J D MillerDesignLocal · Found · Booked
← All posts

AI for small business

The AOL stage

By Joel Miller · · 6 min read

In 1997, a lot of people thought AOL was the internet.

You can't blame them. AOL made it easy. The CD showed up in your mailbox, sometimes every few weeks. You put it in, listened to the modem screech, and a friendly voice told you that you had mail.

Once you were in, everything was laid out for you. News. Weather. Sports. Chat rooms. Shopping. All of it sat in neat little boxes with big buttons.

Even TV commercials sent you there. Plenty of companies didn't end their ads with a web address. They ended them with an AOL keyword.

So for a lot of families, the internet and AOL were the same thing.

They weren't.

Outside AOL's walls, something much bigger was going on. People were building their own websites. Small shops were selling to customers three states away. Search engines were sorting through millions of pages. New jobs were showing up that didn't have names yet.

The people who stepped outside the walls early learned how the web really worked. They figured out what a domain was, how a page got built, and how a stranger found you in a search. By the time everyone else caught up, they had years of practice.

The people who stayed inside AOL still got their email. Nothing bad happened to them. They just fell behind a little at a time, and most of them didn't notice until it was obvious.

We're doing it again.

Today, a lot of people think ChatGPT is AI.

Using ChatGPT today is a lot like using AOL in 1997. It's a real way in. It's just a small part of what's out there.

And again, you can't blame them. ChatGPT made it easy. You type a question and you get an answer. It's the first AI most people tried, and for many it's still the only one.

ChatGPT is a good tool. Nothing here is a knock on it.

But it's one tool. It's the front door to a very big house.

Here's some of what's in the rest of the house.

Some AI tools work right inside your files. You don't copy and paste into a chat box. You point the AI at a folder, and it reads your documents, your notes and your past work before it writes a word.

Some tools learn how you do a job and do it that way every time. You write down your process once, the way you'd train a new hire, and the AI follows it. Your voice and your rules, without you repeating them.

Some tools check each other's work. One AI writes the draft. A different one looks for the holes.

Some make images. Some build web pages, clean up spreadsheets, sum up a long video, or turn a meeting recording into a to-do list.

And some don't look like a chat window at all. They run in the background and handle the dull parts of the week while you do the part that needs a person.

If all you've ever done is type questions into ChatGPT, you've seen the lobby.

Here's a simple test.

When you use AI, do you start from a blank chat every time? Do you explain who you are, what you do and how you like things done, over and over? Do you copy the answer out, fix the same problems by hand, and then do it all again tomorrow?

That's the AOL stage.

It works. Sort of. But you're doing a lot of the work the tool should be doing for you.

The people further along made a different choice. They taught the AI once and stopped repeating themselves. They gave it their real files instead of a pasted paragraph. When it got something wrong, they fixed the instructions, so it didn't make that mistake again.

That's the gap between asking AI questions and putting it to work.

So what's the problem with sticking to the one tool you know?

Time.

Every week, the person who set AI up properly gets a little faster. Their tools get a little better at knowing how they work. The person who starts from a blank chat starts from zero every single day.

The gap grows quietly. That's what makes it easy to miss.

It went the same way with the web. In 1997, the difference between an AOL user and someone who could build a web page didn't look like much. A few years later, one of them had a skill businesses were paying for, and the other was still waiting for the CD.

You don't need to be technical to get out of the AOL stage.

Most of the people who left AOL in the 90s weren't programmers. They were curious. They clicked on things. They asked a friend how something worked, then tried it themselves.

Same here. A good start looks something like this.

Try a second AI tool this week. Give it the same task you'd give ChatGPT, and notice what's different.

Pick one job you do over and over, like a weekly email, a client proposal or a social post. Write down how you like it done. Give those instructions to the AI and see how close it gets.

When it misses, don't just fix the draft. Fix the instructions.

None of that takes a weekend. Most of it fits in a lunch break.

Collecting tools won't get you very far. Learning how the house is laid out will.

AOL was a fine way to meet the internet. It was a bad place to stay.

ChatGPT is a fine way to meet AI.

Don't stay in the lobby.

Want a guide for the rest of the house?

Tell me what you do every week, and I'll show you where AI can take some of it off your plate. No CD required.

Let's Talk →

Get new posts by email

About once a week. Short reads on getting found online and putting AI to work in a local business. No spam, and you can leave anytime.