The Discovery Path

A multi-stream audit of your working life that ends in a 90-day plan aligned to your ikigai.

Picture the Venn diagram of your life: things you love, things you are good at, things that pay. Now picture the fourth circle, the honest one: things you actually do. For most working people that fourth circle has drifted away from the other three. Quietly, one calendar entry at a time.

This path measures the drift with evidence, then closes it with a plan. Your AI runs the streams. You make the calls. The output is not a website or a rebrand or a productivity system. It is a dated, adopted 90-day plan where everything you do has an honest verdict about which circles it lives in.

I ran this on my own practice. The shapes below are real, with the specifics stripped out so you can pour your own life in.

00How to use this

1. Pick your entry below; skip what does not apply.
2. Every stage has You do (judgment, minutes), Your AI does (the heavy lifting), an Artifact, and a Gate. A stage without its artifact did not happen. A gate without your sign-off does not open.
3. One chat window is enough. You can run the steps one at a time: each stream as its own turn, each critic as a fresh prompt.
4. If your AI has persistent memory, everything compounds. If not, keep one running state document and re-paste it each session, or build the memory layer first (an afternoon; guide linked above).

01Pick your path

"I'm busy constantly and can't say where it goes." Run everything in order. The time stream will hurt the most and help the most.

"The work is fine but the money is wrong." All streams, but lean on the value-chain lens. Your answer probably lives at the front of the chain.

"I'm burned out." Say so up front. Rest, health, and the people you live with become real audit lanes, not footnotes.

"I'm pivoting." The kill/park/hobby verdicts are your whole game. Be ruthless in the interview.

Solo or team? Solo: run it as written. Team: one interview round per principal, rules get named owners, and verdicts need everyone whose hours move in the room.

Have time data? Yes: feed it in raw. No: start a seven-day log today and run the early stages on what you already have. Remembered time lies. Measured time does not.

02Stage 1: The evidence streams

The audit looks at you, not your tools. Several streams, gathered before any judging, because conclusions formed before evidence are just moods with formatting.

You do (an hour): hand over the raw material. Time data, accounts in rough numbers, every project and commitment, the folder of half-finished things you avoid. Do not curate. Curation is the drift defending itself.

Your AI does: one stream per area, same fields per item (what it is, pipeline stage, hours it eats, money attached, papered or unpapered, last touched, anything owed). The streams: offerings, time (where hours actually went vs where you think they go; the gap is a finding), money (true net per hour), infrastructure and content, pipeline, and the life lane (household, partner, health, rest; a plan that ignores them gets vetoed by reality inside a month).

03Stage 2: The adversaries

Findings are just claims until something hostile has tried to kill them. Three roles, run separately:

1. The skeptic re-checks every finding against the evidence. "You said this client is unprofitable. Show the arithmetic."
2. The operations lens asks whether a tired human could act on each finding. A true finding you can't act on is trivia.
3. The completeness critic hunts for what nobody mentioned: the tax deadline, the unsigned agreement, the partner with opinions, your sleep. In my run this role found the most and changed the top recommendation.

04Stage 3: The interview

The streams hold the facts. You hold the meaning. Your AI asks about eight questions, covering two things the evidence cannot:

Facts only you hold. The contract you signed but never filed. The dead-looking thread with a warm relationship behind it. The project you can't remember the point of (that's an archive signal; let it go).

Wants only you hold. Here is where what you love enters as data. Which items would you do free forever? Which do you dread no matter the pay? What did you do with your last free Saturday? If one project got your best hours for a quarter, which one makes you sit up? What does "enough money" mean as a number you would actually accept?

Answer plainly. This is the one stage where trying to impress hurts you.

05Stage 4: The second lens

One analysis can be wrong with full confidence. So run a second pass through a different frame and make it reach the same verdict. Where the two agree, trust them. Where they disagree, that disagreement is your most important finding.

