You've heard about it at dinner. You've seen it in the news. Maybe your kid won't stop talking about it. But you've never actually tried it.

Starting is easier than you think

Easier than setting up a new phone. You don't need to know anything about technology, and you don't need anyone to walk you through it. If you can send a text message, you already have every skill required.

That's worth saying plainly, because the people who put this off the longest aren't the ones who couldn't figure it out. They're the ones who assumed it would take an afternoon they didn't have. It takes five minutes, most of which is typing your email address.

Step 1: Pick one and go

Two AI assistants have free versions and run in a regular web browser, nothing to download. ChatGPT at chatgpt.com and Claude at claude.ai. Create a free account with your email and you're in.

Don't overthink it. The tool matters less than starting. People spend more time comparing options than they'd spend just picking one and finding out what it can do. You can always try the other one next week. Neither choice is a mistake.

Step 2: Talk to it like a person

When you're looking at that empty text box, just type. Plain sentences. The way you'd explain something to a colleague down the hall. No special commands, no tricks, no exact phrasing you need to get right.

Try something low-stakes:

Or something entirely your own. Hit Enter and see what happens.

Why so many capable people freeze at the blank box

I find this part interesting, because the people who hesitate most aren't the ones who struggle with technology generally. Plenty of business owners run circles around complicated equipment, juggle five subcontractors and a shifting schedule in their head, and price a job to the dollar without blinking. Put the same person in front of an empty AI chat window and they freeze.

Part of it is that there's no obvious first move. A track saw has a trigger. A quote sheet has fields. A blank text box has nothing, and nothing feels like it could go wrong if you type the wrong thing. It can't. There's no wrong first message. The worst outcome is a mediocre answer, and you just try again with a better question.

The other part is that most of us learned software by pushing buttons in a fixed order, and this doesn't work that way. There's no button. You just say what you want, in your own words, however you'd normally say it. That's genuinely the whole trick, and it feels too simple to be the real answer until you've done it once.

I've watched this happen in person more than once. Someone stares at the box for a solid thirty seconds, types a hesitant half-sentence, deletes it, and finally just asks the thing they actually wanted to ask. The answer comes back reasonable, occasionally impressive, and the tension in the room drops immediately. Nobody's ever told me afterward that it was harder than they expected. Every single time, it was easier, and slightly anticlimactic in a good way.

The hesitations I hear most, and why they don't hold up

"I'm not a computer person." Neither were most of the people I've watched use this well. It's not a computer skill. It's a conversation skill, and if you've ever explained a job to a new apprentice or talked a client through a change order over the phone, you already have it.

"I don't have time to learn a new system." You don't have to learn one. There's no manual, no certification, no settings menu you need to master before it's useful. The five minutes it takes to create an account is close to the entire learning curve. Everything after that is just typing sentences.

"It'll probably get it wrong anyway." Sometimes, yes, and that's worth knowing going in rather than finding out the hard way on something that mattered. For a first conversation about something low-stakes, being occasionally wrong costs you nothing. You'll get a feel for where it's solid and where it isn't long before you're relying on it for anything that counts.

"My business is too specific for a generic tool like this to help." This one comes up constantly from trades and construction owners, and it's usually backwards. A general-purpose AI assistant won't know your supplier pricing or your crew's schedule, true. But drafting an email, summarizing a document, or explaining an unfamiliar term doesn't require industry-specific knowledge. Those are exactly the tasks a general tool handles well, regardless of what kind of business you run.

"Isn't this a bit like cheating?" I get this one from owners who grew up being graded on doing everything themselves, and I understand the instinct. But nobody thinks using a calculator to add up a materials list is cheating, or using a level instead of eyeballing a wall. This is the same category of tool. It doesn't do the job for you. It removes the part of the job that was never the actual skill in the first place, the retyping, the blank-page staring, the first-draft grind, so you can spend your attention on the part that is.

