Twenty Years of 'That's How We Do It' Is Now an AI Problem

Every organisation has processes that nobody would design from scratch today.

They may have started with perfectly good reasons. A particular system worked a certain way. A regulatory requirement needed to be accommodated. Two systems couldn't communicate, so somebody created a manual step between them. A spreadsheet filled a gap. A person developed a workaround that made their job easier.

Then something interesting happens.

The original reason disappears, but the process remains.

Eventually nobody asks why anymore.

"That's how we do it."

Processes have a history

I've been looking closely at processes that have evolved over many years, and one thing keeps striking me: how difficult it can be to separate what an organisation genuinely needs to do from what it has simply learned to do.

A process can contain years of accumulated decisions.

Some steps exist because they're necessary. Others exist because of the limitations of technology that has since been replaced. Some protect against genuine risks. Others may simply replicate a way of working that made sense ten years ago.

From the outside, they all look like "the process".

Until somebody starts asking why.

AI changes the question

Traditionally, when organisations introduce new technology, we often begin by mapping the existing process and working out how to reproduce it in the new system.

That makes sense when continuity is the objective.

But AI gives us an opportunity to ask something different.

If we were designing this process today, knowing what technology can now do, would we design it this way?

Perhaps the answer is yes.

But perhaps three steps could become one. Perhaps information doesn't need to be manually transferred. Perhaps somebody doesn't need to read twenty documents to find three pieces of information. Perhaps a process doesn't need to wait for one person who happens to know what happens next.

And perhaps some work doesn't need to happen at all.

Don't automate "that's how we do it"

This is where AI adoption creates both an opportunity and a risk.

AI can make existing processes dramatically faster.

But speed isn't necessarily improvement.

If we automate a process without understanding why it exists, we may simply embed yesterday's limitations more deeply into tomorrow's organisation.

That's why I think one of the most useful questions we can ask before applying AI to a process is also one of the simplest:

Why do we do it this way?

Not to dismantle everything that came before.

Not because old processes are necessarily bad.

But because AI gives us capabilities that weren't available when many of those processes were designed.

Twenty years of organisational history can teach us an enormous amount.

It just shouldn't automatically become the blueprint for the next twenty.