01 / DEFINE
◎
Find the friction
Which task is repetitive, slow, expensive or error-prone? Who experiences the problem?
EXPERT INSIGHTS / AI IN BUSINESS
AI is not the strategy.
A new tool can look impressive in a demonstration. But if nobody can explain which business problem it solves, it is technology looking for a purpose.
Organisations are under pressure to adopt AI. The temptation is to start with a platform, announce a pilot and look for somewhere to use it. A more useful question comes first: what are we trying to make better?
THE WRONG START
“We need an AI strategy.”
Buy a tool. Run a demo. Ask teams to find use cases. Then wonder why nothing has changed.
THE BETTER START
“What problem costs us time?”
Find the friction. Establish the baseline. Test whether AI can improve the outcome.
Start where the work hurts
Think about the everyday problems that drain capacity: searching across disconnected knowledge bases, rewriting the same documents, manually categorising requests, or waiting days for information that should take minutes to find. Those are potential opportunities—but not automatic reasons to deploy AI.
For example, a service team might spend hours sorting incoming requests. AI could help classify them and suggest a response. Yet the real goal is not “introduce AI classification.” It is reduce the time from request to correct resolution, without increasing mistakes or exposing sensitive information.
02 / MEASURE
↗
Set the outcome
Choose a meaningful measure: turnaround time, quality, customer satisfaction or hours saved.
03 / CHECK
▦
Test the foundations
Do you have usable data, clear permissions and reliable processes?
04 / PROTECT
◇
Keep people in control
Decide what AI may recommend—and what a person must review or approve.
05 / LEARN
✓
Pilot before scaling
Compare results with the original baseline. Expand only when the benefit is demonstrated.
Sometimes the answer is not AI
If the underlying problem is unclear ownership, inconsistent processes or missing data, an AI assistant may only make the confusion faster. A better form, a simple automation or clearer accountability might solve it more reliably and at lower cost.
The strongest AI opportunities combine a defined business need with responsible design: appropriate data access, security, testing, human oversight and a clear owner for the result.
Don’t measure how much AI you deploy. Measure what gets better.
Start small, solve something real, and let the evidence decide what comes next.
Where could AI genuinely help your organisation?
Share your perspective →Artwork credit
Original AI-themed illustrative artwork.