Module 00
Baseline
Record one working week exactly as it happens, before changing anything about how you work.
You cannot redesign what you have not measured.
Movement I — Find out what is true. One week recorded as it actually happens, before anything changes. Everything after this is measured against it.
Why this comes first
Every account of how someone works is a reconstruction, and reconstructions are flattering. Asked to describe your week you will produce a coherent story: you delegated the routine work, reviewed carefully, and spent the remainder on the hard problem. The week itself was not coherent. It contained a change you accepted without reading closely because it was late, an hour lost re-explaining context an agent had already been given twice, and a task you did by hand because handing it over felt like more effort than doing it.
None of that is visible in the reconstruction, and all of it is where the redesign has to happen. So the first module produces no improvement at all. It produces a record.
What you are recording
Four things, and only four. What you handed to an agent. What you rewrote or repaired afterwards. What you reviewed, and how closely. Where the time actually went.
The fourth is the one people skip, and it is the one that carries the finding. Time spent supervising, re-prompting, re-explaining and repairing is real work that no tool reports, and it is invisible in every metric your organisation already collects. If it turns out to be a third of your week, that is the most important number you will produce in this program.
Record decisions, not keystrokes. This is not time-tracking software and it is not a productivity audit — a note per meaningful handoff is enough, written at the time rather than reconstructed on Friday.
Method
- 01Pick an ordinary week. Not a release week, not a quiet one.
- 02Keep one running note, appended to as things happen. Reconstructing at the end of the day defeats the exercise.
- 03For each handoff to an agent, record: what you asked for, what came back, what you did to it before accepting it, and roughly how long the whole cycle took.
- 04At the end of each day, note anything you chose not to delegate, and why. The refusals matter as much as the handoffs.
- 05At the end of the week, total the time in three buckets: work you did, work you supervised, and work you repaired.
How this goes wrong
Recording a typical week
There is no typical week, and the attempt to find one delays the module indefinitely. Take the next ordinary one and note the ways it was unusual.
Changing how you work while measuring
Observing yourself makes you better at the thing you are observing. Expect this, and note where you caught yourself performing rather than working — that note is itself a finding about which parts of your practice you already know are weak.
Counting output instead of decisions
Commits, lines and tickets are already counted by systems that were not designed for this and will mislead you. The unit here is the decision to hand something over, and what happened next.
The exercise
- Record one working week without changing how you work.
- Produce the three time totals: done, supervised, repaired.
- Write two paragraphs on what surprised you. If nothing surprised you, write why you think that is — either your practice is unusually deliberate, or the record is not honest yet.
Baseline week
The sections your artifact contains:
- The week, and how it was unusual
- Handoffs — what was asked, what came back, what it took to accept it
- Refusals — what you kept, and why
- Time: done / supervised / repaired
- What surprised you
Done means. Numbers rather than impressions, and at least one entry you would not want to show a colleague. A baseline with nothing uncomfortable in it has been edited.