When you think about your role, do you see yourself as a mason or an architect?
The mason delivers the work. The architect also helps decide what needs to be built, how it fits together and what it should achieve.
Both matter. But how much of your working week is spent completing tasks, fixing the same problems and following processes you had little say in designing?
And what would change if you had more influence over those processes?
For me, this is where AI gets interesting. Not just another way to produce a report faster, but a way for people who understand the work to start shaping it. You don’t need to become a programmer to explore an idea, compare alternatives or put together something your colleagues can test.
AI doesn’t make you an architect automatically. It can make taking that first step more accessible.
Start with the decision, not the report
Imagine you work in finance and prepare the monthly margin report. You chase missing costs, reconcile the figures, explain movements and send it out. Next month, you do it again.
That is valuable work. But does the report help anyone decide which jobs to reprice, which costs to challenge or where operations needs to act?
The mason asks: “How can I get this report finished?”
The architect also asks: “What decision should this report help us make?”
AI can help you explore that second question. Using an approved tool and dummy figures, ask it for three ways to present the margin problem. Ask what each approach might hide. Ask what an operations manager would need before acting on it.
Then challenge the answers. Has it treated costs consistently? Has it confused timing with a genuine deterioration? What evidence supports its explanation?
Use it to draft a sample review page showing the movement, evidence, proposed action and person responsible. Take that to operations and ask whether it would help.
You are still applying your finance skills. But you are also helping design how the business makes a decision, rather than stopping at supplying the numbers.
This isn’t just a finance opportunity
In operations, are you the person who arranges urgent purchases, or someone who can help reduce the need for them? AI could help you compare approval options and develop fictional shortage scenarios to work through with the team.
In sales, are you getting quotes out, or helping make customer promises more dependable? Use AI to draft a better handoff between sales and operations, then let the people doing the work pull it apart.
In HR, are you completing an onboarding checklist, or making sure someone is ready to contribute? AI could help you explore the gaps in a fictional new starter’s journey and propose a better approach.
The point isn’t to accept whatever AI produces. It is to have something concrete to question, improve and discuss. Use approved tools and authorised data, and test before changing live work.
What does this do for you?
The company benefits when an improvement produces clearer decisions, less rework or better coordination. You benefit by developing—and demonstrating—judgement, influence and a broader understanding of the business.
“I use AI” doesn’t tell a manager much.
“I identified a recurring problem, worked with the team and tested a better approach” gives you a much stronger conversation.
Keep the evidence. What changed? Did it actually help? Include checking time and any work pushed onto someone else. An attractive prototype is not the same as an improvement.
Leaders have a responsibility here too. You cannot ask people to act like architects while filling every hour with delivery. Give them time, training and authority. Agree what existing work will reduce.
On Monday, choose one recurring frustration. Use approved AI to explore two alternatives. Discuss them with the colleague affected and seek approval for a small test.
You don’t have to stop being a good mason. But with AI helping you explore and test your ideas, why identify yourself only as the person who does the work—and not someone who can help design it?

Comments
Moderated — email required. Don’t paste confidential or client data. Community guidelines.