Module 01 Activity

Scenario

Someone has proposed replacing a manual process at your organisation with an agent. Before agreeing, establish whether it needs one.

What you produce

A one-page structure and autonomy decision for a real workflow, argued from the run-time-sequence test rather than from preference.

Steps

1. Pick a workflow you actually run. Write its steps as a numbered list.
2. Against each step, write yes or no: does what happens next depend on a value that only exists once the run has started? Count the yeses.
3. Choose the smallest structure that fits, and name the specific requirement that forces anything above a fixed sequence.
4. Choose the autonomy level from the four in Unit 02, and write one sentence saying how long a wrong action could go unnoticed at that level.
5. Find a vendor demo for something similar and write down the answer it gives to each of the six questions in Unit 03. Mark the ones it does not answer.
6. Write the decision as three sentences a colleague could disagree with.

Evidence to hand in

• The step list with the run-time-dependency column.
• The chosen structure and the requirement forcing it.
• The autonomy level and the exposure window it implies.
• The six demo questions with the demo's answers.
• The three-sentence decision.

Review checklist

• Every step is marked for whether the next step depends on it.
• Model calls are not counted as evidence of needing an agent.
• The autonomy level is chosen from reversibility, not from ambition.
• The demo questions are answered from what was shown, with unanswered ones marked as such.
• If the answer is 'a fixed sequence', that is written down plainly.

Mark complete