Start With One Thing

Someone will tell you this year that you need an AI strategy.

You don't. You need a Tuesday with two fewer hours in it.

That distinction is the whole of this issue, and it's where the season lands — because the failure mode of everything we've written since August is that it becomes a project. Projects don't get finished by people who are already at capacity. What gets finished is one small thing that visibly works.

The season, in five questions

If you're weighing something up — a tool, a quote from someone like us, a free trial you're three days from being charged for — run it through these. It's the five issues, compressed.

Does it give me hours back, or a new thing to manage? If you can't name what it removes from your week, nothing else matters. (Issue #6)

Is it one of the four boring jobs? Same-thing-with-variations, out-of-hours answering, mess into structure, first drafts. Outside that band, be sceptical. (#7)

Where's the checkpoint? What it may say alone, what always comes to a human, and who owns it when it says something daft. (#8)

Have I done the real sum? Subscription plus setup plus correction, against hours actually returned. (#9)

Does it sound like me? And has anybody given it your actual words to work from. (#10)

Passes all five — go ahead. Fails one — you've just found the thing to fix before you spend anything, which is worth more than a yes.

Then pick one. Just the one.

Go to the top of your repeats list. Not the most interesting item — the most frequent one. The dullest thing you do fourteen times a week.

That's the one. Not a plan for the business, not a review of your systems, not a workshop. One task, and it wants to be small enough to feel slightly disappointing.

Small and dull is the whole strategy, because small and dull finishes. Ambitious gets abandoned in week three when a job goes wrong and everything sensible goes out of the window — and then the story you tell yourself for the next two years is “we tried AI, it didn't work for us”. It wasn't the AI. It was the size of the bite.

Give it thirty days, and measure one thing

Thirty days is long enough to get past the fiddly part and short enough that you'll actually check.

And measure one thing only — not whether the technology is impressive. Did the task stop landing on you?

That's it. Not accuracy. Not how clever the output is. Not whether a newer one came out last week. Did the work leave your plate, and stay off it.

If yes, you've got the hour back and — more usefully — you've got a template. The second one is always easier than the first, and the fifth is very nearly free.

If no, kill it. No ceremony, no sunk-cost speech. You've lost a month and learned where the real bottleneck sits, which is cheap information. Nine times in ten it isn't the tool: it's that the knowledge it needed was still in your head, or that nobody owned it. Both are fixable, and neither is solved by trying a different app.

What this was all for

Season one was about seeing your week honestly. This season was about whether the most-hyped tool in a generation actually helps with any of it.

The honest answer, from people who are paid to build the stuff: sometimes, in a narrow band, if it's finished properly and somebody owns it. That's a far smaller claim than the adverts make. It's also a much more useful one, because it's the only version you can act on without getting burned.

Everything else is as it always was. Name the leak. Take the dullest repeated thing. Hand it over properly, with a checkpoint on anything that matters. Then keep the work that needs your judgement, and guard it — because that was the work you started the business to do, and it's the first thing to get buried.

The tools change every eighteen months. That part hasn't changed at all.

This week's homework: name it, and start Monday

One sentence, written where you'll see it.

For the next thirty days, this task does not land on me: ______

Fill in the blank. Put a date on it. Tell somebody, so it's real.

That's the whole homework, and it's the only one this season that costs you anything — the other five were about looking, and this one's about doing.

If you don't yet know which task to pick, the Founder Audit will tell you where your week leaks hardest — a couple of minutes, and it comes back with which of the four to fix first. And if you'd rather talk it through, book a demo and we'll tell you straight whether the thing you're considering is worth building, buying, or ignoring.

Thanks for reading this season. Genuinely — the replies have been the best part of it. Keep them coming: what's the one task you're giving thirty days? Tell us and we'll tell you honestly whether we think it'll stick.

Built by operators, for operators.

— Kane & George