Don't Automate Confusion.

Automation has a reputation for seriousness. It looks like progress. It produces dashboards. It is easy to announce. Confusion does not photograph as well, so it is often left in place and given a login.

Automating a broken process does not fix it. Digitising unclear responsibilities does not create accountability. Building technology around poorly defined requirements does not guarantee the right solution. It encodes the mess — and then charges you to maintain it.

The pattern is familiar. A team is slow, so a workflow tool is bought. Handovers still fail, because nobody agreed who owns the decision at the join. Approvals still stall, because the approval was never a step; it was a conversation that happened in a corridor. The tool now has a record of the stall. That is not the same as removing it.

Confusion is not a software problem. It is a problem of people, process, and the stories an organisation tells itself about how work gets done. Those stories drift. New joiners invent their own versions. The official process and the real one stop speaking to each other. Automation then has to choose which version to believe. It usually chooses the official one, which is the one that does not run the business.

The cheaper move is the less glamorous one. Write down what actually happens. Find the gaps. Decide what should happen. Align the people who will have to live with it. Then — and only then — decide what is worth automating.

Understanding comes before implementation. That is not caution. It is how you avoid spending twice: once on the system, and again on the work of discovering what you should have known before you bought it.