Before You Buy More AI, Work Out What Problem You’re Trying to Solve

There’s a lot of pressure on businesses to “do something with AI”.

I attended Microsoft’s AI Tour in London in both 2025 and 2026, and one thing that struck me was how many conversations reflected a concern about falling behind with AI.

And it isn’t just larger organisations.

In conversations I’ve had with smaller businesses at Chamber events, I’ve heard much the same concern. AI is moving quickly, and there’s a feeling that if you’re not already doing something with it, perhaps your competitors are getting ahead.

That sense of urgency is understandable. But there’s a danger too.

When the motivation becomes “we need to do something with AI”, it’s very easy to start with the technology instead of the problem.

So rather than asking:

“Which AI tools should we be using?”

I think there’s a better question:

“What problem are we actually trying to solve?”

Start with the business, not the technology

Before considering licences, training or implementation, take a step back and look at how people actually work.

Where are people spending time on repetitive tasks?

What information takes too long to find?

Which processes involve unnecessary manual effort?

Where are people repeatedly recreating similar reports, documents or emails?

Which useful tasks are being neglected because nobody has enough time?

Those questions start to reveal where AI might make a meaningful difference.

But sometimes the answer won’t be AI at all.

It’s very easy with any new technology to start experimenting simply because the capability is there. You can spend hours working out how to make a piece of technology do something, only to step back afterwards and realise:

We didn’t actually need to do this in the first place.

You may have learned something from the experiment, so the time isn’t necessarily wasted. But there’s an important distinction between:

“Can we do this?”

and:

“Is this worth doing?”

AI makes that distinction even more important because there are now so many things we can do.

A process might need simplifying rather than automating. Information might need organising properly. An existing capability in Microsoft 365 might already solve the problem. Or perhaps the task doesn’t need doing at all.

Automating an unnecessary task simply allows you to do something unnecessary more efficiently.

You may not be as far behind as you think

One of the interesting things about my conversations with businesses of very different sizes is that the uncertainty is often remarkably similar.

Where should we start?

What could AI realistically do for us?

Which people should use it?

Is our information ready?

How do we know whether we’re actually getting any value?

Those are sensible questions.

Asking them doesn’t mean you’re behind.

Taking time to identify worthwhile opportunities can be a better starting point than rushing to deploy technology because you feel you ought to.

And those opportunities don’t have to be transformational.

Perhaps somebody spends several hours each week summarising information from different documents. Maybe a manager regularly creates similar reports from meeting notes and emails. Perhaps staff spend significant amounts of time drafting routine communications or trying to find information scattered across Microsoft 365.

A small improvement that saves an hour every week may not make a dramatic AI case study.

But multiply that across several people and throughout the year, and it starts to become much more interesting.

Finding the opportunity is only the start

Suppose you identify something that Microsoft 365 Copilot could help with.

The next question shouldn’t necessarily be:

“How many licences should we buy?”

It should be:

“Are we ready to use it effectively?”

Do people understand what Copilot can and cannot do? Are they confident about checking AI-generated responses? Is the information they need accessible and well organised? Are permissions appropriate? Do people understand why the organisation wants them to use AI in the first place?

Training matters. I’ve spent much of my career helping people learn technology, and building skills and confidence is essential.

But successful adoption requires more than teaching someone which button to press or which prompt to type.

We need to move the question from:

“How do I use Copilot?”

to:

“Could Copilot help me do this particular piece of work better?”

That’s when AI starts becoming a practical business tool rather than simply an interesting piece of technology.

AI should earn its place

I’m very enthusiastic about Microsoft 365 Copilot and AI.

But enthusiasm shouldn’t stop us asking whether the technology is solving a worthwhile problem.

The objective isn’t to get everybody “using AI”.

The objective is to help people work better.

Sometimes AI will play a significant part in achieving that. Sometimes a small use of Copilot might save someone an hour every week.

And sometimes the sensible conclusion will be:

“We don’t actually need AI for this.”

That isn’t a failure.

So if you’re concerned that your organisation is falling behind with AI, perhaps don’t begin with:

“What AI should we buy?”

Start with something simpler:

“Where could we make work better?”

Once you understand that, you have a much better basis for deciding where AI deserves a place.