Often not. When I review a business's software before quoting any AI work, a good part of the automation wish list often turns out to be already available inside software the business pays for every month, switched off or never tested. Check what you already own before you buy or build anything.

The list that didn't need building

A financial planning practice came to me a few weeks back with a list of automations they wanted AI to build: client onboarding steps, meeting follow-ups, document chasing, the usual. Before I quoted anything, I asked to see how they were using their existing tech stack.

That review showed a lot of the list could be handled by capability already sitting inside the practice management software they pay for every month. Not built by us. Not bought as an add-on. Already there, just switched off, not connected, or never tested because nobody had the time to investigate.

We hadn't built anything yet. We'd just looked closely at what they already owned. And the client was very happy.

Why businesses buy before they look

I see the same pattern in many of the AI conversations I have with business owners. The instinct is to buy the fix: a new tool, a new subscription, sometimes a whole new system. The audit step (what does the software you already pay for actually do?) gets skipped because it feels less exciting than the shiny new thing, and because most people have never been shown how to look.

My rule, on every assessment: nothing gets built or bought until I've checked what's already switched on. Some of the best AI work I do isn't AI work at all. It's turning on a feature someone paid for a year ago and forgot about, or never knew existed.

Try this week: test one feature you already pay for

Open the admin or settings menu on the one piece of software your business can't run without. Look for anything with "AI", "automate", "smart" or "suggest" in the name. Test one of them, once, on a task that costs you nothing if it gets it wrong.

Then make it stick

Before you commission or buy anything new from here on, make "what can our existing stack already do?" the first question you ask.

If you're about to spend on AI, ask what you've already got first, and whether anyone has actually looked.

Sources