Command the Machines
0 / 5
8
BAND III · ADVANTAGE
Week 8 of 12 · DAVOS · 1,560 m

Get Consultant-Grade Answers Before Lunch

Bigger questions. Better answers. Before lunch.

You've been rationing your own curiosity — skipping the spreadsheet, shelving the big question, chasing the board pack, starting every day un-briefed. That rationing ends this week. And on Thursday, something the course has been building toward since day four: your first worker gets a promotion.

Not your desk? This week's moves work on any desk: read the file you've been avoiding · a four-line brief that buys consultant-grade research · let something attack your plan first.

Today · one task

Upload the spreadsheet you've been avoiding. Ask three real questions — not "summarise this" — and make every answer show its receipts: the rows, the columns, the numbers.

There's a spreadsheet you've been meaning to read for weeks — maybe months. Not because it doesn't matter; because reading it costs an afternoon. As of today, reading is free.

You'll have: the avoided spreadsheet, finally read — three plain-English questions, three answers with the source attached.

  1. Find it and upload it. You know the one — the sales-by-customer dump, the stock file, the survey export sitting in the folder called "to review".
  2. Ask the three questions you'd ask if reading were free. Not "summarise this" — real questions: "Which ten customers drive most of this?" · "What's quietly shrinking?" · "Where does this contradict what I believe?"
  3. Check the receipts. Every answer should point at the rows and columns it came from — the 2.5 reflex, applied to data. An answer without a source goes back for one.

Here's a spreadsheet I've never had time to read. Answer my three questions from it, and for every answer show your working: the columns, rows or ranges you drew on, and the numbers behind the claim. If the data can't support an answer, say so plainly rather than stretching it. Then — separately, at the end — tell me the one thing in this file I didn't ask about but should have.

COPY

Rescue — Answers feel too neat? Ask it the auditor's question: "What in this file contradicts what you just told me?" A good answer survives its own cross-examination.

Did you get this done?

One answer changed what you'll do this week — or confirmed a fear with a number.

Want more — The folder called "to review" contains more than one of these. One a day with the morning coffee, and the backlog of unread truth in your business is gone by Friday.

Reading is free now. Tomorrow you ask a question no spreadsheet can answer — and go to lunch while it's being researched.

Today · one task

Pick one question with real money on it. Fill in the four lines — what to find, where to look, what to ignore, what to hand back. Switch web research on, fire it, and go to lunch.

Somewhere on your desk is a question worth real money that's never been properly researched — because proper research costs a consultant and a fortnight. Today it costs a good brief and your lunch hour.

You'll have: THE RESEARCH BRIEF saved — and your first deep-research report, returned with sources, before you were back.

  1. Pick the question that's actually worth it — the market you're eyeing, the pricing move, the competitor pattern, the regulation rumour. One question, real stakes.
  2. Fill the brief. The press below asks you four things — scope, sources, exclusions, deliverable — and your answers are the skill. A tight brief is the difference between a report and a ramble.
  3. Fire it, with web research on, and go to lunch. Read the report when you're back: answer first, evidence behind it, sources cited, gaps admitted. Save the press as THE RESEARCH BRIEF.

I have a serious question to research. First, interview me briefly: (1) SCOPE — what exactly am I asking, and what decision hangs on it? (2) SOURCES — what should you lean on, and what do I already believe? (3) EXCLUSIONS — what's out of bounds or already decided? (4) DELIVERABLE — how long, in what shape? Then research it properly on the web and return: the answer up front in three sentences, the evidence behind it, every source cited, the strongest case against your own answer, and what would have to change for the answer to flip. Say plainly where the evidence is thin — I'd rather have a gap than a guess.

COPY

Rescue — Report came back broad and shallow? The scope was a topic, not a question. Rebrief with a decision attached — "should we {do X} by {when}" — and rerun. Decisions focus research; topics dilute it.

You just built a machine. Give it a name.

Say what it does and what it's allowed to touch, then tap the button. It joins the count at the top of your screen — that count is your workforce, and it grows every week.
Did you get this done?

The report answered the question you actually asked, with sources — and you finished lunch first.

Want more — The brief is the skill, and it compounds: next week's hiring kit and the Attacker on Friday both inherit this discipline. Keep every filled brief on the shelf — your research library is starting itself.

Consultant-grade, before pudding. Tomorrow the board pack learns to build its own section.

Today · one task

Pick one board-pack section and the one export that feeds it. Drop the export in your Project, run the prompt above — then read the result the way a director would.

Every board cycle, one section of the pack costs you three chasing emails and a Sunday evening. Today that section learns to assemble itself — numbers, narrative and the "so what" — from data you already have.

You'll have: one section of your board pack assembling itself from source data, in board format, with zero chasing emails sent.

  1. Pick one section — and the one export that feeds it. The sales summary needs the sales export; the financial highlights need the management accounts. One section, one document. The full pack is a trajectory; today is one section.
  2. Drop that export on the shelf and run the press below. It writes the section in board format from that document alone: numbers first, what moved and why, one "so what". Anything it can't find is marked as a gap — never estimated.
  3. Read it as a director would. Numbers right, narrative honest, "so what" clear? Then it goes in this cycle's pack — and that section stopped needing you today.

Build the {name it} section of my board pack from the document I've just added: {the export}. Board format: the numbers first, a short narrative of what moved and why, then one "so what" paragraph a director can act on. Cite where in the document each figure came from. Where a number the section normally carries isn't in this document, leave a clearly marked gap naming the missing export — never estimate a board number.