The second frame is a value chain: the path work takes from showing up to getting paid, mapped to a one-person practice. Origination (how work finds you) → Scoping and Contracting → Production and Delivery → Retention and Expansion → Productization, with ops, capacity, tooling, and inputs beneath. Fill every box with evidence (counts and ages, never adjectives), then read the shape first. The most common one: full in the middle, empty at both ends. Great production, starved origination, leaky paperwork. If that is you, your highest-value hours are not more production. They are the unglamorous front of the chain.

Make every point end in a one-sentence "Move:". No analysis gets to end without a next action.

06Stage 5: Competing plans

Do not let your AI write "the plan." Have it write three rivals with named theses: revenue-first, leverage-first, focus-first. Each one commits to a 90-day horizon, a first week, and what it kills. Then a judge, a fresh prompt with no loyalty to any of them, scores all three against the findings and what you said you want, picks a winner, and grafts in the best of the losers. It keeps a count, so you can see exactly what got merged.

A plan that maximizes money while scheduling away everything you love is not a plan. It is a resignation letter with extra steps.

07Stage 6: The plan, compiled to executable

The difference between a plan and an essay is the compile. Required parts: a two-paragraph TL;DR (the verdict, then the prescription, then the one rule the plan runs on); week one as a table with owner splits ("AI drafts, you confirm" / "only you can do this"), time budgets in your hours, and hard dates; a slip order decided while you are calm (what folds first when the week overloads, and what never slips); hour budgets per lane in your hours; and risks, each paired with the fix already built into the plan. Treat burnout as a hard constraint, not a footnote.

And the heart of it, the ikigai ledger: every item gets exactly one verdict, and the verdict says which circle it lives in.

Plus standing rules, numbered and named so you can call on them in future sessions: revenue asks before polish; a hard rate floor with the agreement signed before work starts; a tightening cap on hours for anything unpapered; the intake fence (everything new enters as a hobby unless it arrives with a paying customer); a published-numbers gate (no load-bearing number goes public until you have measured it); and no gaps left uncovered.

08What this is not

This path produces the plan. It does not execute it. Execution comes later, and doing it inside the discovery corrupts the discovery: you start auditing toward the answer you are excited to build. Finish the verdict first, then build with the plan as referee. The memory layer that makes it all compound is the companion guide above. And the best execution habit is one this path already taught you: before any real plan runs, let hostile critics tear into it.

09The re-entry rhythm

Weekly: refresh the week table and apply the slip order without renegotiating the plan. Quarterly: a mini-audit. One time pull, verdicts re-checked, bets resolved, adjust once. On drift: a month of off-plan hours is not a discipline failure. It is a stale ledger. Re-run the interview, re-cut the verdicts.

10Failure modes to preempt

1. Verdicts before evidence. Judging during inventory ruins the count.
2. Skipping the completeness critic. The deadline nobody mentioned blows up week one.
3. The flattering interview. Impressive answers produce a plan for the person you wish you were, and the person you are will not run it.
4. The unledgered number. An impressive figure shipped without checking and turned out wrong by a factor of six. Measure it or cut it.
5. The unadopted plan. Written, admired, never tracked. A rejection wearing a compliment.
6. The starved hobby fence. Monetize everything you love and soon nothing is loved. Keep one circle pure.
7. The quietly dead background run. Ask a long AI job to prove it is still alive. Do not trust a status that just says it's fine.

11Trigger phrases

• "count, don't judge" sends a stream back to inventory mode.
• "run the critics" calls in the three adversaries.
• "circle check" asks of any item: which circles is it in, and does the verdict still match?
• "hobby-fence it" assigns the protective verdict on the spot.
• "adopted" closes the plan, with a tracker entry to prove it.

12The shape of the whole thing

Gather the streams without judging. Let the adversaries attack the findings. Tell the interview the truth, including what you love. Make a second lens reach the same verdict. Force rival plans to fight, adopt the winner the same day, and give every piece of your working life its honest circle. Then come back, smaller, whenever the drift returns.

The fourth circle never stops drifting. The point is to own the instrument that measures the drift, and the rhythm that pulls it back, with your AI holding the streams together at the center.

Adapt freely. Credit if you publish a fork.