A Reader’s Companion

The Field Guide

Every protocol in the book, gathered for the working week.

Ideas persuade once; protocols get used weekly. This section collects every working tool from the preceding chapters in one place, stripped to the steps, each with its home chapter for the full reasoning. Dog-ear this section. It is the part of the book built to get dirty.

Sequencing Your Work

The Sequence Rule (Chapter 5)

Engage → Augment → Evaluate. Never Augment → Accept. Before any AI tool touches significant work: fifteen to twenty minutes of your own thinking, however rough. The order in which you and the tool meet the problem decides whether the tool deepens you or replaces you.

The Position Brief (Chapter 10)

Twenty-five minutes. Then instruct the AI to attack the brief, not merely complete the task. Attention goes to wherever the output and the brief diverge — one of you is wrong, and finding out which is the actual work.

The Stakes Dial (Chapter 14)

Full delegation for routine, judgment-free tasks — feel nothing. Full framework where your judgment is the product or your name carries the consequence. Human-first, AI-as-critic-only for the highest-stakes personal communications. Choosing the setting takes five seconds and is itself a judgment skill that sharpens with use.

Evaluating What the Machine Produces

The Verification Rule (Chapter 10)

Any fact your conclusion stands on, you touch at the source — personally, every time, no exceptions for polish. Identify load-bearing facts by asking: if this one is wrong, does the recommendation collapse? Those are yours. Everything else can be delegated.

The Three-Pass Review (Chapter 2)

Pass one, structure — headings and first sentences only: does the argument build or merely flow? Pass two, claims — circle every sentence a hostile reader could challenge; the circles you cannot defend are your verification list. Pass three, voice — one page read aloud; replace every phrase you would never say to the person’s face. Twenty-five minutes total, and each pass trains the eye that runs the next one faster.

The Byline Test (Chapters 6, 11)

Before shipping: would someone who knows your work recognize this unsigned? If it could have come from anyone with the same prompt, put yourself back in until it couldn’t. For creative work, the stronger form: generate many, kill most with written reasons, transform the survivors until they carry your signature. The kill-reasons file is your taste, compounding on paper.

Protecting the Capacity

The Deep Block Protocol (Chapter 3)

Expect two uncomfortable weeks; expect the tally to fall. Start at forty-five minutes if ninety is impossible. The falling tally is your attention recession, reversing, in your own handwriting.

The Monthly Unassisted Task (Chapters 14, 18)

Once a month, one significant task in your domain completed with no AI at all — chosen by you, logged without ceremony. Not ideology: the capability check the endoscopists never scheduled. If the unassisted version has become alarming, you have learned the most important fact on your dashboard.

The Ten-Year Ledger (Chapter 4)

One page, revisited annually. The slow skill your career most depends on, named specifically. This year’s deposits: two stretch assignments you might fail at, one honest post-mortem per quarter, two mentor sessions on your patterns. And the running list — ‘what I can do now that I couldn’t a year ago’ — because slow-skill growth is invisible daily and unmistakable annually.

Keeping the Human Infrastructure

The Mentorship Hour (Chapter 7)

Monthly, structured: twenty minutes on one real artifact, reviewed out loud with the standard visible; twenty on one live problem, reasoning made audible; fifteen on pattern — what the mentor observes in you across months; five on one written commitment, checked next time. Runs in both directions: keep one mentor, be one mentor. Remote works if presence is total.

The Unrecorded Room (Chapter 8)

Quarterly. Four to eight people whose trust matters. Phones in a basket, reasons stated. Two questions only: what is actually going on that we don’t say in official rooms, and what are we pretending is fine? Three rules: nothing attributed, nothing summarized afterward, silence allowed. The first one is awkward once. What returns arrives through channels no dashboard carries.

The ‘You First’ Household Rule (Chapter 13)

The working script for homework nights: ‘Show me your try first — then let’s see what it says and whether we agree.’ Then the question that does the teaching: ‘Where was your answer different? Who do you think is right?’ Written together, the rules become identity, which is the only form a ten-year-old still carries at twenty.

Leading and Measuring

The Decision Journal (Chapter 12)

For every consequential call, before any AI counsel: the decision, your reasoning, the predicted outcome with a date. Note the machine’s analysis after yours. Review quarterly against reality. Five minutes per entry — judgment’s gym, and the only feedback loop most executive judgment ever gets.

The Leader’s Floor Hours (Chapter 12)

Twice weekly, protected like board meetings, at whatever level the reports are quietest about. Two questions, then genuine waiting: what is harder than it should be right now, and what do you know that you think I don’t? Expect pleasantries for a month; the channel opens after you are seen acting on something uncomfortable without punishing its source. Friday close, one line: what did I learn this week that no system reported?

The Depth Deficit Index (Chapter 16)

Twenty statements, five dimensions — Direct Engagement, Evaluation, Presence, Tenacity, Held Capability — scored 0–5 for the last thirty days. Depth Score out of 100; DDI is 100 minus it. Anchor with evidence: the calendar’s actual deep blocks, the position briefs that actually exist, the date of your last unassisted work. Then one 90-Day Cycle: lowest dimension, mapped practice, retake. Trend line over single score, always. Mirror, never weapon — tie it to ratings or pay and Goodhart’s Law collects immediately.

The Team On-Ramp (Chapter 17)

One team, one quarter. Weeks 1–2: anonymous baseline — individual Index plus the five-condition audit, leadership and staff rating separately; publish both, including the gap. Weeks 3–6: fix the single lowest condition in one real workflow. Weeks 7–10: add monthly unassisted tasks — the team’s manual flight hours. Weeks 11–13: re-measure, publish the trend, let the team choose the next condition. Permanent rules: anonymous and developmental forever, and leadership takes the Index first, visibly.

Fifteen protocols. None requires permission, a budget, or a platform — only the decision this book has been arguing for all along: use the machines for everything they are good at, and keep doing, on purpose, the work that keeps you worth augmenting.