admin/BusinessNotes
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity
Files
60176d2fdc74d7730b10625610464cd6b42e375f
BusinessNotes/log.md
EugeneTes 60176d2fdc all
2026-07-30 11:15:52 +02:00

77 KiB
Raw Blame History

Log

Append-only chronological record of wiki operations. Newest entries at the bottom.

2026-07-17 — init

Operation: Initialize vault storage.

Changes:

  • Created folder tree: raw/sources, raw/assets, wiki/sources, wiki/entities, wiki/concepts, wiki/timelines, wiki/comparisons, wiki/queries, wiki/lint-reports.
  • Created wiki/overview.md (#overview).
  • Created index.md catalog.
  • Created log.md (this file).

State: Empty wiki, no sources ingested. Awaiting first source in raw/sources/.

Open questions: What is the first source to ingest? Which entities/concepts anchor this domain?

2026-07-17 — ingest

Operation: Ingest everything from raw/sources/ (2 files, full folder). First content ingest.

Sources ingested:

Changes:

Findings:

  • The two sources independently converge on one thesis: AI substitutes undifferentiated execution and raises the value of judgment → sell productized outcomes with accountability, not hours. 7 of 10 concepts carry evidence from both, which is the vault's main confidence signal.
  • Both sources argue the same side. The vault has no adversarial material on productization; agreement here is unreplicated advocacy, not confirmation.
  • Every quantitative claim across both sources is uncited (90% below-minimum-wage, 80%/5th-touch, 91% vs 11% referrals, 47% subject line, $1:$6 AI spend, 30–50% margin, 15–25% webinar). Flagged on each page and in overview; not to be repeated as fact.
  • The RU source is a second-order distillation — its 5–7 upstream notes are outside this vault under PARA/4. Archive/15-06-2026 Inbox/, so no claim from it can be traced to origin.
  • Recorded 3 live contradictions rather than resolving them: calls vs. chat (sales-discipline); "sound unique" vs. "buy a proven method" (methodology-as-moat); cloning vs. moat (cloning-over-originality).
  • Marked Status: tentative on the 3 single-source pages: offer-ladder, cloning-over-originality, solution-vs-staff-augmentation.

Open questions:

  • Import the 5–7 upstream notes from PARA/4. Archive/15-06-2026 Inbox/ into raw/sources/? Highest-value next ingest — would let the RU claims be traced.
  • Find an adversarial source: where does the productized model fail?
  • Which market/currency do the $1K floor and $200/mo substitute assume, and do they transfer to the vault owner's market?
  • When to productize: offer-first or after 2–3 identical projects? Unresolved across sources.
  • Should Pabrai/Buffett/Munger/Graham get entity pages, or is the material too thin?

2026-07-17 — ingest (batch 2)

Operation: Pulled the eight PARA/4. Archive/15-06-2026 Inbox/ notes into raw/sources/ and ingested them alongside the Sebastian interview (already in the source folder). 9 new sources.

Raw changes: copied 8 files from D:\Projects\Notes\Vault\PARA\4. Archive\15-06-2026 Inbox into raw/sources/ (originals left intact; kept source filenames for provenance). No existing raw file modified.

Sources ingested (9): Sebastian interview → 2026-07-06-sebastian-interview-ai-and-software-engineering; and the June-15 batch → 2026-06-15-rodenko-selling-development-expensively, 2026-06-15-17-ways-first-client, 2026-06-15-how-to-get-rich-cloning, 2026-06-15-more-clients-dev-agency, 2026-06-15-making-money-with-ai-2026, 2026-06-15-konspekt-aphorisms, 2026-06-15-linkedin-mail-template, 2026-06-15-meta-analysis-selling-dev-in-ai-era.

Wiki changes:

Findings / what changed in understanding:

  • Provenance resolved. The June-15 distillation was second-order; its 7 upstream notes are now ingested, so statistics and claims trace to named speakers (dmitry-rodenko, ab-analytics, Tony, mohnish-pabrai) — but the roots are promotional YouTube videos, so figures are attributable, not verified.
  • New domain branch + new contradiction. The Sebastian interview added engineering-work concepts and the vault's sharpest live conflict: online outreach ("Big zero") vs. multi-channel outbound. Recorded, not resolved; best current read is audience-dependent (enterprise vs SMB/startup).
  • Second moat. Relationships-as-moat now competes with methodology-as-moat.
  • eugene is plausibly the vault owner (his insin project + custom Claude Code harness match the owner's tooling) — recorded as tentative inference.

