I talk to businesses about AI every week. And every week, I hear the same thing.
'We want to start using AI.'
My response is always the same. 'For what?'
That's where most conversations hit a wall. Many businesses want to use AI - they just don't know what for. They've seen articles. Tinkered with the free version of Chat GPT. A mate at a networking event said his business was all over it and it's awesome. So now they feel like they should be doing something.
But 'something' isn't a strategy.
The Question That Actually Matters...
Before we talk about AI, I ask a different question.
What bugs you?
Not 'what are your strategic priorities?' Not 'where do you see the business in five years?' Just - what irritates you and your staff in the day-to-day grind? What are the tasks that eat time for no good reason? What do people complain about on a Friday afternoon over drinks?
That list - the frustrating, the unglamorous, the why-are-we-still-doing-this-manually list - is your starting point.
It might sound like:
'We manually pull the same data into a report every single week.'
'Every new client gets the same email. Someone writes it from scratch every time.'
'Staff waste 20 minutes searching old email threads for information that should take 30 seconds to find.'
'We answer the same questions from customers over and over.'
None of these feel like exciting AI use cases. But they are.
They're exactly the kind of repetitive, pattern-based work that AI handles well - and that quietly costs businesses hours every week.
Lean Thinking Makes This Practical...
There's a methodology called "lean" - originally from manufacturing - built around one idea: eliminate waste.
Any step in a process that doesn't add value is waste, and waste should go.
It's a useful lens for finding where AI fits.
Map a process that annoys you. Write down every step. Then ask honestly: which steps involve waiting, re-entering data somewhere else, reformatting the same thing, or repeating a communication that never changes? Those are your targets.
Then ask one more question: Could a well-trained assistant do this?
That's essentially what AI is. If the task is repetitive, clearly defined and based on a pattern - yes.
If it requires real judgement, relationships or context that lives only in someone's head - then probably not yet.
An Example...
A client runs a small logistics business. Every week, their admin team transfers notes from driver debrief forms into a spreadsheet, then manually build a client report from that spreadsheet. Hours of work. Prone to errors. Everyone hates it.
Map the steps. Find the duplication. Build a simple digital form that auto-drafts the client report. The team reviews it and it gets sent. The manual transfer of data disappears.
No expensive software. No transformation project. Just a real frustration, fixed.
Before You Talk to Anyone About AI...
Get clear on these four things first:
What specific task or process frustrates your team most?
How long does it take per day, per week?
What does the output look like, and who uses it?
What would 'fixed' look like?
If you can answer those, you're ready for a useful AI conversation.
If you can't, do this groundwork first. It'll save you a lot of time and money.
The Point...
AI isn't magic. It's a tool. And like any tool, it only works when you know what job you're giving it.
Start with what bugs you. That's where the value is.