Workflow Automation Readiness Checklist

01 / Start here

Choose one task

Keep the exercise small: choose one recurring process, not an entire department or business.

I have chosen one recurring task to map.

What is the task?

Use a short, plain-language name.

Who handles it today?

What starts it?

Note the event, request, schedule, or signal.

What should be true when it is done?

02 / Current state

Map today’s process

Describe what actually happens now, including the handoffs and moments that do not fit a tidy diagram.

I have captured the current process as it happens.

What are the current steps?

What information or materials go in?

Which tools are used?

Where does work move between people or tools?

What comes out of the process?

Where does it stop, wait, or need a person to step in?

03 / Friction & repetition

Measure the repetition and friction

Rough observations are useful. You do not need perfect reporting data to notice where the work repeats or drags.

I have noted where repetition or friction shows up.

How often does it happen, and at what volume?

How long does one occurrence usually take?

Where do delays or queues appear?

Where do copying, rework, or duplicate entry happen?

What errors or missed steps occur?

Which part feels most costly, frustrating, or distracting?

04 / Judgment & safeguards

Separate rules from judgment

A repeatable sequence can still contain decisions, exceptions, approvals, and information that needs careful handling.

I have marked the rules, decisions, and safeguards in the process.

Which steps follow predictable rules?

Which steps require variable decisions or context?

What exceptions or edge cases change the path?

Where are approvals or sign-offs required?

Does the process touch sensitive information? What should be protected?

Where must a person review, correct, or make the final call?

05 / Fit, not hype

Consider the fit

Use these signals to frame a conversation, not to produce an automated score or a feasibility verdict.

There is a clear trigger.The core rules are repeatable.The tools and access are stable enough to inspect.The desired output is measurable or easy to review.I have recorded the caution signals before deciding.

What caution signals are present?

Consider high stakes, an unclear process, many exceptions, sensitive data, or context-heavy decisions.

Where might AI support help with drafting, sorting, summarizing, or classification?

Keep a person in review and note what they would check.

What would the human review checkpoint need to look like?

06 / Next small step

Choose a small next experiment

A useful next step can be a clearer process map or a bounded test. It does not need to automate the whole task.

I have defined a small, reviewable next experiment.

What is the smallest useful starting point?

Who owns the next step?

What signal would tell you it is useful?

Where will a person review the result?

What is the fallback or manual path if it does not work?

After reviewing the notes, which is the best description for now?

Promising to explore · Needs more clarity · Keep manual for now