Search courses, chapters, or pages...
Identify the event that starts a piece of work, such as a form submission, email, calendar time, or status change. Practice separating a real trigger from a vague request like “keep an eye on this.”
Use what you learned in the previous lesson to solve real-world problems.
State what the task needs before it can begin and what it must produce when it is finished. Use a clear “done” condition so the automation has a target instead of an open-ended activity.
Check what you understood with a short quiz.
Rewrite vague work like “process the order” into observable actions: check, copy, compare, send, update, or file. Build a short sequence that another person—or a machine—could follow.
Draw a boundary around one task by marking where it starts, where it ends, and what stays outside. Learn why automations fail when they quietly include extra favors, follow-ups, or side conversations.
Look for moments where the work can go down more than one path, such as approve/reject, urgent/not urgent, or match/no match. Capture the question being answered without trying to model every exception yet.
Compare decisions based on clear criteria with decisions that require taste, empathy, ethics, negotiation, or accountability. Decide when automation can execute the rule and when it should only support a human.
Trace who receives the result of a task, what they need from it, and how they know it is ready. Recognize handoffs as part of the work, not an afterthought after the automation runs.
Check whether the same kind of work happens often enough, and consistently enough, to be worth automating. Use volume, time spent, delay, and error rate as practical clues.
Test whether the task usually follows the same path with the same kinds of inputs and outputs. Learn to flag work that must be standardized before automation will help.
Reason through whether the work already happens in apps, forms, messages, or systems that automation can access. Notice when paper, screenshots, hallway conversations, or missing records make the task harder to automate.
Classify tasks by what happens if the automation is wrong: easy to undo, costly to fix, legally sensitive, or harmful to people. Use risk and reversibility to decide when approval or human ownership must remain.
Choose whether a task should be fully automated, assisted by automation, or kept human-led. Practice turning “automate this job” into a safer target like draft, classify, route, remind, summarize, or update.
Take a messy process and carve out one stable, valuable path to automate first. Focus on a narrow slice with clear inputs, steps, output, and handoff instead of trying to automate the whole job at once.
Review this chapter with practice based on your mistakes.
