Oprill field notes · work review
Count the work around the answer.
An original workbook for examining preparation, review and correction in an AI-assisted task. Use the synthetic exercise to see which effort a fast first draft can leave out.
A planning exercise, not a survey or a research finding. The example values are synthetic.
1. Define a comparable task.
Choose a result that a reviewer can accept or reject. In this synthetic example, the task is to turn a short meeting note into an action list. Both the manual and assisted versions must identify the same commitments, owners and unresolved dates. A shorter output is not automatically a better one.
Write the acceptance conditions before timing either approach. Use the same source material and require the same level of review. If one version skips source checks while the other includes them, the elapsed times do not describe comparable work.
2. Make the hidden steps visible.
Separate preparation, drafting, checking and correction. Preparation includes choosing approved sources and describing the task. Checking includes opening the source again, comparing each commitment and identifying uncertainty. Correction includes rewriting or removing unsupported statements.
The worksheet starts with a fictional manual total of thirty minutes. Its assisted example uses six minutes of preparation, three minutes of drafting, twelve minutes of review and four minutes of correction. Those numbers describe an illustration only. They are not measurements of a tool, a customer or a typical workplace.
3. Read the comparison carefully.
The calculator adds the four assisted stages and subtracts that total from the manual baseline. A positive difference means fewer minutes in this particular example. A negative difference means the assisted path takes longer. Neither result establishes a broad productivity claim.
Try increasing review time while keeping drafting time constant. This exposes an important question: what makes the reviewer’s job difficult? It could be missing citations, unclear source versions or a draft that blends proposals with commitments. Improve the workflow only after identifying the specific cause.
4. Record quality alongside time.
Keep a separate record of unsupported claims, omitted commitments and unresolved contradictions. Time and quality answer different questions. A fast draft that causes an incorrect external action should not be treated as successful because its arithmetic looks favourable.
Before using workplace content, agree on source access and the action boundary. Record who can stop the exercise and who decides whether the evidence supports another trial. Keep the original cases available so a later change can be compared against the same expectations.