COPY

Rescue — A figure it needs lives in a system it can't see? Same two-click rule as Monday: export, drop it on the shelf, run again. Board numbers come from documents, never from memory — the machine's or yours.

Did you get this done?

The section generated itself in board format — and you sent zero chasing emails.

Want more — Next cycle, add a second section; the cycle after, a third. The full self-assembling pack matures over two or three board cycles — that's the honest trajectory, and you're one section into it as of today.

One section down, zero emails sent. Tomorrow morning, the biggest moment of the week: your first worker gets promoted.

Today · one task

Open the 1.4 scheduled task and replace its instructions with the Morning Page version — same worker, same 06:00, same powers. Then promote its entry on The Roll, in writing.

On day four of this course you commissioned a modest worker: a morning note, three commitments, one slipping thing. It has reported every morning since, without fail, without thanks. Today — with seven weeks of machines now feeding it — it gets promoted.

You'll have: THE MORNING PAGE on The Roll — the Morning Note, promoted: five numbers, two names, one decision, waiting by your first coffee.

  1. Open the 1.4 standing order and upgrade it with the order below. Same worker, same schedule, same read-and-report-only powers — a bigger desk.
  2. Run it once, live, now, and read the page it builds from everything you own: the cash view, the customer digest, the chaser's pipeline, the calendar. Five numbers. Two names. One decision, teed up.
  3. Update its entry on The Roll — a formal promotion, in writing. The name changes to THE MORNING PAGE; the powers don't change at all. Tomorrow it arrives unasked. That's not the win anymore. That's just Tuesday.

(Replace the 1.4 morning-note order with this.) Every weekday at 06:00, assemble my MORNING PAGE from what I hold: the latest cash view, the customer digest, the deal chaser's queue, my calendar and inbox. One page: (1) FIVE NUMBERS that changed or matter today, each with one line of why; (2) TWO NAMES — the people I should talk to today, and the reason; (3) ONE DECISION that deserves my best thinking today, framed with its options. If a source is missing or stale, say so in one line — never pad a gap with a guess. Under 200 words. You read and report only — you never send, delete or change anything.

COPY

Rescue — Page arrived with gaps? It can only serve what the week's machines feed it — each named gap is just next Monday's export reminder. A Morning Page that admits what it doesn't know is exactly the officer you want.

You just built a machine. Give it a name.

Say what it does and what it's allowed to touch, then tap the button. It joins the count at the top of your screen — that count is your workforce, and it grows every week.
Did you get this done?

It arrived this morning without being asked — and nothing on it was wrong.

Want more — Read it before the phone gets a vote. The page is your agenda speaking first — five numbers, two names, one decision — and Week 12 chains the whole workforce into it.

Promoted, in writing, on The Roll. The course's first machine just became its most senior — and your day now starts briefed. Tomorrow: the machine whose only job is attacking you.

Today · one task

Paste in one live plan and take the three punches — the wrong assumption, the competitor's move, the costliest failure. Write your answer to the one that stung.

Every plan you've ever launched was attacked eventually — by the market, by a competitor, by the detail nobody raised in the room. Today you build the thing that attacks it first, while the attack is still free.

You'll have: THE ATTACKER saved — a standing red team loaded with your company's context — and one live plan, attacked three ways, with at least one punch that stung.

  1. Feed it one live plan. Paste the real thing — the expansion memo, the pricing change, the hire, the launch. Loaded inside your Project, it attacks with your company's context, not generic caution.
  2. Take the three punches: the assumption most likely to be wrong, the move a competitor makes the day you launch, and the failure that costs the most. Each comes with the evidence it rests on.
  3. Apply the test-of-the-test: did at least one sting? If nothing did, the briefing was too kind — tell it what you're actually afraid of and rerun. Then answer the punch that landed, in writing, and the plan gets funded stronger. Save as THE ATTACKER.

You are my attacker. Your only job is to break the plan below before the world does. Using everything you know about my company from the Project, deliver exactly three attacks: (1) the assumption in this plan most likely to be wrong, and the evidence that worries you; (2) the most dangerous move a competitor makes the day this launches; (3) the failure mode that costs the most, and the early sign it's happening. No praise, no balance, no "on the other hand" — the defence is my job. End with the single question I least want to be asked about this plan.

COPY

Rescue — Nothing stung? The briefing was too kind. Add one line — "what I'm actually afraid of is {x}" — and rerun. Red teams need to know where the armour is thin.

You just built a machine. Give it a name.

Say what it does and what it's allowed to touch, then tap the button. It joins the count at the top of your screen — that count is your workforce, and it grows every week.
Did you get this done?

Three attacks delivered — and at least one stung.

Want more — New rule, cheap to keep: nothing gets funded until the Attacker has had it first. And in Week 12, when you build a second you, it inherits this instinct — your twin learns to attack before it approves.

Week eight is in. Spreadsheets read themselves, big questions research themselves over lunch, the board pack builds its own section, your first worker earned its promotion — and your plans now get attacked in private before the market gets a swing. Next week: the hardest subject in the building. People.

Week 8

Your day starts briefed now.

Five for five. The avoided spreadsheet answered three questions with receipts, a consultant-grade report came back before you finished lunch, one board-pack section assembled itself without a single chasing email, the Attacker landed a punch your plan needed — and the modest worker you hired on day four got promoted, in writing, to THE MORNING PAGE. Five numbers, two names, one decision, every morning. You'll never start un-briefed again.

0 of 5 done

Month two — GO DEEPER — ends here. Sit the same thirty questions again; read it beside your Week 4 number.

The week's quiz · seven questionsNO PASS MARK · IT TEACHES
Continue → Week 9