Automation that outlives the person who built it

Every team has at least one. A script, a workflow, a spreadsheet with macros, that does something important and that exactly one person understands. It works. It has worked for years. And everyone is quietly terrified of the day that person leaves.

That fragility is not a quirk. It is a design failure, and it is the most common one we find.

Working is not the same as durable

An automation that works is solving a problem right now. An automation that is durable keeps solving it when the person who built it is gone, when the tool it depends on changes, and when nobody remembers why it was built the way it was.

Most internal automations are built for the first case and never tested against the second. They work, so nobody questions them. The brittleness stays invisible until the day it does not.

"If only one person understands it, you do not have an automation. You have a dependency."

What durability actually requires

Durability is not about clever code. It is mostly about boring discipline.

• It runs somewhere shared, not on one laptop. If the automation lives on a personal machine, it dies when that machine does.
• The credentials belong to the company, not the individual. A workflow tied to a personal login breaks the moment that person changes their password or leaves.
• It fails loudly. When something breaks, a real person finds out quickly, rather than discovering a month of silent failures later.
• Someone other than the builder can read what it does. A page of plain-language documentation is worth more than the most elegant code.

The test

There is a simple way to check whether your automation is durable. Imagine the person who built it is unreachable, starting today. Could someone else find it, understand what it does, tell whether it is working, and fix it if it broke?

If the answer is no, you do not have a finished automation. You have a prototype that happens to be in production, and a risk that nobody has priced.

We treat this as part of the build, not a favour at the end. The handover is the point. If we vanished tomorrow, what we built should keep running without us. That is the only version of done worth aiming for.