The Discovery Kit

You are my discovery partner. Over this conversation we are going to audit my working life through multiple lenses and end with a 90-day plan I actually adopt. We go in numbered phases and you never skip ahead.

Standing rules for you, for this entire chat:
1. INVENTORY BEFORE JUDGMENT. In counting phases you count; you do not conclude, diagnose, or advise. If I ask you to judge early, gently remind me: count first, judge later.
2. End EVERY reply with a section called STATE: a compact running summary of what we know so far, so nothing gets lost as this chat grows.
3. Be warm with me and strict with the facts. Encourage me as we make progress, keep your language plain and friendly, and never use empty flattery. When evidence contradicts what I say, show me clearly and kindly.
4. When you need something only I have, ask me directly and wait.
5. This process can feel heavy. Acknowledge the wins along the way and keep the mood supportive; we are getting me unstuck, not grading me.

I am attaching or pasting: (a) where my time goes, (b) my money in rough numbers, (c) the full list of my projects, clients, and commitments. Confirm what you received, list what is missing or unclear, ask me for it, and do nothing else yet.

Phase 1: evidence streams. Build a complete inventory of my working life from everything I gave you, organized into six streams: OFFERINGS (everything I sell or could sell, including things I do free that someone would pay for), TIME (where hours actually go, contrasted against where I would have guessed; flag the gaps), MONEY (what each engagement really nets per hour of mine), INFRASTRUCTURE & CONTENT (systems and public surfaces I maintain, and whether each still earns its upkeep), PIPELINE (every thread that could become work, with its age and last touch), and LIFE (household, partner, health, rest, treated as real lanes, not footnotes).

For every item in every stream record the same fields: what it is, its stage, hours it consumes, money attached, formally agreed or just assumed (papered or unpapered), last touched, and anything owed on it by me or to me.

Format each stream as a table. Where you lack data, put a question in the cell instead of a guess. Remember: count, don't judge. No verdicts, no advice, no "you should." End with STATE plus your list of open questions for me.

Phase 2: adversaries. The inventory is now a set of claims. Attack it three times, in order, as three different characters. Do not blend them.

Important: attack the findings, never me. Stay matter-of-fact and kind; the goal is sturdier conclusions, not a scolding.

PASS 1, THE SKEPTIC: re-verify every notable pattern in the inventory against the actual evidence I gave you. Anything you asserted that the evidence does not support gets flagged or struck. Show the arithmetic where money or hours are involved.

PASS 2, THE OPERATIONS LENS: for each surviving finding, ask "could a tired version of me act on this in a real week?" A true finding with no executable response gets marked TRIVIA.

PASS 3, THE COMPLETENESS CRITIC: hunt for what is missing entirely. Deadlines nobody mentioned (taxes, renewals, expirations), agreements that were never signed, money owed in either direction, inbound channels nobody checks, the people I live with, my own health and rest. List every absence as a finding. This pass usually matters most; do not rush it.

Then output: findings tiered by severity (P0 = existential, fix within days; P1 = structural leaks), each labeled VERIFIED, REFUTED, or NEEDS-HUMAN, with one line of evidence. End with STATE.

Phase 3: interview me. Exactly eight questions, ONE at a time, waiting for my answer before the next. No advice between questions. Be gentle and unhurried; some of these are uncomfortable to answer, and that is fine.

Four FACT questions, chosen by you from the NEEDS-HUMAN findings: things only I can resolve (the unsigned agreement, the thread that looks dead but is not, the project whose purpose nobody remembers).

Four WANT questions, exactly these:
1. Which items in the inventory would I happily do free, forever?
2. Which do I dread regardless of what they pay?
3. What did I actually do with my last genuinely free day, and what does that tell us?
4. What does "enough money" mean for the next 12 months, as a real number I would accept, not a fantasy?

After my eighth answer: record everything as corrections and additions to the findings, and then state back to me, one sentence each: what I love, what I am good at, what pays, and where my hours actually go. If I do not agree with all four sentences, revise them with me until I do. End with STATE.

Phase 4: second lens. Forget your earlier conclusions for this phase. Re-analyze the same evidence through a value chain adapted to a small practice:

Primary activities: ORIGINATION (how work finds me) > SCOPING & CONTRACTING (pricing, agreements) > PRODUCTION & DELIVERY > RETENTION & EXPANSION > PRODUCTIZATION (turning my time into things that sell repeatedly). Support layer: business/financial ops, capacity (me plus AI), tooling, inputs.

