AI makes it very easy to begin in the wrong place. A new model appears, somebody demonstrates an impressive agent, or a competitor announces an AI initiative. Suddenly the conversation becomes about tools before anyone has properly examined the work.

That is backwards.

When I look at a workflow, I deliberately put the tool discussion aside at this point. Otherwise the technology starts dictating the problem we think we have.

Curiosity comes before automation.

Before deciding what to automate, I want to know how the business actually works. Where do people lose time? Where does information get copied from one system to another? Which decisions repeatedly require the same research? Where are customers waiting? Which tasks frustrate capable people because the process around them is unnecessarily complicated?

Those questions are less glamorous than choosing an AI platform. They are also much more valuable.

Do not run before you can walk.

A lot of companies are trying to run before they can walk when it comes to AI. They are adding copilots, agents, chatbots and automation tools across the business because they feel they should be doing something with AI. But if the underlying workflow is unclear, inefficient or full of unnecessary steps, adding AI can simply make a poor process faster and more complicated.

Before introducing new tools, look closely at how the work moves today. Where are the hand-offs? Which steps are duplicated? Where does information get re-entered? Which approvals add value and which exist only because “that is how we have always done it”? A messy workflow does not become a good workflow because AI has been layered on top of it.

First understand the existing workflow. Then improve that workflow — through better process design, automation, AI, or the right combination of them. And keep a human in the loop wherever judgement, accountability, context or oversight matters. The goal is not to remove people from the process; it is to let technology handle the work it does well so people can focus on the work that needs human judgement. Technology should strengthen the way the business works, not be layered on top of a workflow that has never been examined.

No AI for the sake of AI. Just the right technology, in the right places, for your business.

The answer may not be AI.

Sometimes a problem needs an intelligent assistant. Sometimes it needs automation between existing systems. Sometimes a workflow should simply be redesigned. And sometimes the most useful recommendation is to leave something alone because the cost and complexity of changing it would outweigh the benefit.

That is why a business-first assessment matters. Technology becomes a response to evidence rather than a solution looking for a problem.

Look inside the business before looking outside it.

Most organisations already contain more opportunity than they realise: knowledge hidden in documents, repetitive processes people have stopped questioning, useful data trapped between systems, and small inefficiencies multiplied hundreds of times each month.

Curiosity opens those drawers. Good analysis decides which ones are worth exploring. Only then should AI enter the conversation.

That is the idea behind the Curiosity Cabinet.

This is where I will examine practical AI, automation, workflows and emerging technology without treating every new tool as a revolution. Some drawers will contain useful ideas. Some will contain warnings. Some may contain something unexpectedly valuable.

The point is not to use more AI. The point is to discover where AI can make work genuinely better.