SHEET 05 · THE TRUST ARCHITECTURE DIAGNOSTIC
You're deploying AI on your team.
You're running it on yourself.
Nobody has mapped either.
Three 90-minute sessions over roughly a month, 1:1 with me. Your calibration, your floor, and a prescription you can run without me.
THE DIAGNOSTIC
3 × 90 minutes · 1:1 · roughly a month
$2,000
£1,575 · DIRECT BOOKING · NO PROCUREMENT
The company pays. The work is with you.
The entry point. The Floor Map runs this across your leadership line.
A-01
WHO THIS IS FOR
DEPLOYING
AI in your team
You've bought the tools, the dashboard says something, and you don't fully believe it. You can feel the gap between reported adoption and what's actually happening on your floor.
RUNNING
AI on yourself
You use it heavily, daily, across everything. It's made you faster and possibly more scattered. And your team adopts to the level of what you actually do with it, not what the rollout plan says.
DEEP USERS
The relational layer
For leaders whose AI use has become genuinely personal - daily, load-bearing, sometimes hard to describe. What's happening, what's real, and how to hold it well.
Most leaders are at least two of these at once. One buyer, one calendar. No procurement process, no discovery workshops, no deck. We book, we start.
A-02
THREE SESSIONS · THE GAPS ARE WHERE THE EVIDENCE GETS MADE
SESSION 01
WEEK ONE
≈ TWO WEEKS
SESSION 02
≈ WEEK THREE
≈ TWO WEEKS
​
SESSION 03
= WEEK FIVE
Map the Leader · the ceiling first
We start with you, because your calibration is the deployment's ceiling. Your own AI practice, your history with it, the pressure sitting behind this deployment, and what happens to you if it fails. Before we meet you'll have completed a preparation questionnaire - about thirty minutes, and it shapes everything. You leave with an evidence brief: a structured fortnight of watching your own floor. Not opinions, not labels - observable behaviour. Who submits AI-touched work without saying so. Who asks permission nothing requires. What the usage data shows against what people claim in meetings.
You collect. I stay out of it.
Map the Floor · through the evidence
We walk your evidence piece by piece and I type the patterns against the taxonomy - always to the engine underneath, never just the visible behaviour, because the same face can run on three different engines and each one takes a different intervention. Just as important: who never appeared in your evidence at all. What you couldn't see from your chair is its own finding, and usually the most valuable one. You leave with one intervention - small, low-stakes, chosen together, briefed with its failure mode and a defined observable.
The intervention runs, as management — not as an experiment.
The Prescription · the road ahead
We debrief the intervention - whatever happened, it's data, including if it backfired exactly the way the failure mode predicted, which means the map is reading your team correctly. Then the full prescription: every typed pattern matched to an intervention, sequenced, each with its failure mode and its observable. Your own practice change is in the sequence with a date on it, because the ceiling doesn't move by aspiration. The measurement framework becomes yours, on a monthly rhythm small enough to survive your calendar. And where the work has limits, I name them - if direct sessions with specific individuals would move things, I'll tell you which ones and why, as a finding. You decide. The Diagnostic stands complete either way.
A-03
WHY THIS ISN'T A QUESTIONNAIRE
Why this isn't a questionnaire
The framework this rests on says self-report is identity performance - ask people how they use AI and you get an account of who they'd like to be at work. A fair reader will ask why a diagnostic with a preparation questionnaire survives its own rule. Three reasons.
01 Evidence, not recall
The map is built from a structured fortnight of watching your actual floor - who submits AI-touched work without saying so, what the usage data shows against what people claim in meetings. Observed behaviour, collected against a brief. Not your impressions of your team.
02 Accounts against accounts
Three sessions exist so accounts can contradict each other. What you write before session one gets tested against the evidence in session two and what the intervention does in session three. A survey captures one performance at one moment. This is built to catch the performance moving.
03 The performance is the data
Your account is read as material, not taken as record. Where your telling of the floor is vague, or defensive, or a little too fluent, that's a finding. The distance between the questionnaire and what's on the map by session three is usually one of the most useful things the Diagnostic produces.
A-04
WHAT THIS IS, AND WHAT THIS ISN'T
What this is, and what it isn't.
The taxonomy comes from three years of documented practice with frontier AI systems, inside organisations, inside my own work, recorded as it happened. It is a working practitioner framework. It is not a validated psychometric instrument, and I won't dress it as one, because "validated" is a claim that has to survive an audit and I hold you to the same rule I hold everyone: evidence, not adjectives.
What it does instead is make itself testable on your floor. The intervention you leave session two with carries a named failure mode and a defined observable. By session three, the map has either read your team correctly or it hasn't and either result is data, produced by your organisation, about your organisation. Most frameworks ask you to trust results from other people's companies. This one generates its own evidence inside the engagement, which is the only kind an auditor actually wants from you anyway.
A-05
WHAT'S INCLUDED
Preparation questionnaire
Before session one. Thirty minutes, focused, not corporate. It's your baseline account, built to be tested against the evidence, not taken as the map.
3 × 90-minute sessions
Google Meet, recorded and shared. Yours to keep and revisit.
Written documents after sessions one and two
What we found and what you're doing next - within three working days, written for you, not from a template.
The prescription pack
Within five working days of session three: the typed map of your floor, every intervention with its failure mode and observable.
The measurement one-pager
The tracking framework translated into your context, on a monthly rhythm - so the work survives my exit.
A 30-day check-in
One email, on your first observable. Because the map only matters if it's being read.
No PowerPoint · No procurement · No corporate theatre
A-06
THE RULES · SHORT FORM
I work from evidence, not adjectives - yours included. No one on your team is ever told a type, by me or by you. Interventions read as management, not experiment. And if the engagement extends to sessions with your people, nothing an individual says ever travels back to you: patterns and recommendations only, never quotes, never verdicts. That last rule is in the contract, because it's the one that matters most.
A-06
EXTENDING THE WORK
Some Diagnostics surface individuals whose patterns need direct work, usually two or three people, named because the map says so. Individual sessions are priced per person, participation is voluntary, and they're governed by the same confidentiality rule. This only ever arrives as a finding from your Diagnostic, never as a pitch.
A-07
SCALING THE WORK
UPWARD · LEADERSHIP LINE
The Floor Map
The Diagnostic reads one leader. The Floor Map runs the same engine across three to five of your leaders and maps the patterns into one synthesis, each leader's map, the cross-floor read, one intervention sequence for the whole line. Same rules: nothing an individual leader tells me travels to you, or to each other.
$12,000 · £9,450
Three to five leaders
A-08
FROM THE FIELD
CLIENT RECORD
I was using Claude before I met Gail, what I didn't have was a practice. She helped me build a daily routine with it, taught me how to make my emails sound like me instead of like AI, and showed me what to ask our tech team so Claude could actually work with our files. She even had me complete Anthropic's prompting course. Six months ago I was experimenting; now management knows me as one of the company's AI power users and that was part of the plan too.
- SALES LEAD, EDUCATION SOFTWARE · DIAGNOSTIC 1:1
Book the Diagnostic
I work with a small number of leaders at a time.
If you can feel the gap between what your adoption dashboard says and what's actually happening on your floor and you'd rather know what's in it - email me. Tell me your role and roughly what's happening with AI in your world.
I'll reply with the right starting point, the terms, and session times.