You're Not Behind. You're Just Waiting for Someone to Show You the Door.

A practical, no-hype guide to starting with AI, from someone who learned it between school runs and client calls.

A sunlit open door in a soft plaster hallway

There was a moment, maybe eighteen months ago, when I realised that every piece of content I was reading about AI was written by someone who had nothing to lose from me feeling overwhelmed.

The developers writing technical breakdowns. The VC-backed founders launching tools I'd never use. The content creators cranking out "10 AI prompts that will change your life" threads while quietly being paid to promote the very platforms they were reviewing. None of them were writing for me: someone with a non-tech background who wanted to learn but found it all a bit overwhelming.

The information was everywhere. A genuine starting point was nowhere.

So this is that starting point. Written by someone who came to this not as a technologist, but as a working parent running her own business from the west of Ireland, with twenty-odd years of marketing and consumer data behind her and precisely zero desire to become an AI guru. I'm not here to convert you. I'm here to show you the door that nobody pointed me toward.

What AI actually does in my working week

Not in theory. In my actual working week, between a school run and a client brief, with a coffee going cold on the desk.

I use AI to think out loud when I'm stuck. I paste in a half-formed idea that came to me while out walking the dog and ask it to push back, spot the gaps, or suggest three different angles I haven't considered. It's not a search engine. It's closer to having a well-read colleague in the room who has no ego about your draft.

I use it to do the first pass on things I find tedious: repurposing a long article into Pinterest descriptions, turning a voice memo into a structured outline of a blog, drafting an email I've been putting off because the tone felt tricky. It gets me to 70%, and then I take it from there.

I use it to research quickly and critically. Not to accept what it tells me without checking (we'll come to that), but to get a working understanding of something fast, then go and verify the parts that matter.

What it does not do: write my posts for me, make my decisions, or, critically, replace the judgement I've spent two decades building. I know what feels right.

What I tried and quietly abandoned

Prompt libraries. I downloaded several and I used almost none of them. The prompts that work are the ones you build yourself, for your specific situation, over time. Someone else's library is just someone else's thinking. But don't let that intimidate you. Your AI tool of choice will help you build great prompts, you just have to ask.

Chasing every new tool. There is always a new tool. Every day something new comes out. The people telling you this week's release will change everything are usually the people who told you the same thing about last week's. Pick one tool, learn it properly, and only look up when you hit a wall it genuinely can't help you over.

The one shift that made it actually useful

I stopped treating it like a search engine and started treating it like a conversation.

When you type a short question into an AI tool, you get a generic answer. When you give it context (who you are, what you're trying to do, what you've already tried, what good actually looks like for your business), you get something entirely different. Useful. Specific. Worth acting on.

Think of it like briefing someone properly instead of barking a one-liner at them and wondering why the output is mediocre.

The quality of what you get back is almost always a mirror of what you put in.

The other shift: I stopped expecting it to be right and started expecting it to be useful. These are not the same thing. AI will occasionally tell you something confidently incorrect. That's not a reason to distrust it entirely. It's a reason to verify anything that matters. Treat the output as a capable first draft from someone who hasn't met your clients, doesn't know your market, and has never felt the specific Friday afternoon panic of a deadline you underquoted.

What you don't need to know to start

You do not need to understand how large language models work. You do not need to know what a token is. You do not need API access, a developer, a course, or a prompt framework.

You do not need to have read everything that's been written about AI in the last two years. (Honestly, most of it will either age badly or be irrelevant to your actual situation.)

You do not need to be an early adopter. The tools are more capable and more accessible now than they have ever been. Showing up today is not too late.

What you do need: curiosity, patience with your first few attempts, and a genuine problem in your business that you're trying to solve. Start with the problem, not the technology.

Where to actually begin

Start with ChatGPT. I know that might surprise you coming from someone who now works primarily in Claude, but ChatGPT is where I started and it's still the easiest on-ramp. The interface is intuitive, the free tier is very functional, and there is more beginner-friendly content written about it than anything else out there. It is the best place to learn how to have a useful conversation with an AI tool, which is the foundational skill for everything else.

Once you are very comfortable with ChatGPT and ready for more, that's the moment to move to Claude. Not before. You'll know when you're ready because you'll have a specific frustration that a better tool can solve.

The tool might change but the skill of directing it won't. Learning how to communicate clearly with an AI, how to give it useful context, how to push back on a weak output and iterate toward something better: that transfers across every tool, every update, every new release. Build the skill, not a dependency on a specific platform.

One last thing: start with something small and real. Not something hypothetical, but a thing you actually need done this week. An email you've been avoiding, a social caption that isn't landing, an offer description you can't quite articulate. Use it for that. Notice what works and what doesn't. Adjust and give feedback.

That's it.

One question before you go

What's the one thing in your business you've been thinking AI might help with, but haven't tried yet?

Tell me. Your answer might just become next week's post →

← Back to The Edit Where does AI fit in your business? →