One sitting, the whole engagement
Northbridge x Cadence started as a brief in session 1 and became a shipped recommendation by session 9: the $30 per seat enterprise tier, Q1, piloted on the 210 accounts already in procurement or security review. Today the engagement compresses into a single run - not because you need new material, but because the skills only count when they fire in sequence, fast, without you consulting your notes. That is what the next real engagement will demand.
The arc, one breath 6 min live
Nine stations, one engagement, each with a single takeaway worth keeping. Where a station has a signature number, it is on the map - 19 to 100 on the ladder, 8 in the slop trap, 15 at the gate. If any station feels foggy, its session is one click away; the capstone assumes them all.
LiveStation by station, one takeaway each4 min▶
Say these out loud with the map up - the recall is the exercise. b1: the frontier is jagged and invisible; map it, and obey the two rails that outrank every gain. b2: the hypothesis tree with explicit exclusions - what you will NOT investigate is half the scope. b3: an AI-supplied fact is a lead, not a fact, until a human sees its source; the log starts day one. b4: findings become insights without flattening - keep the outlier interviews the summary wants to average away. b5: the pyramid, and the ladder from 19 to 100 - answer first, quantify, so-what, anchor.
b6: the slop trap - the all-AI deck scored 8, the consultant-owned one 100, same tool, different division of labor. b7: the voice is yours, especially for difficult news. b8: the gate caps everything at 15 the moment client-identifying material appears, regardless of craft - an 87 on prose is still a 15 with a leak. b9: the citation audit (VERIFIED, VENDOR-CLAIMED, CUT) and the hostile-client rehearsal, then the readout: answer in two minutes, close with the ask.
What transfers on Monday 6 min live
Cadence was a training rig. The honest translation table says exactly what each course artifact becomes when the client is real - and what does not transfer at all.
LiveThe translation table3 min▶
| In this course | On your Monday |
|---|---|
| Cadence's data room: 62%, 210, 94% vs 71%, $38-45, $30, 2.1x | Your client's actual numbers - the anchors change, the write-with-them discipline does not |
| The scorer's six measures: answer-first 20 · quantified 20 · so-what 15 · weasel 15 · anchors 15 · length 15 | Your self-edit checklist - write it out once by hand today, and run it on every summary before anyone else reads it |
| The forbidden list behind b8's gate | Your engagement kickoff ritual: day one, name what never enters a prompt for THIS client - names, figures, board material |
| The citation audit from b9 | Your ship gate: nothing leaves without every fact stamped VERIFIED, VENDOR-CLAIMED, or CUT by someone other than the author |
What does not transfer: the fixed facts themselves, and the comfort of a scripted ladder. Real drafts do not come with a next button - which is why the paste box exists.
LiveThe three habits that survive real deadlines3 min▶
Under deadline, checklists shrink to habits. If you keep only three from this whole course, keep these. Answer-first drafting: the recommendation in sentence one, always - it forces every other sentence to earn its place, and it is the single biggest lever the ladder ever showed you. Lead, not fact: every AI-supplied number is a lead until you have seen its source; the habit takes five seconds per fact and would have saved Deloitte Australia a headline. Anonymize, then prompt: strip client-identifying material before it enters any tool, even the approved one - the gate taught you that craft cannot buy back a leak.
The jagged frontier, revisited 4 min live
In session 1 you drew a frontier map of your real task list: INSIDE, EDGE, OUTSIDE. Reread it now with nine sessions of hindsight - and notice what moved.
LiveReread your b1 map, then keep re-sounding it3 min▶
For most people, something migrated from EDGE to INSIDE - interview synthesis once you learned to keep the outliers, hostile-question generation once you learned the answer-with-source-row discipline, first-draft summaries once the six measures became your edit pass. The frontier did not move; your knowledge of where it runs did. That was the point: the study's real lesson was never "AI is good" or "AI is bad" but that knowing where the edge runs IS the consulting skill.
So the standing habit: re-sound the frontier quarterly. It moves - models improve, your tasks change, your firm's approved tools change. One hour, your current task list, three columns, honest EDGE entries. And when the day comes that you run the practice instead of the workstream, the leader track is waiting: a1 for the evidence, a3 for the policy anatomy behind the rails, a5 for the firm's AI operating model.
The ladder + your draft ▶ ★ 25 min · everyone scores their own
Three presses replay the ladder's key rungs one last time - then the box opens for YOUR text. The drafts are scripted rewrites, the measures are real pattern checks, and nothing pasted leaves the browser. That includes your real summary.
1 · Press "1 · Intern draft" - it scores 19. Before you look at the table, name the three missing measures from memory. Then check. If you named answer-first, quantified support, and client anchors without hesitating, the course did its job.
2 · Press "2 · Answer + numbers" - it scores 78. Say out loud which two moves bought those 59 points: the answer moved to sentence one, and every claim took its number. Two moves, 59 points - the ladder's whole economics in one press.
3 · Press "3 · So-what + anchors" - it scores 100. The ceiling. Notice why the anchors matter: so-whats a model can attempt, but the client anchors - Cadence named, the 210-account mechanics, the bottleneck called - are the points AI cannot supply, because your client's specifics live in your engagement, not its training data.
4 · Now the box. Paste YOUR OWN executive summary - the rewrite from b5's homework, or write one right now for a real engagement you are on. Press "📏 Score my draft".
5 · Read the coaching line honestly. Under 50 is intern territory, 50-79 is a solid skeleton, 80+ is readout-grade. Then read the table under the number - the missing points are usually so-whats and anchors, the two measures that demand you draw consequences and name specifics.
6 · Fix ONE measure and re-score. Add the missing so-whats, or the client anchors, or move the answer to sentence one - just one move. Watch the number respond. That is the ladder working on YOUR text, and it is the edit loop you will run for the rest of your career.
7 · Note the gate still runs on pasted text. If the leak scan fires on your real summary, do not be embarrassed - you just learned something important for free, in a browser, instead of in a client meeting. Anonymize and re-score.
8 · The graduation sentence. The scorer stays here, free, forever. Come back before your next readout - paste, score, fix one measure, ship.
Homework, permanent edition 📝 ◐ ongoing
- Score your next real executive summary before sending it. Paste, read the band, fix one measure, re-score. Make it the last step of your writing process, the way spell-check used to be.
- Run one real citation audit on the next deliverable you ship: fact inventory, three stamps, fresh eyes, a signature. The first one takes an hour; the fifth takes fifteen minutes.
- Teach the ladder to one colleague. Walk them from 19 to 100 with the four moves. Teaching it is the retention trick - and the moment you explain why anchors are the points AI cannot supply, you will know you own it.
Three questions before you go 🎯 ◐ 90 seconds
1 · The nine stations of the engagement, in order, are...
The arc follows the six weeks: meet the engagement, scope it, research it, analyze it, structure the story, build the deck, manage the client, protect the data, then audit and present. The deck never precedes the storyline, and the audit always precedes the room.
2 · The three keep-forever habits are...
The answer in sentence one forces every other sentence to earn its place; every AI fact stays a lead until a human sees its source; client-identifying material gets stripped before any prompt. Three habits, and every headline incident this course cited traces to skipping one of them.
3 · The paste box's coaching bands say...
The bands grade craft signals, not truth - the citation audit guards the facts. And the pattern holds on real text: drafts most often bleed points on consequences never drawn and clients never named, the two measures AI cannot supply for you.