AI is being pushed at small businesses from every direction right now. Most of it is solving problems you don't have. The uncomfortable truth is that the best system is rarely the most sophisticated one — it's the one your team will actually use on a busy Tuesday.
There's a particular kind of regret that small-business owners know well. You bought the tool. You watched the demo, you could see how it would change everything, you paid for the annual plan. And six months later, half the team has quietly gone back to the spreadsheet, and the shiny new platform is just another tab nobody opens.
The shelfware problem
It happens constantly. A business reaches for software to fix a friction point, and a big share of what gets bought ends up as shelfware — paid for, logged in once, never adopted. The money stings, but the real cost is subtler: the original problem is still there, now wearing a thin coat of "we tried that."
Why does it happen? Almost never because the tool was bad. It's because a tool isn't a system. Software is just one ingredient. A system is the whole recipe — who does what, when, in what order, with what information, and what happens next. Drop a powerful tool into a fuzzy process and you don't get clarity. You get a more expensive version of the same confusion.
The best systems aren't the most sophisticated ones. They're the ones that stick.
The AI version of the same mistake
Right now the pressure is amplified, because the tool everyone's being sold is AI. There's a real fear of being left behind — so businesses bolt on an AI feature, an AI assistant, an AI something, without first asking the only question that matters: what problem, exactly, is this solving for us?
AI is genuinely useful. I use it, I build with it, and for the right task it's transformative. But "we should be using AI" is not a plan — it's anxiety with a budget. Plenty of the friction in a small business isn't an AI-shaped problem at all. It's a missing checklist, an undefined handoff, or a form that three people re-type. No model fixes those. A clear system does.
The test that actually matters
Before you adopt any tool or automation — AI or otherwise — run it through one question: will the people doing the work actually use this when they're busy and under pressure?
If a system is even slightly harder than the old way on a bad day, people will route around it the moment things get hectic — and a bad day is exactly when you need the system most. So the bar isn't "is this impressive?" It's:
- Is it easier than the current way? Not eventually — immediately, for the person doing it.
- Does it fit how they already work? The best automation is often invisible; it happens in the background, not in yet another app to check.
- Does it survive a busy week? If it only works when everyone's calm and remembering to follow it, it isn't a system. It's a suggestion.
A simpler starting point: before buying anything, write down the actual process as it happens today — messy reality, not the tidy version. Nine times out of ten, the fix becomes obvious, and it's smaller and cheaper than the tool you were about to buy.
Start with the work, not the tool
The right order is the opposite of how it's usually done. Don't start with "which tool should we get?" Start with "how does this work actually flow, and where does it snag?" Once the system is clear, the question of whether you need software — and which kind, and whether AI belongs anywhere in it — usually answers itself. Often the answer is a smaller, simpler change than anyone expected.
That's the whole philosophy behind how I work: solidify and optimise the systems a business already runs on, and only reach for a new tool when it genuinely earns its place. No buzzwords, no shelfware, no adding to the noise — just systems people actually use.