AI Sanctuary

For senior executives

The AI briefing you'd want, if you had a chief of staff who'd read everything.

Six executive briefings. No jargon on the surface. Detail on demand. Practice on the tools the world actually uses — ChatGPT, Claude, Perplexity, Lovable. Ascend from novice to guru.

Roughly 30 minutes of reading. A morning to master.

What the Compass measures

The Executive AI Compass

AI is the fastest ground-shift most executives will live through. The temptation, as always, is to respond with more — more tools, more automation, more efficiency. I have watched organisations make this mistake with every wave of change: investing in the machine, forgetting to ask what it leaves for the human being.

The future won't belong to the best title or the most decorated CV. It will belong to those who know what they alone bring — seeing a system whole, understanding what moves people, turning complexity into a clear path.

AI doesn't replace the person. It replaces the version of the role where the person had quietly become a mechanical function.

The Executive AI Compass asks one question: What is the strength in me that technology can amplify, but never replace?

Four capabilities — Understand, Direct, Redesign, Lead. Know the answer, and you steer the technology. Skip it, and someone else steers it for you.

Four capabilities

Understand — see systems, tell real capability from a sales slide.
Direct — explain complexity, brief experts, price the result.
Redesign — turn chaos into a clear path, rebuild work with control points named.
Lead — understand people, defend evidence, capital and accountability.

How the journey works

Four places, one climb

Compass

See where you stand across Understand, Direct, Redesign, Lead.

Temple

Eight executive briefings — summary first, appendices on demand.

Lab

Apply each briefing to your own work with the real tools.

Playbook

Your artifacts, decisions and prompts, kept in one place.

Five stages

The master's path

Rank is not years. It shows in composure, in bearing, in how little needs to be said. You begin attentive and unsettled; you end able to see the whole terrain at once.

The Novice

Stage 1

The Apprentice

Stage 2

The Adept

Stage 3

The Sage

Stage 4

The Guru

Stage 5

The curriculum

Eight executive briefs, in order

Executive Brief 01

What AI actually is — for a P&L owner

Modern AI is a probability engine trained on enormous quantities of text and code. It predicts the next useful word or line. That single fact explains its strengths (speed, breadth, fluent drafts) and its weaknesses (confident-sounding errors, no real-time knowledge, no accountability). Treat it as a very fast, very well-read junior analyst who never sleeps — and who must be checked.

Executive Brief 02

Choosing the machine — a selection discipline, not a feature tour

You already know what the machine is and roughly what it costs. The Apprentice question is different: on what basis do you choose one, and what does that choice commit you to? Stop comparing feature lists — vendors rewrite those monthly. Choose on task class, on where your data is contractually allowed to sit, and on how expensive it will be to leave. Every tool decision is a small reversible bet, unless you let it quietly become an irreversible one.

Executive Brief 03

Getting expert work from AI

Take role, audience, length and examples as given — that is the Apprentice's craft. The Adept problem is harder: what belongs in the context window, what must be kept out of it, and how a hard question gets decomposed so the model is never guessing at the part that matters. At this altitude you stop writing prompts and start designing inputs: curated evidence, explicit decision criteria, forced disagreement, and comparison of competing outputs. The model is the same; the ceiling moves because your inputs did.

Executive Brief 04

The Economics of Intelligence

AI is not simply another productivity technology. It changes the marginal cost of producing many forms of knowledge work — the cost of the next report, the next analysis, the next draft. Analysis, drafting, coding, research, translation, synthesis and scenario generation can now be produced at far lower cost and far higher volume than a person could produce them. But cheap production does not destroy value; it relocates it. The question every board now faces is where value has moved, and whether the company has repositioned itself to hold it. When intelligence becomes cheap, judgment becomes expensive — not as a slogan, but as a matter of supply and demand.

Executive Brief 05

Workflow design

You can now get expert output from a general model. The Sage question is organisational: which parts of a real process should be delegated to a machine at all, where the human checkpoint must sit, and how you know afterwards whether the redesign paid. Individual fluency reaches its ceiling fast. Redesigned processes compound — and they fail in specific, predictable ways when the checkpoint is placed for convenience instead of where the liability actually sits.

Executive Brief 06

The AI that works while you sleep — standing agents in your own channels

Everything so far has assumed you are present: you open a window, you ask, you read the answer. The senior move is different. A model can now sit inside the places your team already works — the channels, the tracker, the CRM, the document store — and be given standing instructions rather than prompts. It reads the thread and the connected systems, does the task, posts the result with its sources, and can run to a schedule or act on its own when it sees a reason. That shifts your job from asking well to specifying well: what it may reach, whose identity it acts under, when it speaks, and what it is never allowed to decide.

Executive Brief 07

Decision-grade evidence — adversarial research at the top table

Verification hygiene is assumed: you click the citation, you read the source. The Guru problem is that a well-sourced, entirely accurate synthesis can still be worthless for a decision. Search-based tools return the consensus — the most repeated version of a story — and consensus is precisely what your competitors are also reading. Value at this altitude comes from provenance, from asymmetry, and from research deliberately conducted against your own thesis rather than in support of it.

Executive Brief 08

Capital, control and accountability

Policy, accountability and data classification are settled ground by now — you have them or you are negligent. The final ascent is harder and less discussed: AI spend behaves unlike any line item you have governed before, competitive advantage from it decays faster than the investment cycle that funds it, and the org chart you inherited was designed for a cost of production that no longer holds. This module is about capital allocation, defensibility and structure — the three decisions that cannot be delegated downward.