How to Talk to AI: A Small Business Owner’s Guide

There’s a new book doing the rounds called How to Talk to AI by the technology writer Jamie Bartlett, and it makes a point that we see play out with clients every week: most people still don’t really know how to get useful results out of AI. They type a vague question, get a vague answer, and conclude “AI isn’t that clever really.” Meanwhile, someone else at the next desk is using the exact same tool to save five hours a week.

The difference usually isn’t the AI. It’s how it’s being talked to.

If you run a small business, this isn’t just a curiosity. It’s becoming a genuine business skill, in the same category as knowing how to write a decent email or read a spreadsheet. So let’s break down what “talking to AI properly” actually means in practice, and where the line sits between AI that helps your business and AI that quietly wastes your time (or worse, makes decisions you didn’t mean to hand over).

Why This Is a Business Skill Now, Not Just a Tech Curiosity

A few years ago, “using AI” meant one specialist person in the business experimenting with a chatbot. Now it’s baked into email, spreadsheets, customer service, marketing, even accounting software. Every team member is, whether they realise it or not, having a working relationship with AI several times a day.

The businesses getting real value aren’t the ones with the biggest budgets. They’re the ones where people understand a few basic principles about how these tools actually think, and where they fall short. That’s exactly the gap our Digital Marketing and AI course was built to close: not abstract AI theory, but practical, plain-English skills your team can use on Monday morning.

Five Mistakes Small Business Owners Make When Talking to AI

1. Treating it like a search engine

Typing “marketing ideas” into an AI tool and expecting a tailored plan is like walking up to a new hire on day one and saying “sort out the marketing.” You wouldn’t do it to a person. Don’t do it to the AI. Context is everything: your industry, your customers, your budget, what’s already been tried.

2. Accepting the first answer as gospel

AI tools are confident by default. They’ll state something incorrect with exactly the same tone as something correct. For anything that matters (pricing, legal wording, factual claims about your industry), treat the first answer as a draft to check, not a finished decision.

3. Never telling it who it’s talking to

“Write a social post about our new service” produces something generic. “Write a LinkedIn post for UK small business owners who are sceptical of AI hype, in a practical, no-jargon tone” produces something usable. The audience and tone instructions do most of the heavy lifting.

4. Handing over judgement calls, not just tasks

There’s a real difference between “draft me three reply options for this customer complaint” and “reply to this customer complaint and send it.” One keeps a human in the loop; the other quietly hands over decisions that might need a human touch, a distinction Bartlett’s book spends real time on, and one we think about a lot when we build chatbots for clients.

5. Giving up after one bad experience

A clunky first attempt with a free chatbot puts a lot of business owners off entirely. But a badly prompted AI tool and a well-built AI system for your business are two very different things. Which brings us to the point where general AI chat and dedicated business tools part ways.

A Simple Framework for Better Results

You don’t need to become a “prompt engineer.” You just need to consistently give AI four things:

That last step  the “check”  is the one businesses skip most often, and it’s the one that protects you. AI is a brilliant first draft machine and a risky final decision-maker. Keep it in the first role.

Where This Matters Even More: Customer-Facing AI

Everything above applies double when the “conversation” with AI isn’t just you drafting an email. It’s your customers talking to an AI chatbot on your website, at 11pm, when you’re not there to check its answers before they’re sent.

This is exactly why we build chatbots with clear boundaries: a well-built AI chatbot knows what it can answer confidently, what it should hand off to a human, and what it should never guess at (pricing exceptions, medical or legal specifics, anything genuinely business-critical). It’s the same “role, context, task, check” logic. Just designed in advance, so your customers get a good experience and you’re not left cleaning up an overconfident answer.

If you’ve been putting off adding a chatbot because you’ve had a mediocre experience with a generic one, that’s usually a sign it needed better design, not that the idea doesn’t work. Have a look at our AI chatbot development service to see how we approach it – never missing an enquiry, without ever making a promise your business can’t keep.

The Bottom Line

AI isn’t going to become less present in business over the next few years. It’s going the other way. The businesses that benefit won’t be the ones that avoid it, or the ones that hand everything over to it blindly. They’ll be the ones who learn to talk to it well: give it context, keep a human check on anything that matters, and know exactly where the line is between “AI drafts this” and “AI decides this.”

That’s precisely what we cover, hands-on, in our Digital Marketing and AI course – real tools, real prompts, real business scenarios, explained in plain English.