The problem nobody has fully solved
The consulting apprenticeship worked for a century on one mechanism: juniors produced under review, and judgment rubbed off through the reps. AI now produces the first draft of nearly everything a first-year used to make, faster and often better. If the production reps disappear, where does the judgment come from? Sessions a1 through a5 gave you the evidence, the pricing, the rails, the review discipline, and the operating model. This session closes the track with the two things left: rebuilding the ladder your people climb, and having the AI conversation with clients on your terms instead of theirs.
The apprenticeship problem 8 min live
The traditional ladder was never really about the deliverables - it was a judgment factory disguised as a production line. Juniors learned what good looks like by producing under review, thousands of times. AI removes the production reps without removing the need for the judgment they built. Here is the old ladder, the new one being assembled in its place, and the gap between them that nobody has closed.
LiveThe honest answers emerging4 min▶
Four answers are taking shape across the industry, and each moves something that used to be absorbed into something that is now taught. Juniors become reviewers earlier: the review pyramid from a4 gets taught explicitly in year one, instead of soaked up by year five. Frontier-mapping becomes a taught skill: the b1 exercise - mapping your own task list against where AI helps and where it quietly fails - is now onboarding material, not senior intuition. Deliberate reps on the craft measures: the consultant track's live scorer exists exactly for this - its ladder from 19 to 100 is a training tool, letting a junior feel the difference between an intern draft and a client-ready one in twenty minutes instead of two years. And the institutions are signaling: Deloitte's AI Academy and PwC's upskilling investment show the big firms formalizing what apprenticeship used to deliver informally - cite them as signals, not proof.
And the honest remainder: pattern recognition from volume is not replaced by any of this. A consultant who produced three hundred market scans has instincts that a consultant who reviewed three hundred AI drafts may not build the same way. Fewer reps at producing means slower instinct-building, and no academy has demonstrated a full substitute. Say that to your bench plainly - they already suspect it, and your credibility on the solved parts depends on your honesty about this one.
Self-studyWhy the scorer is a training tool, not a grading tool2 min▶
The consultant track's live scorer reads real text and prices real habits: answer-first openings, quantified support, so-what density, weasel-word discipline, client anchors, length. Watch what happens when a junior runs the ladder from 19 to 100: the lesson is not the score, it is feeling which specific edits moved it - putting the answer first, attaching the number, drawing the consequence. That is judgment training compressed. The old ladder delivered the same lesson across two years of markups; the scorer delivers a concentrated version in a session. It does not close the volume gap - concentrated reps are not the same as thousands of live ones - but it is the most honest partial substitute this course knows of, which is why the whole consultant track is built around it.
The bench curriculum 7 min live
If production is no longer the curriculum, something explicit must replace it. Here is what your juniors learn now - each item mapped to where the consultant track teaches it, because your bench can literally take that track while you run this one.
| What juniors learn now | Where it is taught | What it replaces |
|---|---|---|
| Frontier mapping - knowing where AI helps and where it quietly fails | Consultant session b1 | Years of trial-and-error delegation instincts |
| Source discipline - the source log; an AI fact is a lead until verified | Consultant session b3 | Learning citation hygiene from one bad incident |
| The six craft measures - answer-first, quantified, so-whats, no weasels, client anchors, length | Consultant session b5 (the scorer) | Two years of partner markups, absorbed slowly |
| Anonymization reflexes - client material stripped before any prompt | Consultant session b8 | Hoping nobody pastes the wrong thing |
| Citation audit - the QA gate before anything reaches a client | Consultant session b9 | Trusting that production details are someone else's job |
| Client empathy + the courage to disagree with a partner | Never automated - taught by example, yours | Nothing. These two never left the human column. |
LiveThe two things that never automated3 min▶
Everything in the table's first five rows is teachable through structured reps. The last row is different in kind. Client empathy - reading the room, hearing the question behind the question, knowing which true thing not to say this week - has no scorer and no academy module. The courage to disagree with a partner - saying "I think the data cuts the other way" up the hierarchy - is the single behavior that catches the errors everything else misses, and it is trained by exactly one thing: watching what happens to the last junior who tried it. If you praised the catch (a4's ritual), you trained courage. If you punished it, no curriculum on this page can help you.
