What an automation actually is, explained without the jargon

People ask me what I actually build, and when I say "automations", I can see the word slide right off them. It sounds technical. It sounds like something for big companies with IT departments.

It isn't. An automation is just a set of instructions that runs by itself. You already follow dozens of them every week. The only difference is that right now, a person is doing the steps.

Every automation has three parts

Once you see these three parts, you'll start spotting automations everywhere.

1. A trigger. This is the thing that starts it. A new form is filled in. An email arrives. A payment goes through. It's 9am on a Monday. Something happens, and that's the signal to begin.

2. The steps. This is the list of things that should happen next, in order. Add the person to the spreadsheet. Send them a welcome email. Let the owner know on their phone.

3. The result. This is what's different afterwards. The customer got an answer, the record is up to date, and nobody had to remember to do it.

Here's a real example from a gym. The trigger is someone filling in the free-pass form. The steps are saving their details, sending a friendly email and adding a reminder for the front desk. The result is that a new lead gets a reply in seconds instead of tomorrow.

So where does AI come in?

A plain automation is great at steps that are always the same. Copy this, send that, move this over there. It's fast, it's reliable, and it never gets bored.

What it can't do is use judgement. It can't read a message and decide whether the person sounds ready to buy. It can't write a reply that actually answers their question. It can't pick up the phone and have a conversation.

That's the part AI handles. In the systems I build, AI is one step in the chain, used only where a step needs a bit of thinking. It reads, decides or writes, and then the plain automation carries on with the rest.

So when you hear "AI automation", picture an ordinary checklist where a few of the steps are done by something that can read and write.

What the tools are called

You might hear names like Zapier, Make or n8n. They're all tools for building automations, and they all work the same basic way: you choose a trigger, then you chain the steps together. I mostly use n8n, because it can run on a private server that the business controls, and it handles the more complex systems well.

You don't need to learn any of them to get the benefit. That's my job.

When you shouldn't automate something

Not everything should be automated, and anyone who tells you otherwise is selling something.

• If it happens once a month, it's probably not worth it. Automations pay off on the jobs that repeat every day.
• If the process changes every week, settle it first. Automating a messy process just gives you a faster mess.
• If the relationship is the product, keep a person in it. An upset customer or a big, sensitive decision deserves a human.

The sweet spot is work that's repetitive, time-sensitive and a little bit annoying. Answering new enquiries, following up with people who went quiet, booking appointments and copying information from one app to another are the classic examples.

The quick test

Here's a question you can ask about any task in your business: "If I wrote down exactly how to do this, could someone follow it without asking me anything?"

If the answer is yes, it can almost certainly be automated. If the answer is "mostly, but they'd need to use their judgement now and then", that's where AI comes in.

If you've got a task in mind and you're not sure which group it falls into, tell me about it. I'll give you a straight answer.