The best way to use AI: Think first, prompt second

Let’s talk about how we’re actually using AI   

ChatGPT. Claude. Em-dashes.  

AI is everywhere right now. It’s all anyone seems to be talking about. Every tool promising to solve “this” or automate “that”, it’s no surprise how many businesses are bringing it into their workflow.  

The thing is, AI can be useful. As an operational tool it can be incredibly effective and help us to produce better work for our clients.  

Where things go wrong is when it’s used to replace thinking. That’s the conversation we should be having about it. 

Think first. Prompt second. 

The quality of content is dropping – fast.  

You can see it most clearly in marketing and communications. Everything is starting to sound the same. Brands that once had a clear tone of voice are now blending into one generic, AI-generated style. It's easy to spot and you can never unsee it.  

More people are skipping the thinking stage and jumping straight to generate. Then not taking the time to check or validate what’s being produced. That leads to more time spent rewriting, fixing inaccuracies and correcting things that should’ve been right in the first place. 

AI has an answer for everything – even when it shouldn’t.   

Ask for stats? It’ll produce them. Insights? You’ll get them delivered in a way that sounds just right.  

It isn’t being held accountable. There’s no challenge, no verification, no pushback.  

Most current AI is passive. It follows instructions without questioning them – even when they’re wrong. That responsibility sits with us, especially in creative industries. 

So, how do we use AI properly?  

If you’re going to use AI (and you most likely already do and will), here’s are a few awesome. tips:  

1. Use it as an operational tool, not a thinking tool 

AI should help you execute, not replace your thinking.  

Use it to understand concepts, structure ideas or get a starting point – but don’t take its output as fact without reflecting on it. 

2. Don’t skip the prep work 

The quality of what you get out of AI depends on what you put in.  

If your prompt is one sentence but you expect something detailed and accurate, you haven’t briefed it properly.  More information = better outcome. 

The more context you provide, the less guesswork AI has to do. 

3. Be super specific 

Instead of asking AI to rewrite everything, be specific. 

Ask where you can reduce word count or improve clarity – not replace your work entirely. This keeps you in control of the output, rather than handing it over completely. 

The more focused your prompt, the more useful the response. If you ask something broad, you’ll get something broad back. If you ask something precise, you’ll get something you can actually use. 

4. Get organised  

AI is great for planning and structuring, but only if you give it the right inputs.  

If you want a useful plan, you need to treat it like a proper brief. That means giving context, not just a task.  

For example: 

• Include deadlines, time limits or buffers  
• Highlight constraints like holidays, availability or dependencies  
• Add any fixed points that can’t move  

The more detail you give, the more realistic and usable the output becomes. 

Last thought from me 

Ultimately, I think it all comes down to us being more mindful about how and when we use AI.  

Yes, it can speed up processes, but we should use it to get things done the right way. Not just quickest.  

We should be holding it accountable, always. Check its outputs. Ask for sources. Challenge what it gives us. 

And never let it replace some good ol’ fashioned thinking. It produces better work, that you can actually stand behind.