The AmplifierPractice · Programme
The First Fortnight: your first full run
"Apply the framework to one piece of real work in your life within two weeks of finishing the book. Do not wait until you feel ready. You will not feel ready. Begin anyway." Here is how.
Planned for release phase 2. The page is complete as a draft; interactive parts run in the browser and do not yet save to an account.
In brief
Fourteen days, one real project that is already on your desk. Days 1–2 are Step 1, alone. Days 3–6 are Steps 2 and 3. Days 7–10 are Steps 4–6. Days 11–13 are evaluation, integration, and the structural problem you should expect to find. Day 14 is the write-up, by hand. The point of the first run is not to do it perfectly; it is to do it consciously.
Days 1–2 · Define the End State (Step 1)
A weekend, ideally. A notebook, ideally — the point is that this is done somewhere the tools are not. Write a heading and sit with the blank page; not knowing what to write is the actual starting point. Then paragraphs: who is this for, named; what should it feel like when they pick it up; what are the constraints. Read it back once and ask whether it is a deliverable list. If it is, go again. Then stop. Do not push past the step you are on.
Days 3–4 · Map the Flow (Step 2)
Phase A on paper: every task, sequenced, dependencies drawn. Expect embarrassment that you have never done this for this project before. Phase B: the review prompt from the workbench. Phase C: argue with each gap; add, shrink or decline with a reason. Budget an afternoon.
Days 5–6 · Assign Tasks (Step 3)
Every task through the four questions. Almost none will land cleanly the first time, and this is not failure — it is the filter telling you the tasks are too coarse. Decompose. Some tasks will merge. By the end you have three columns: human, AI, hybrid with the parts named. This is the slowest part of the fortnight and the part that makes the rest fast.
Days 7–8 · Select tools and prepare (Steps 4 and 5)
Category first, name second, for each AI task. Then, before opening anything: context, expected output, criteria, in that order, criteria written down. Brief in your own language; let the tool optimise. If you catch yourself giving a three-line brief and getting fluent generic output back, you have found the most common Step 5 failure — do it properly and compare the two drafts side by side. The difference is the lesson.
Days 9–10 · Execute (Step 6)
Do the human tasks; direct the tool on its tasks with what you prepared. Notice the speed. If the earlier days were done honestly, this phase will move faster than you are used to, and the proportion of your time spent thinking versus doing will feel backwards. It is not backwards. It is the framework.
Days 11–12 · Evaluate and integrate (Steps 7 and 8)
Every output against the written criteria. Expect small fixes — a renamed thing, a stale reference, a missing buffer — and handle them without alarm. Then assemble everything and look at it as one piece against the Day 1 end state.
Day 13 · The redo
Usually a few hours, rarely a day. Rewrite the end state with the audiences named. Reframe the pieces that were for nobody in particular. Reassemble. Check for a centre of gravity.
Day 14 · Write it up, by hand
A status note, a summary, a message to whoever it was for — written yourself, no tool. It will be clumsier than what the tool would write. It will also, unmistakably, be yours. Then note three things in the notebook: where you skipped, where you struggled, where old habits pulled. Those three notes are the brief for your second run.
Then do it again.
The pattern needs to hold across three or four projects before the new posture is settled. The second run is faster because the friction of remembering is gone.