← Back to the notebook

Practical AI

Does your automation create another job?

Count the checking, corrections, permissions, and maintenance before deciding a workflow is saving you effort.

A workflow can finish successfully and still make your day harder. The task appears in the right app, but the date is wrong. The summary looks complete, but you have to reread the source to trust it. The integration works until a permission expires.

Include that work in your assessment. Keep a simple record of what you spent on the original task and what you spend using the setup: capture, review, corrections, and maintenance. Use your own examples rather than an estimated saving from someone else’s demonstration.

Look at the exceptions. Which inputs produce uncertain results? Which corrections happen repeatedly? An exception that needs the same fix every time is a reason to change the instructions, narrow the task, or remove that step from automation.

Make ownership explicit. Someone needs to notice a failed connection, check whether the source changed, and decide how to handle incomplete output. For a personal setup, that person may be you. For a team, assign the responsibility before relying on the workflow.

Keep a way to recover the original information. A task with a link to the conversation is easier to check than a task with no context. An extracted value beside its source is easier to correct than one presented as an unexplained fact.

Review the setup after you have used it across several real examples. Keep the parts that remove effort. Simplify or stop the parts that create more checking than they replace. The goal is a day that is easier to manage, including the work needed to keep the system running.

Thinking is better with a conversation.

Continue on LinkedIn ↗