If typing isn't your thing

Not everyone wants to sit and type sentences, and you don't have to. Most phone keyboards have a microphone icon that turns speech into text, which means you can talk your way through a message the same way you'd leave a voicemail, and let the AI clean it up from there. Both ChatGPT and Claude also have mobile apps, if a phone feels more natural to you than opening a browser on a laptop at the end of a long day. There's no wrong device for this. Whatever you'd actually use is the right one.

Have a conversation, not a transaction

Most first-timers ask one question, get an answer, and stop there. Keep going. That's where it gets useful.

If the answer isn't quite right, say so. "That's close, but I was thinking more about..." or "Can you make that shorter?" The AI adjusts. Think of it less like a search engine that hands you a list of links and more like a knowledgeable colleague who actually tries to help, and who doesn't mind being corrected.

This is the single biggest difference between people who get real value out of these tools and people who try it once and shrug. The ones who stick with it treat the first answer as a rough draft, not a verdict. Ask a follow-up. Push back. Ask it to try again from a different angle. A little back and forth turns a decent answer into a genuinely useful one, and it takes less time than most people expect.

The first answer is a draft. The conversation is where it actually gets good.

What it's good at, and where it struggles

First draft of a letter? Good fit. Plain-English explanation of a confusing document? Also good. Ideas you haven't thought of, a second opinion on a decision, a way to organize a messy pile of information into something readable? All solid.

Where it struggles is anything that needs to be exactly right without you checking it. AI can state incorrect information with complete confidence, so for medical, legal, financial, or technical questions, especially anything involving a code requirement, a material spec, or a number that's going into a quote, treat the answer as a starting point and verify it elsewhere. More on that next week.

A few first uses worth trying in a trades or construction business

If you want something more relevant than a general knowledge question, here's where the early wins usually show up:

None of these require any setup. They're exactly the kind of low-stakes, high-annoyance tasks that eat twenty minutes here and there all week, which makes them a good place to build the habit before you touch anything that actually needs to be precise.

Pick whichever one on that list annoys you the most right now. That's usually the best place to start, not because it's the easiest, but because you'll actually notice the time it saves. A task that doesn't bother you won't teach you much about whether this is worth your time. A task you dread every week will.

What happens after the first conversation

The habit builds faster than people expect. Most owners I talk to describe the same rough pattern: the first week, they use it for one or two things out of curiosity. By the third or fourth week, they've stopped thinking of it as a separate task and started reaching for it the way they'd reach for a calculator, without deciding to.

That shift matters more than any individual use case. The value isn't really in the first email it drafts for you. It's in reaching the point where you instinctively know which tasks are worth handing off and which ones aren't, so the tool becomes background instead of a novelty you tried once and forgot about. That instinct only comes from actually using it, a little, regularly, on things that don't matter much. It doesn't come from reading about it, which is a strange thing to admit in an article, but it's true.

If you hit a point where a task feels like it should be doable but the answer keeps missing what you actually meant, that's usually a sign the question needs more context, not that the tool has failed. Add a sentence about what you're trying to accomplish and why. Mention who's going to read the result. The difference between a mediocre answer and a useful one is often just a bit more of the context you'd naturally give a person.

One thing to do this week

Open a browser, go to claude.ai or chatgpt.com, and have one real conversation. It doesn't need to be work-related. Ask something you've been curious about for years and never bothered to look up, or pick one of the prompts above and see what comes back.

You might be surprised how natural it feels. Most people who put this off imagined something closer to programming than to typing an email, and the gap between that expectation and the actual experience is usually the biggest surprise of the whole exercise.

It won't feel life-changing the first time. It'll feel like a decent conversation, maybe a mildly useful email draft, and that's exactly what it should feel like. The businesses that end up getting real value out of this aren't the ones who had one dramatic breakthrough conversation. They're the ones who did this small, unremarkable thing once, then did it again the next week because it had quietly saved them ten minutes, and kept going from there.

If you get through that and start wondering where AI could actually save your business time, not just as a novelty but as something built into how you quote, document, and communicate, the free assessment is a good next step.

Next week: "Don't Believe Everything the Robot Says." How to tell when AI is confident versus when it's guessing, and what to do about it.