A slow process gets automated. A confusing workflow gets redesigned. A meeting gets shortened. All of that can look like progress.
But what if the process, workflow, or meeting did not need to exist?
I think we often start one question too late. We accept the thing as necessary, then put all our effort into making it better.
Ask why it is there
A requirement may come from an old system, a regulation, or a decision made under very different constraints. Over time, the reason gets harder to find while the process remains.
Then a new team arrives and optimizes it.
Suppose opening an account takes seven steps. Making those steps easier sounds like a sensible UX problem. But first I would want to know why there are seven.
Maybe three are legally required. Two serve an internal reporting system. One was added after an incident. Another has simply never been questioned.
Now there are several problems to investigate, not one seven-step flow to polish. Some steps may still be necessary. Others might disappear if we fix the reason they were added.
Removing work is harder to show
It's easy to show that a process is faster. It's harder to get credit for deciding we didn't need it, even if that saves more work.
But even an automated workflow needs maintenance, attention, and governance. Making unnecessary work cheap does not make it free.
This is the sequence I find useful:
Question Remove Simplify Accelerate Automate
Sometimes the existing process is necessary. I just want to understand why before spending time improving it.
AI makes it easier to skip that question
We can now generate reports, summarize meetings, and move information between systems with much less effort. The temptation is to automate because it is possible.
But a summary does not make an unread report useful. An agent moving data between two systems does not tell us whether both systems are needed.
When a request arrives, I think it helps to ask what happens if we do nothing. A feature can be reasonable without being important enough to build.
Then reframe the request around its intended outcome:
| Instead of asking | Ask |
|---|---|
| How do we improve this approval process? | What risk is this approval process supposed to control? |
| How do we automate this report? | Who makes a decision because this report exists? |
| How do we make this feature easier to use? | What is the customer actually trying to accomplish? |
| How do we speed this workflow up? | Does this workflow need to exist at all? |
| How do we add AI here? | What problem would AI actually remove? |
| How do we build this requirement? | Is this requirement real, or inherited? |
Those questions do not guarantee a different answer. Sometimes the existing process is there for a good reason.
But they give us a better basis for deciding what to keep, what to change, and what to stop doing.
Before making the process faster, I'd ask what would happen if we stopped doing it.