The Toolkit
Same OTA system, new engine.
Parts II and III of the book build a human Toolkit for running the OTA loop deliberately. Part IV does not add a second one: it shows what changes when AI becomes the engine, with human leadership in charge of it. This page is the map, not the manual — each entry says what a tool does in a sentence or two. The instructions and the worked examples are in the book.
Where these tools come from
A human Toolkit, then three kinds of change
AI changes what’s possible with that same Toolkit in three distinct ways — which is why the entries below split into three groups, not one flat list. Nothing has been retired for being wrong; the human-tool lineage stays visible throughout.
- Group 1 — AI-augmented: same tool, new engine.
- The human tool is unchanged in function; AI runs it faster, wider, and with less manual effort. Core Competence Identification, for instance, is the same audit — AI assists the benchmarking. The signal brief sits here as an output of the observation system, not as a Toolkit tool.
- Group 2 — AI-transformed: recognisable logic, greater continuity or integration.
- The human tool’s purpose carries forward, but AI changes its scope, continuity or integration — from periodic to continuous, from a document to a live shared record. A new name is used only where a genuinely distinct software component appears: the Decision Record becomes the Decision Ledger, Reality Instrumentation becomes the Reality Lattice, and Red Team becomes the Adversarial Engine. Elsewhere the human name stays — the Automated Pre-Mortem, continuous Scenario Mapping, continuously monitored OKRs.
- Group 3 — New software components and a governance mechanism.
- No pre-AI equivalent exists. The Calibration Tracker, the Sentinel, the Meta-Loop, and the Strategy Coherence Engine are only possible once AI can watch the other tools operate. The Autonomy Ladder sits alongside them as a governance mechanism rather than software: part of Decision Architecture, it is how AI earns, holds and loses decision authority.
Trace any “Evolved from:” label below back to the human tool it names — the book treats each one fully in Parts II and III, and Part IV explains only what AI changes.
Group 1 — AI-augmented
5 entries
Signal briefOutput
OTA phase: Observe
A short synthesis of the most significant signals, prepared before key meetings — AI-generated from instrumentation data; a human edits and approves. An output of the observation system, not a separate Toolkit tool.
Human-era form: Parts II–III. What AI changes: Part IV, Chapters 13–16.
The Thinking Pipeline
OTA phase: Think
A six-step structured-reasoning method (facts vs. interpretation, competing hypotheses, load-bearing assumptions, reference classes and base rates, first- and second-order effects, falsification); AI increases data depth, option generation and consistency.
Human-era form: Parts II–III. What AI changes: Part IV, Chapters 13–16.
Five Modalities Diagnostic
OTA phase: Act
A structured assessment of organisational strength across Direction, Structure, Processes, Capability, and Culture; AI assists the data gathering.
Human-era form: Parts II–III. What AI changes: Part IV, Chapters 13–16.
Core Competence Identification
OTA phase: Think → Act trigger
Identifies the capabilities an organisation must protect and develop to remain viable; AI benchmarks them against external evidence.
Human-era form: Parts II–III. What AI changes: Part IV, Chapters 13–16.
Decision Architecture
OTA phase: Substrate
Maps who — human or AI — may decide what, at which level and under what escalation conditions. AI becomes a possible decision-maker; its authority expands or contracts by decision class through the Autonomy Ladder, the governance mechanism within Decision Architecture.
Human-era form: Parts II–III. What AI changes: Part IV, Chapters 13–16.
Group 2 — AI-transformed
9 entries
Reality Lattice
Evolved from: Reality Instrumentation + Instrument Audit
Merges periodic instrument audit and sensing into continuous monitoring, and serves as a shared data source for the rest of the Toolkit.
Human-era form: Parts II–III. What AI changes: Part IV, Chapters 13–16.
Automated Pre-Mortem
Evolved from: Pre-Mortem
Fires automatically on significant Decision Ledger entries, drawing on historical cases, conditions in the assumption register and Adversarial Engine challenges; its flags feed the Sentinel.