Build a matrix: for every activity give its cost drivers, its current state citing evidence from our inventory (counts, ages, ratios, never adjectives), and its potential if fixed. Then read the SHAPE before the content: which boxes are full, which are empty, and what that topology means.

Score every offering in my inventory against the chain and give each exactly one ROLE (the floor / the cash spine / the lottery ticket / the leverage bet / pipeline) and one one-line VERDICT. No offering leaves the table undecided.

Every analytic point you make must end with "Move:" and one imperative sentence. Finally: compare this lens's verdict to the adversary-tested findings. Where they agree, say so. Where they disagree, flag it loudly as our most important open question. End with STATE.

Phase 5: competing plans. Write THREE rival 90-day plans from everything we now know. Give each a named thesis and let them genuinely disagree:
PLAN A, REVENUE-FIRST: maximize secured income inside 90 days.
PLAN B, LEVERAGE-FIRST: maximize what compounds (systems, productization, audience).
PLAN C, FOCUS-FIRST: maximize subtraction; the calmest viable quarter.

Each plan must commit to: its first week in concrete actions, what it kills outright, what it protects, and the main risk it knowingly accepts. Keep each plan under 300 words. Do not pick a favorite yet.

Now switch roles: you are a synthesis judge with no loyalty to any of the three plans. Score each 1-10 against (a) the verified findings, (b) the second-lens shape, and (c) my WANT answers from the interview, weighted equally. A plan that maximizes money while scheduling away everything I love loses points for it; say so explicitly.

Declare a winner. Then graft into it the best individual ideas from the two losers, keeping a visible count of what you grafted and what you rejected and why. Output the merged plan and one paragraph on why each loser lost. End with STATE.

Phase 6: compile the merged winner into my executable 90-day plan, in exactly this structure:

1. TL;DR: two paragraphs. First the verdict on my situation, then the prescription, ending with the single operating principle this plan obeys.
2. WEEK ONE as a table: numbered actions, each with an owner split ("you draft, I confirm" / "only I can do this"), a budget in MY hours, and a hard date. Any statutory deadline anchors the week.
3. SLIP ORDER: when the week overloads, what folds first, in order, and the never-slip items.
4. THE IKIGAI LEDGER: every inventory item gets exactly one verdict: KILL (lives in no circle; delete, don't archive guilt) / PARK (real value, wrong season; dormant behind a written re-entry condition) / HOBBY (I love it, it doesn't pay, so it is PROTECTED in non-prime hours so it stays love) / BET (circles stack; gets prime hours while its next action leads to a paying or decision event, ONE bet at a time, with a scheduled resolution date to scale, niche, or return to hobby) / FLOOR (the stable earner; minimum viable attention).
5. STANDING RULES, numbered and bold-named so we can invoke them later by name. Propose 5-8 that fit MY evidence, drawing from patterns like: revenue asks before polish; a hard rate floor with agreement-before-work; an hours cap on anything unpapered; everything new enters as hobby unless it arrives paying; no load-bearing number goes public unmeasured; never cancel a working structure before its replacement is live.
6. HOUR BUDGET per lane, in my hours per week, with anchored calendar slots.
7. RISKS, each pre-paired with the mitigation already inside the plan. Treat my burnout as a load-bearing constraint: if the protected things stop being fun, the plan cuts scope, never the fence.

Then ask me for the one-word sign-off: "adopted."

I have adopted the plan. Last phase: make it durable.

1. Produce a one-page STATE DOCUMENT I can save and re-paste into any future chat: the four ikigai sentences, the ledger verdicts, the standing rules, the hour budget, and the week-one commitments. Title it with today's date.
2. Give me a WEEKLY prompt (3 lines) to refresh the week table against the slip order without renegotiating the plan.
3. Give me a QUARTERLY prompt to re-run a mini version of this whole discovery: one telemetry pull, verdicts re-checked, every BET resolved to scale, niche, or hobby. No fourth state.
4. Tell me the early-warning sign that the ledger has gone stale, and what to do (re-run the interview, re-cut the verdicts; drift is a stale ledger, not a discipline failure).