The method
A portable way to work out how much of any workflow can be automated, which parts must stay with a person, and where to start.
- Status
- stable
- Trust
- human-reviewedhuman:rrizwan98 ·
- Owner
- human:rrizwan98
- Approved
- human:rrizwan98 ·
This room is the method, not a story about one job. It applies to any workflow in any business: software delivery, design, recruitment, support, operations. Whoever is reading, the procedure is the same.
Read it in this order
- Count your own workflow: write the run down, tag every step, and get your own number instead of borrowing someone else's.
- What automates and what does not: the rules that decide a step, in any domain.
- Worked counts: the same method applied to five different businesses, so the pattern is visible rather than implied.
- From one task to an AI-native company: what happens after the first workflow works.
The one idea underneath all of it
AI made the typing fast. It did not make delivery automatic, because a workflow is not one action: it is a run of steps, and a person is still doing four jobs inside it.
- writing every instruction
- re-explaining the project after every context reset
- redoing the work the AI got wrong
- remembering the steps nobody wrote down
Automation means moving those four out of a person's hands, and leaving the decisions in them. So the question is never "can AI do my job". It is: in my run, which steps are decisions, and which are just typing?
What a number from this method means
A percentage produced here is a count of steps in one named workflow, on a stated date, with each step tagged. It is not a promise about time saved, money saved, or headcount. Two businesses in the same industry will get different numbers, because they run different steps.
If you want the number for your own work, the counting page is a procedure you can run yourself in twenty minutes, and the audit is the same procedure done together.