Practical AI
Choose one task before choosing another AI tool.
A practical way to select a first workflow: recurring work, clear inputs, a checkable output, and a manageable cost of being wrong.
Start with a task you did recently. Write down what arrived, the steps you took, and what you handed over. “Make marketing more efficient” is too broad to test. “Turn the weekly campaign updates into a draft report for review” gives you an input and an output.
Choose something that repeats. A one-off task can be worth doing with AI, but a recurring task gives you more opportunities to test the setup and decide whether maintaining it is worthwhile.
Make the output checkable. Before building anything, describe what a useful result contains. For a report, that could mean the source for each number, the reporting period, and a separate list of missing information. For a task, it could mean the action, the owner, the date, and the conversation it came from.
Start where a mistake is easy to catch and correct. A draft you review is a smaller commitment than a message sent automatically to a customer. Decide where your approval belongs before connecting a destination.
Try several examples, including one with missing information. Does the workflow ask for clarification, mark the gap, or quietly invent an answer? Keep the original inputs beside the output while you review it.
Count the review and upkeep as part of the work. If checking the result takes longer than the original task, change the scope. A useful first setup earns its place in your day through repeated use.
Write your starting brief in one sentence: When this input arrives, prepare this output, keep this context, and wait for this person to review it. That is enough to begin a focused setup conversation.
Thinking is better with a conversation.
Continue on LinkedIn ↗