You know you need AI, but where do you start?

Most business owners I talk to around the Bay say something like this: “I know we should be using AI. I just don’t know where to start.”

That’s a completely normal place to be. AI is everywhere in the news. Every software company says it has AI now. Your nephew uses ChatGPT for his homework. And you’re running a business, so you don’t have spare hours to work it all out.

The good news is you don’t need a big plan, a big budget or a technical team. You need one small win. Here’s the simple five-step approach I use with businesses from Katikati to Whakatāne.

Step 1: Find the job that eats the most time

Don’t start with the technology. Start with your team’s week.

Ask everyone one question: “What do you do every week that feels like a waste of your skills?” You’ll get honest answers fast. Things like:

• typing the same email reply over and over
• writing up quotes at night after a day on site
• copying details from one system into another
• answering the same customer questions on the phone
• writing job notes, reports or meeting minutes

Write the list down. Next to each job, add a rough guess of how many hours it takes each week. You now have something more useful than any AI strategy: a list of real problems, in order.

Step 2: Pick one job, not ten

This is where most businesses go wrong. They get excited, try AI on everything at once, and end up with a mess nobody trusts.

Pick one job from your list. The best choice is usually the one that:

• takes a lot of time each week
• is fairly low-risk if something goes wrong
• someone on your team is keen to fix

That last point matters more than you’d think. A person who is fed up with a task is the perfect person to test a new way of doing it. They know what “good” looks like, and they’ll tell you straight away if the AI isn’t up to it.

Step 3: Try it with a person in charge

Now try AI on that one job, with one simple rule: AI does the first draft, a person checks it.

For example, if the job is replying to common emails, the AI writes a draft reply. Your team member reads it, fixes anything that’s off, and presses send. Nothing goes to a customer without a human looking at it first.

This does two things. It keeps your quality high while everyone learns. And it builds trust, because your team can see exactly what the AI is doing, rather than worrying about it happening behind their backs.

Use a business version of the tool, not a free personal account. Business versions of tools like ChatGPT, Claude and Microsoft Copilot give you more control over your data. I explain why in my post on whether AI is safe for your business.

Step 4: Measure it, honestly

After two or three weeks, sit down with the people using it and ask:

• Is it saving time? Roughly how much each week?
• Is the quality as good as before, or better?
• What’s annoying about it?
• Would you go back to the old way?

Be honest with the answers. Sometimes AI is a clear win. Sometimes it saves a little time but needs tweaking. And sometimes it’s just not the right fit for that job, and that’s fine too. You’ve learned something without spending much, and you can try the next job on your list.

When it works, you’ll usually notice it in small, human ways first. Quotes go out the same day. The inbox is clear before morning tea. Someone leaves on time for once.

Step 5: Write it down, train the team, then do the next one

Once a job is working well, write a short, one-page note about it:

• what the AI does, and what it doesn’t do
• which tool you use, and what information goes into it
• who checks the results
• who to ask when something goes wrong

Then make sure everyone who needs it knows how to use it. Not just the keen early adopter. Everyone. AI works best when the whole team is comfortable with it, and nobody feels left behind or nervous about it.

Then go back to your list and pick the next job. One at a time, each one proven before you move on. In a few months you’ll have a handful of jobs running smoothly, a team that trusts the tools, and a much clearer idea of where AI could help next.

What not to worry about

A few things people often stress about at the start, that don’t need to slow you down:

• You don’t need new software. Most AI help fits inside the tools you already use, like Gmail, Outlook, Xero or Google Sheets.
• You don’t need to understand how AI works. You need to understand your business. That’s the hard part, and you already know it.
• You don’t need to be first. You just need to start. Small and steady beats big and rushed.

The short version

Find the job that eats the most time. Fix that one job, with a person in charge. Measure it, write it down, train the team. Then do the next one.

That’s it. No big bang, no jargon, no ripping out your systems.

If you’d like a hand with step one, that’s exactly what my free AI chat is for. We spend 30 minutes at your place or on Zoom, talk about where your team’s time goes, and you leave with three clear first steps, whether you work with me or not.

Book a free AI chat