AI Workforce · Level 0

See the work, not the tools

Continue

0.1 One person is not an organisation

Level 0 · 21 min · 2 knowledge checks

0 of 5 gates cleared0 of 14 visible items read

Most independent professionals and small business owners operate as though every task requires them personally. Some of it genuinely requires their judgement. Much of it does not, and the difference between the two is invisible until someone looks for it deliberately. This level is that look. Before you select a model, write a prompt or open an account, you will build an inventory of what your work actually consists of, because the first step in building an AI workforce is not selecting tools, it is understanding the work.

The constraint on a one-person organisation is not talent and it is not hours. It is that everything routes through one person's attention. A one-person smart organisation is something different: a person who has deliberately designed a system around themselves so that routine, repetitive, analytical, administrative and preparatory work does not depend entirely on their personal effort. The person keeps what only a person can hold: direction, judgement, relationships, priorities, quality, approval, accountability. Everything else becomes a candidate for a seat in the workforce you will build from Level 1 onward.

This week produces two artefacts: a classified inventory of one real working week, and a short list of tasks that stay on your desk only because no system exists. It also installs two numbers, a 50 per cent figure and an 80 per cent figure, that will govern every delegation decision in the rest of the course. The reason for measuring rather than reflecting is in this level.

Level 0 lessons

0 of 4 done

You cannot delegate work you have not defined.

0.1One person is not an organisation21 min 0.2What the capability numbers actually say18 min 0.3Reliability is not capability22 min 0.4The bottleneck test20 min

Level 0 exercises

0 of 2 done

Done against a real organisation you choose, not a case study. The gates below are about what these produce.

Exercise 0.A - The tracked week

Track five consecutive working days. Record every meaningful task as an event: what you did, what it produced, and why you personally did it, in one line each at the time or at day's end. Recording takes about fifteen minutes a day. At the end of the week, classify every entry into the six kinds of work and apply the four bottleneck questions to each. The produced artefact is the classified inventory: a list of real events, each carrying a category and a bottleneck verdict. The exercise fails if tasks are recorded as categories rather than events. "Admin, 2 hours" is not a record; "prepared and sent the March invoice to a client, 20 minutes, did it myself because it was faster than explaining" is. The bottleneck test cannot be run on a category, and an inventory of categories is a diary of impressions, which you already had.

Exercise 0.B - The bottleneck findings

From the inventory, name the three largest blocks of your week that failed all four bottleneck questions. For each, write what a system would need to exist for the task to leave you: what inputs it would need access to, what output it would have to produce, what check would confirm the output is right, and what would still come to you. Three findings of a paragraph each, roughly 45 minutes. The exercise fails if every task somehow requires you, which is a finding about honesty, not about the work. A week in which nothing fails all four questions has not been tested; it has been defended.

Level 0 gates

0 of 5 cleared

You can list the tasks you performed last week and assign each to one of the six kinds of work without a residual miscellaneous pile.

Not started

You can name, for any task on your list, which of the four bottleneck questions it fails, or state that it fails none.

Not started

You have classified ten real tasks as human-led, AI-assisted, AI-delegated with review, or a candidate for automation, and can defend each classification.

Not started

You have identified one task you perform only because no system exists, and written what the system would need.

Evidence needed: Attach the artefact, or attest with a date if it lives somewhere you cannot upload from.

Not started

You can state the difference between a capability figure quoted at 50 per cent reliability and one quoted at 80 per cent, and say why the second governs delegation.

Not started

Tools in this level

Work inventory

Your real tasks from last week, run through the bottleneck test.

Open and get it reviewed

The ten statement check

Rate yourself before and after; read the lines, not the total.

Open and get it reviewed

Desktop only

Drag-and-drop tools stay on a wider screen. Lessons and checks work anywhere.