AI knows everything about business, but nothing about your business

A writer called Balazs Balint ran a little test back in February. He asked ChatGPT a simple question: 'I need to get my car washed and the car wash is only 300 feet away. Should I walk or go by car?'

It told him it would be best to walk.

Which, of course, is ridiculous. How on earth is he going to get his car washed if he walks there?

So, four months on, I wanted to retest this. I tested five different models, because I wanted to know whether the issue was still there or whether the AI now had the context around it.

I put the same question into Google's Gemini. Walk.

ChatGPT. Walk.

Grok was worse. It gave me all the reasons why walking was better.

Copilot told me walking was probably quicker.

Even Claude, the model everybody's raving about, told me to walk, and helpfully added that my car would then be 'in the optimal position for washing'.

The car I had just left behind.

Every one of them sounded clever, and yet they all missed something a child knows: you cannot get your car washed if you leave the car behind.

So what is the problem? The challenge is that AI does not know the context. What it focuses on is how far you are from the car wash and whether you want to get there. It is not looking at whether you want to get there with the car, when the whole point of a car wash is that it is where you get your car washed.

And here is the thing. AI could solve the problem all along. When I pointed out to all the models that this was what I actually wanted to do, they all quickly changed their answer and said, 'Oh, you would be better driving.'

So it is not just that AI does not know. It does not know the context. And context is a critical part of the language we need to use with AI.

Because as humans, we use context all the time.

When you go away on holiday and arrive at your hotel, you do not drive your car into reception. You park in the car park outside. There is no sign on the door telling you that you cannot drive into the hotel with your car. It is just something we know. It is a rule, even though it is not written down anywhere, because the context tells us we do not behave that way.

AI does not know this. It does not know it is meant to behave that way. So sometimes it will suggest things that seem stupid, crazy, or even irrelevant.

When we prompt, and when we use AI, we need to give it that context all the time. With the context, AI can understand the question you are actually asking and give you a far better answer. Without it, it will happily tell you to walk the 300 feet to the car wash rather than drive, even though the point was never just to get to the car wash. It was to get there and get the car washed while you were at it.

So when you ask it to help you write your next business plan, or to build an agent to help you make decisions in your business, it is vital you give it the context. Because without it, AI is just going to guess. And guessing is not a great way to run a business.

And this is what we see when we train business owners and people on our courses, all the time. They have been prompting AI for months. They just have not been giving it context.

Then they learn how to do it, how to structure a prompt properly and how to hand over what they know, and it changes the way they use AI and the results they get. That is the whole job of our training. To shortcut your AI journey. To get you on top of the wave instead of under it. Less time, less effort, far better results.

And the key to all of it is context.

So if you have tried AI and found it underwhelming, do not write it off. It is not the tool. It is the context, the water you swim in, the things you know so well you forget to say them out loud.

Give it that, and the grey answer turns into your answer.

Because AI will walk to the car wash every single time. People who have learned to use it know to bring the car.

If that sounds useful, get in touch. We teach this hands-on, built on 21 years of running businesses ourselves, not on how the tech companies talk.