By Bryan McIntyre, DelegatedOS

You tried AI. You got an underwhelming result. And somewhere quietly, you decided it probably wasn't for your business.

I hear this constantly. Owners who had a go with ChatGPT, weren't impressed, and concluded AI is a bit overhyped and not really worth the bother.

The tool wasn't the problem. You were judging the whole thing on about 5% of what it can do.

I've watched this play out from both sides. We started DIY ourselves, so I know how easy it is to stall. And I've since sat with business owners from many different industries who'd written AI off completely, then watched them change their mind in a single afternoon.

So the question isn't whether AI works. It's whether you're better off learning it yourself or hiring an AI consultant to do it with you. This is the honest version of that answer, written for a NZ business owner, not a tech audience.

First, a definition, because the word gets thrown around loosely. An AI consultant is someone who works out where AI actually fits your specific business, then builds and installs it, so you get the result without having to become the expert yourself.

What can you realistically do with AI yourself?

More than you think. The first move for any owner is to learn how to prompt properly and understand roughly how these tools work. That skill alone puts you ahead of most businesses, and it costs you nothing but a bit of practice.

The bigger shift is mental. Most owners hit a problem and deal with it the way they always have. The AI-first habit is to pause and ask, "how could AI help me with this?" before you default to the old way. Skip that question and you miss most of the value.

There are three places AI helps. The first is the thinking layer, using tools like ChatGPT to solve problems and pressure-test decisions, which is where almost nobody starts. The second is agents and automation. The third is industry or task-specific tools, like Gamma for building slide decks.

DIY genuinely works for the simple end of the first bucket. Drafting an email. Thinking through a hiring decision. Tidying a document. Low stakes, and if it gets it wrong, you'll spot it in seconds.

One NZ-specific warning while you're there. These tools default to a foreign setting. The spelling comes out Americanised, and ask for marketing copy and it'll sound like it was written for a US audience, not a New Zealand one. You have to tell it plainly who it's writing for.

If you want a proper starting point, our honest guide to ChatGPT for small business walks through how to actually get value out of it.

So why do most DIY AI attempts stall?

Because tools get collected instead of problems getting solved. An owner signs up for four or five AI subscriptions, each doing a slice of something, none of them connected, and nothing in the actual workflow changes. You don't have an AI strategy. You have a subscription problem.

The numbers back this up. An MIT study in 2025 found that 95% of generative AI pilots never make it into production. And in the 2026 PwC Global CEO Survey, 78% of the 103 New Zealand CEOs surveyed reported little to no financial impact from their AI spend so far.

That's not a tools problem. New Zealand businesses are already using AI. The AI Forum NZ's August 2025 productivity report found 87% of organisations use some form of it, but only 12% have scaled it across the business. Nearly everyone has started. Almost nobody has finished.

Here's the pattern I see most. People use a generic tool like ChatGPT as their finishing point, when it's really a starting point. They'll try two, three, four different models but use them exactly the same way, and wonder why the results don't improve. The real leverage sits in AI trained on your business and the exact task you need it to do.

What does an AI consultant actually do?

A good consultant works out where AI belongs in your business, then builds and trains it around how you actually operate. It's a working system, tied to a real process, that produces the right output without you steering every step.

In practice, that looks like this. A client of ours needed to turn a quote, with hundreds of chargeable line items, into a clean scope of works the customer could actually read. No customer wants to see every screw and length of timber.

He tried hard to get generic AI to do it. It couldn't. So we built and trained an AI to handle exactly that transformation. He plugged it in and it worked instantly.

The job used to take him a good 60 minutes or more, on top of the time he'd wasted trying to force a general tool to do it. Afterwards it took about three minutes to generate, plus five to ten to check. That's his own before-and-after, on a task he ran constantly.

That's the difference. DIY got him a tool that nearly worked; trained AI got him the result.

It was never really a conversation about AI

If you forget everything else, remember this one thing. This isn't a conversation about AI. It's a conversation about your business and what you want from it.

You didn't get into business to learn prompting. You got into it to build something. The goal was never to become an AI expert. It was to hit the goals you set for the business in the first place.

DIY has a real place, and I'll always tell an owner to learn the basics. But the point of getting help isn't the technology. It's getting more of what you actually want out of your business, faster, without carrying yet another thing to master on your own.

That's the whole job of a good consultant. Getting you the result, not turning you into an AI expert.

The bottom line

You can do more yourself than you think. And there's a clear point where doing it yourself quietly costs you more than help would.