed over. If you did that homework, your list of candidates is already written.

Two: being awake when you aren't

The enquiry that lands at ten past eleven on a Sunday doesn't need a genius. It needs an answer — do you cover this postcode, roughly what does that cost, could someone come out this week — while the person is still holding their phone.

That's response latency from Issue #3, and it's where the gap between doing nothing and doing something small is enormous. Not because the reply is clever. Because it's there.

Three: turning mess into structure

This one surprises people. Point it at a rambling two-minute voice note and get a tidy job spec out of the other end. A photo of a supplier invoice, and get the line for your spreadsheet. Six months of customer messages, and get an honest list of the ten questions you're actually asked — which is usually nothing like the list you'd have guessed.

Every operator has a pile of unstructured mess: the WhatsApps, the notes app, the scribbled measurements on the back of something, the inbox. It's full of information about your own business that you've never been able to read as a whole, because it isn't in a shape anyone can read. This is the bit that quietly does the most for knowledge captivity — Issue #4 — because it lets you get things out of your head without sitting down to a documentation project you'll never finish.

Four: the first draft of anything

Not the final version. The first one.

The blank page is where the hours go. The website copy you've meant to redo for two years. The terms you keep saying you'll write down. The reply to the complaint you're dreading opening. Getting from nothing to something-mediocre is the expensive part — and that turns out to be precisely the part a machine can do.

Then you fix it, because the mediocre version is not good enough to send and never will be. But you've skipped the hard bit, which was starting.

And one thing that isn't on the list

Judgement.

Pricing the awkward job. Deciding whether this customer is worth keeping. Knowing from thirty seconds on the phone that this one's going to be a problem. Whether to eat the cost on a job that went wrong to protect a relationship worth ten more.

None of that is on the list, and anyone telling you otherwise is selling. The four things above are the work anyone could do. What's left is the work only you can do — which was the whole point of last season.

This week's homework: sort the pile

Take your repeats list — the Twice note from Issue #2, or just five minutes and an honest memory. Put every item into one of four buckets.

Same thing, small variations. Needs answering when I'm not there. Mess that needs a shape. Blank page.

Anything that lands in a bucket is a live candidate — a real job for a machine, today, with off-the-shelf tools you could sort yourself this month.

Anything that won't go in a bucket is almost certainly judgement. Leave it alone. Protect it, in fact — it's the good part of your job, and it's the first thing that gets squeezed when the other four fill your week.

Most operators find the great bulk of their repeats land in bucket one. That isn't a disappointing result. That's a very cheap fix hiding in plain sight.

The Founder Audit scores all four dimensions in a couple of minutes, if you'd rather see which is costing you most before you sort anything.

And hit reply — we read every one. Which of the four buckets does most of your week fall into? We're keeping a tally, and it's already changing what we build first.

Built by operators, for operators.

— Kane & George