Before asking how AI could improve something, ask whether it still makes sense to do it that way at all.
Processes, rules and ways of working are usually created for a reason. The problem is that the reason can disappear while the solution stays.
AI gives leadership teams a reason to reopen that logic before they automate it.
1. WHAT PROBLEM WAS THIS DESIGNED TO SOLVE?
Find the origin.
What was happening when this process, approval, role, meeting or rule was created? Perhaps information was difficult to access. Two systems could not communicate. Getting something wrong carried a high cost. A customer expected something that technology could not provide another way.
Until you understand the original problem, you cannot judge whether the current work still makes sense.
2. DOES THAT PROBLEM STILL EXIST?
Test the relevance.
Sometimes the answer is yes. Keep the control. Improve the process if necessary. AI does not mean everything old is wrong.
But if the original constraint disappeared years ago, making its solution faster may be the wrong objective.
If the problem no longer exists, why are we making its solution faster?
3. WHAT IS POSSIBLE NOW?
Expand the possibility.
What could be removed rather than improved? What can AI, data or technology make viable today that was not viable when the current model was designed?
This question should not stop at productivity. It can lead to customer value, new services, scalable customisation, new customer segments and different revenue models.
4. IF WE DESIGNED IT TODAY, WOULD WE BUILD IT THE SAME WAY?
Force redesign.
Forget the inherited process for a moment. Start with the customer, the outcome you want to create and the technology available today.
How should information move? Where should decisions sit? What should people do? What should AI do? What would you measure? Would you even sell the same thing in the same way?
Then compare that with the model you have today.
THE DECISION
If the logic still makes sense, keep it and improve it.
If it no longer serves a purpose, stop it.
If it needs to change, ask one more question: Is the future model clear?
No → Rethink. Find and prove what the better model should be.
Yes → Rebuild. Design how the new model should actually work and launch it.
TWO DIRECTIONS
Rethink the Work — question how the business operates.
Rethink What You Can Sell — question how the business creates value and revenue.
The same discipline sits underneath both: challenge the assumptions before deciding what AI should change.
Challenge the logic before you change the work.