The night AI ruined a perfectly good networking event

Someone said the word "AI" at a networking night last month, and I watched the whole room's interest get piqued at once. Not loudly. More like everyone quietly leaning in.

One business owner started talking about a scam email that had nearly caught out his bookkeeper. Another admitted, a bit sheepishly, that she'd just been avoiding the whole topic. A third laughed it off, but not really. You know the laugh.

Three different reactions, same room, same conversation. That's the thing about AI right now, everyone's starting from a completely different place with it, and most of the advice out there is written as if we're all standing in the same spot.

We're not. So let's sort that out properly.

The fear isn't really about the technology

I get why this feels uncomfortable, and I don't think that discomfort means anything's wrong with you. Strip it back and the anxiety usually isn't about AI or automation as concepts at all.

It's about not having answers to some very ordinary questions. What happens to the information you type in? Can the system be trusted? Will you make a mistake you don't even know how to spot?

None of that makes you behind. It makes you someone who hasn't been given a straight answer yet.

Ignoring the topic won't make you cautious, it'll just leave you unprepared. Signing up to five tools in a weekend isn't the fix either. The steadier path sits in the middle, understand where the real risks sit, decide what you're comfortable with, and put a few clear boundaries around it.

AI is a tool. It's not a plan.

AI gets talked about like it's a complete business strategy you can buy in a box. It isn't. It's a tool, and like any tool, its value depends on how it's used and who checks the work afterwards.

It can draft a reply, summarise a pile of notes, or knock a rough idea into some kind of shape. What it can't do is understand your business the way you do, know your regulars by name, or take the blame when something goes sideways.

That part is still you. It was always going to be you. The goal was never to hand over control, it's to lift some of the repetitive weight off your shoulders while you keep both hands on the wheel.

What it looked like when someone got this right

A retailer I know in Lilydale had heard AI could help with customer enquiries and stock planning. She wasn't about to bet the shop on it straight away.

She was worried about feeding sensitive information into a system she didn't fully understand, and worried automated replies might sound cold, or worse, tell a customer something untrue. Fair worries, both of them.

So she started small, with one task. She used AI to draft first-pass replies to the questions her team fielded over and over, the ones they could practically answer in their sleep.

Nothing went out automatically. Every message got a human read first. And the team set one rule from day one that never moved: payment details, passwords and anything confidential stayed well away from the tool.

Nothing about it was dramatic. There was no transformation montage. The team clawed back real hours, replies got more consistent, and she felt more in control, not less, because she was still the one making the calls.

That's what sensible AI adoption actually looks like. Quieter than the hype suggests, and a fair bit less interesting to post about on LinkedIn.

You're probably not starting from the same place as the business next door

A lot of the pressure comes from the idea that every business has to adopt AI right now, across the board, or get left behind. That idea just isn't true, and it ignores something fairly obvious: you're not starting from the same place as the retailer down the road, or the tradie you had a coffee with last week.

You don't need to use every tool. You don't need to automate every process. You definitely don't need to touch the parts of your business that already run fine without it.

Instead, ask where you're actually losing time. Are you writing some version of the same reply over and over, whether that's an email, a text or a DM? Are you re-explaining the same job details every time a new enquiry lands?

Is there a pile of scribbled notes from a job sitting there un-actioned, because turning it into something useful feels like a whole separate job of its own? Is admin quietly eating into hours that should go to customers, or to the tools in your hand?

That's your shortlist, and it'll look different to the next person's. Start with something low-risk and repetitive where you can check the output before it goes anywhere near a customer.

Protect what actually matters

Most of what makes AI safe or risky comes down to one habit: think before you paste. Ask whether what you're about to share is confidential, whether it identifies a real customer, employee or supplier and whether you'd be comfortable with it sitting outside your business.

Often you can strip the sensitive part out and keep the task. Use a general description instead of a real name. Give the tool the context it needs, not the whole file, the whole quote or the whole contract.

Keep passwords, payment details, private records, contracts and health information well away from public AI tools. The less you feed in, the less there is to worry about later.

Scams have had a bit of a glow up

Here's the part that worries me more than the tools themselves. AI has made it a lot easier for scammers to sound polished. Bad spelling used to be the giveaway, and it's not a reliable warning sign anymore, more's the pity.

A scam email can read as thoroughly professional now. A fake invoice can look completely legitimate. A message can nail a supplier's tone well enough to catch someone out on a busy day, and everyone's day tends to be a busy one.

Which means verification matters more than it ever has. If a payment request or an urgent instruction feels even slightly off, check it through a different channel. Call the person on a number you already have, not the one in the message.

Pause. Verify. Then act. Same rules as always, just with rather higher stakes attached. Use your Spidey senses. 🦸

Keep a human in the loop, always

AI can sound completely confident while being completely wrong. That's why someone still needs to check its work, much the way you'd check an apprentice's first few jobs before letting them loose unsupervised.

Before anything AI drafted goes out the door, ask whether it's accurate, whether it sounds like your business, and whether you'd be comfortable putting your name to it. This matters most for customer communication, financial decisions, contracts and anything touching personal information.

AI can support that work. It can't be accountable for it. Only you can.

Start with one safe experiment

You don't need a grand strategy to begin, just one small, low-risk task that suits where you're actually at, not where someone else's business is at. Try using AI to draft a routine reply, turn scribbled job notes into a proper list, or clean up a paragraph that's gotten away from you.

Then actually look at what it gave you. What worked? What still needed your judgement? Did it genuinely save time, or did fixing it up take just as long as doing it yourself?

That one small experiment will teach you more than any explainer article, this one included.

Check your foundations first

Before any shiny new tools, it's worth an honesty check on the basics you already have. Are your passwords actually unique, or is it still the dog's name and a "1"?

Is multi-factor authentication switched on where it's available? Would your team recognise a dodgy message if it landed in front of them? Are your files backed up somewhere that isn't just your own head?

AI doesn't replace any of that. Solid security habits are what make it safe to explore new tools with any real confidence at all.

Ask for help, but keep your hands on the wheel

You don't have to work this out solo. A trusted adviser can walk you through how a tool actually works and whether it makes sense for your business, without turning it into a lecture you didn't ask for.

The questions worth asking are simple. What problem does this actually solve? What information does it access? Who checks the output? Can we stop using it easily if it doesn't work out?

A good adviser helps you make the decision. They don't talk you into one, and a good one will start by asking where you're at, not where they think you should be.

So where does the confidence actually come from?

Fear grows in the vague spaces. Confidence shows up the moment the next step becomes clear, wherever you happen to be starting from.

You don't need to understand everything about AI. You need to understand the task you're trying to improve, the information you're protecting, and the checks you'll keep in place around it.

Start small. Keep the sensitive stuff out. Check every important output. And keep asking, honestly, whether the technology is helping, or just adding another layer of noise on top of the ones you already had.

AI shouldn't leave you feeling less in control of your own business. Used carefully, it should hand you back time, clarity and a bit more confidence in the decisions you're already making, whichever room you walked in with them.

If you feel like everyone else had this figured out and you haven't, come along to my talk on 27 August. We'll go through exactly how AI can be used at each level of a business, from the small time-saving stuff through to the bigger shifts, with real examples across trades, service businesses, retail and admin-heavy roles. Book here.