Human-era form: Parts II–III. What AI changes: Part IV, Chapters 13–16.
Adversarial Engine
Evolved from: Red Team
Maintains continuous challenge against the organisation's live assumption base, with external provenance required.
Human-era form: Parts II–III. What AI changes: Part IV, Chapters 13–16.
AI-enabled Scenario Mapping
Evolved from: Scenario Mapping
Continuous Scenario Mapping: periodic review becomes continuous branch monitoring. Branches link to their assumptions, probabilities update as evidence shifts, and pre-committed thresholds trigger out-of-cycle alerts.
Human-era form: Parts II–III. What AI changes: Part IV, Chapters 13–16.
Decision Ledger
Evolved from: Decision Record
A shared, queryable record of each decision — what, when, by whom, on what information and assumptions, with what expected outcome and pre-committed triggers. Contradicted assumptions resurface affected decisions; outcomes feed the Calibration Tracker.
Human-era form: Parts II–III. What AI changes: Part IV, Chapters 13–16.
Continuously monitored assumption register
Evolved from: Assumption register
Maintains the assumptions behind decisions and scenario branches. New evidence updates confidence, scenario probabilities and dependent decisions, and exposes contradictions no individual could hold in working memory.
Human-era form: Parts II–III. What AI changes: Part IV, Chapters 13–16.
Continuously monitored OKRs
Evolved from: OKRs / direction alignment
Progress and underlying assumptions both monitored in real time — when an assumption breaks, the system flags the objective to the leader who set it.
Human-era form: Parts II–III. What AI changes: Part IV, Chapters 13–16.
AAR + machine-scale double-loop learning
Evolved from: After-Action Review
The After-Action Review's single-loop learning, combined with automatic cross-cycle pattern detection across the organisation's AAR archive.
Human-era form: Parts II–III. What AI changes: Part IV, Chapters 13–16.
Automated Domain Expert Sign-Off
Evolved from: Domain Expert Sign-Off
Automatically classifies claim-heavy decisions green, amber or red. Red blocks ratification until an independent expert review is recorded; where AI has earned Pathway Two authority on the Autonomy Ladder, it may perform the sign-off itself.
Human-era form: Parts II–III. What AI changes: Part IV, Chapters 13–16.
Group 3 — New
5 entries
Calibration Tracker
Cadence: Continuous
Scores probability-bearing outputs — from humans and AI alike — against eventual outcomes, producing calibration profiles by individual, team, topic and time horizon.
No human-era form. Introduced in Chapter 13; developed in Chapters 14–16.
Sentinel
Cadence: Continuous
Surfaces the signatures of trouble before they compound, including the six pathogens and other anomaly classes.
No human-era form. Introduced in Chapter 13; developed in Chapters 14–16.
Meta-Loop
Cadence: Continuous tracking; periodic proposals
Watches the Toolkit itself and proposes additions, retirements, recombinations and reconfigurations to human leadership for ratification.
No human-era form. Introduced in Chapter 13; developed in Chapters 14–16.
Strategy Coherence Engine
Cadence: Continuous intake; surfaced at least once per planning cycle
Reasons across mission, vision, values, strategy, OKRs, governance, incentives and budgets as a graph, surfacing contradictions within the corpus and against observed reality.
No human-era form. Introduced in Chapter 13; developed in Chapters 14–16.
Autonomy LadderGovernance mechanism
Cadence: Continuous; periodic proposals
Not a software component: the mechanism, within Decision Architecture, through which AI earns, holds and loses decision authority — one rung at a time, separately per class of decision, on Calibration Tracker evidence. Pathway One and Pathway Two are the two pathways within it.
No human-era form. Introduced in Chapter 13; developed in Chapters 14–16.
Looking for the concepts?
Calibration, WWHTBT, calibrated trust…