AI won't run your business. It changes what you do.
Most AI advice is about tools. I think that's the wrong place to start. For an owner, the useful question is about judgement.
There's a new AI tool every week.
Trying to keep up with all of them is a losing game. Thinking clearly about AI is not.
This isn't from one book. It's my own thinking on how an owner should approach it.
Start with the work, not the tool
Most owners start by asking which tool to buy. I'd start somewhere else. Look at the work.
List the tasks in your business that are repeated, rule-based and easy to check. For example:
- First drafts of emails, proposals and posts
- Summaries of calls and meetings
- Answers to common customer questions
- Sorting and tagging leads
That's where AI fits today.
Trust, judgement and relationships stay with people. Don't hand those over.
Three questions before you start
Before any task goes to AI, I'd ask three things.
- If this goes wrong, who notices, and how fast? AI can make confident mistakes. Start where errors are cheap and easy to catch.
- What does my team do with the time saved? If you don't decide, the time just disappears.
- What data am I handing over? Treat an AI tool like any outside vendor. Ask the same questions you'd ask them.
If you can't answer all three, the task isn't ready yet.
The real shift is in your role
Here's what I think most AI advice misses.
When drafting and sorting get cheap, judgement becomes scarce. And judgement is the owner's job.
Which draft is right. Which lead deserves a call today. What to say to the client who's about to leave. No tool decides those well. You do, or your best people do.
When the work gets cheap, judgement gets expensive.
So AI doesn't take the owner out of the business. It moves you. Less time producing. More time deciding.
That's the change worth planning for.
A sensible way to start
Keep it small and keep it honest.
- Pick one repeated task this month. Just one.
- Run AI alongside the current way for two weeks. Don't switch over yet.
- Compare quality, not just speed. Faster and worse is not a win.
- Keep it only if customers wouldn't notice, or if it got better.
Say, hypothetically, your team writes twenty follow-up emails a week. Let AI draft them for two weeks while the team still writes their own. Put both side by side on Friday. The answer usually becomes obvious.
"Are we using AI?" is the wrong question. A yes or no tells you very little.
The better question is this: are our best people spending their judgement on what matters most?
If AI helps you answer yes, use it. If it doesn't, it's just another tool.
Try this week
Write down the five tasks that take up most of your own week. Mark each one: judgement or routine.
Pick one routine task. Run the two-week test on it. Then give the time back to something only you can decide.