Self-studyThe hiring signal changes too2 min▶
If the bench curriculum is judgment and verification, the hiring filter should be too. The old screens quietly rewarded production speed: case math under pressure, polished formatting, fast slide assembly - all things AI now does in the first minute. Test for judgment and verification instinct instead: hand candidates a fluent, confident, subtly wrong AI draft and watch what they do. The candidate who spot-checks the numbers against the source before praising the prose is the one whose instincts fit the new ladder. Formatting speed tells you almost nothing anymore; the willingness to distrust a beautiful paragraph tells you nearly everything.
The client conversation about AI 6 min live
Have it before they ask. The firm that raises its own AI use, calmly and specifically, controls the frame; the firm that waits gets asked in a procurement questionnaire or, worse, after a headline. The script has four beats, and every one of them is a promise you already built in this track.
LiveWhy each beat holds weight4 min▶
Beat 1 - "we use AI, deliberately, inside approved enterprise tools" - preempts the gotcha. A client cannot expose what you disclosed first, and "deliberately" signals an operating model (a5), not a free-for-all. Beat 2 - "your data never trains anyone's model and is anonymized in use" - is a3's rail spoken plainly: enterprise and commercial tiers do not train on business data by default, and you anonymize even inside them. Say it in those words; clients hear precision. Beat 3 - "a named human owns every fact you receive" - is the Deloitte-shaped promise: the incident every sophisticated client now knows was about unverified production details, and your named-owner rule is the exact countermeasure. Beat 4 - "you get the gains as speed and depth, not as a discount surprise" - is a2's pricing posture delivered before the "you used AI, why am I paying?" question arrives, on your terms.
Write your two scripts ✍️ ★ 18 min · everyone works
Two artifacts, both in your own voice, both leaving this room finished: the client script and the bench announcement. Reading them aloud is part of the exercise, not a suggestion.
Draft the four-beat client script in your own voice. One paragraph, all four beats, no jargon. It should sound like you across a table, not like a compliance page.
Draft the internal announcement of the new bench curriculum. What juniors will now be taught (the table from part 2), and why review is promoted work, not demoted work. That sentence, or your version of it, must appear.
Read both aloud. Anything you stumble on, a client or a junior will stumble on harder. Rewrite until it sounds spoken, not drafted.
Cut every weasel. "Leverage," "robust," "best-in-class," "committed to" - the scorer's weasel measure applies to your scripts too. Every cut makes the promise more falsifiable, which is the point.
Test the client script against the hardest question: "did AI write the report we paid for?" Write your honest two-sentence answer. If you cannot answer it in two sentences without flinching, the script is not done - beats 3 and 4 are your raw material.
Date both drafts. The client script gets delivered this month (homework makes it binding); the bench announcement goes out when you can stand behind every line of the table.
Homework 🎓 ◐ this month
- Deliver one script for real this month. The client version to an actual client, unprompted, before they ask. The conversation is easier than you expect and gets easier each time.
- Pick one junior and teach the review pyramid explicitly. Not "shadow me" - an actual session on the a4 pyramid and the six craft measures, with one real deliverable as the material.
- Book the quarterly re-read of the frontier map. It moves. The a5 operating-model review date and this one belong in the same calendar slot, recurring, with your name on it.
Three questions before you go 🎯 ◐ 90 seconds
1 · The apprenticeship problem, in one sentence, is...
The ladder was a judgment factory disguised as a production line. The emerging fixes - review taught earlier, frontier-mapping as onboarding, deliberate reps on the scorer - replace some of it; pattern recognition from volume remains honestly unsolved.
2 · In the bench curriculum, the two things that never automated are...
The first five rows of the curriculum are teachable through structured reps; empathy and upward courage are trained by example - specifically by what happens to the next junior who says "I think the data cuts the other way."
3 · The four beats of the client script are...
Each beat is a promise built earlier in the track: the operating model (a5), the confidentiality rail (a3), the liability rail's named owner, and the pricing posture (a2). Delivered before the client asks, it wins the frame.
What this session covered from the official universe
Certificates, videos, and graded assessments stay with the official providers - this course teaches the working core.