Open questions:

  • Still no adversarial/failure-case source on productization — now the top gap.
  • Which moat (method vs relationships) and which channels apply to eugene's actual buyers? Candidate first durable query.
  • Corpus-flagged gaps unfilled: unit economics, legal/SLA packaging, CIS/EU niches, own case studies, objection scripts.
  • A first lint is now worthwhile (34 pages, multiple tentative/contradiction flags).

2026-07-17 — query

Operation: Answered durable query — "What would be the best method to find the first client?" (Workflow B). First page in wiki/queries/.

Question: What would be the best method to find the first client?

Answer (synthesis): No single magic channel; the corpus converges on a sequence — (1) pick one narrow niche (precondition), (2) for a literal first client favor a warm/repeat-contact in-person channel (Chamber of Commerce / local groups; 3×=regular, 6×=trusted) over cold online, (3) work it with discipline (3 channels × 90 days, 5+ touches Day 0→3→7→14→30, founder sells), (4) convert client #1 into referrals immediately. In-person is the low-variance bet because it is the one thing ab-analytics and sebastian independently endorse; cold online only if buyers are SMB/startup. Tailored note for eugene (likely owner): reallocate LinkedIn/articles effort to recurring in-person events.

Changes:

  • Created 2026-07-17-best-method-first-client (#query).
  • Added inbound links from client-acquisition-channels and eugene (both had this exact question open in "Next Questions"; converted to "partially addressed → see query").
  • Updated index: new Queries section + page count 34→35 + status line.
  • No source or concept claim changed — the answer is a synthesis of existing pages, not new evidence.

Findings / caveats carried into the answer:

  • The online-vs-in-person split remains the governing uncertainty; the answer resolves it by audience (enterprise → in-person; SMB/startup → online viable), not by a universal law.
  • All cited statistics remain attributable-but-unverified (promotional videos); flagged in the query page.
  • Referrals are the highest-converting channel but cannot bootstrap client #1 — noted as a sequencing constraint.

Open questions:

  • eugene's actual niche/buyer type is still unchosen — it determines the online-vs-in-person call. Blocks a truly concrete plan.
  • Is there a warm-intro path (existing content audience, embedded/CV contacts) faster than cold-in-person for the first engagement?
  • Still no adversarial/failure-case source; a first lint remains the other worthwhile next operation.

2026-07-18 — ingest

Operation: Ingest single source — "Information Is Free, Implementation Is Paid" (Workflow A). 12th source; first on demand generation / content marketing as a discipline.

Source ingested: raw/sources/Information Is Free, Implementation Is Paid.md → 2026-07-18-information-is-free-implementation-is-paid (undated YouTube video, 08:45; ingestion date used for the filename prefix). No raw file modified.

Wiki changes:

  • Created 1 source summary in wiki/sources/.
  • Created 1 concept: information-vs-implementation (thesis + 5×10×4 content factory + scramble trick + pain→solution structure). Single-source, marked Status: tentative.
  • Enriched 4 concepts: client-acquisition-channels (organic-content channel developed; added as a fourth, most-bullish position on online — "content is the engine" — sharpening the online-vs-in-person split), pain-discovery (the nuanced-observable-pain AI prompt added as a content-side pain-articulation tool), methodology-as-moat ("sequencing is the paid good" as the marketing-side moat framing), pricing-from-value (free content as an upstream price anchor).
  • Updated overview: source count 11→12; added the content-marketing engine to the Branch-A channel step; new highest-signal bullet ("information is free / scramble"); widened the online-vs-in-person contradiction; noted the source reinforces (not tests) the "sell dear" school.
  • Updated index: +1 source row, +1 concept row, page count 35→37, status line, "eleven of twelve" gap note.

Findings / what changed in understanding:

  • New domain surface: demand generation. Prior sources listed "content marketing" only as one of 17 channels; this is the first to treat it as a system — give away every step, withhold the sequence, and let the giveaway anchor the price.
  • The online-vs-in-person split is now the vault's widest live contradiction, spanning "content is the entire lead engine" (information-vs-implementation) to "content = Big zero" (sebastian) — still best read as audience-dependent, still untested.
  • Reinforcement, not confirmation. Same US "sell dear / give-value" coaching school; anonymous speaker, zero results data. Does not close the standing adversarial-source gap.
  • New moat wrinkle: "sequencing is the moat" agrees with methodology-as-moat but strains cloning-over-originality — a publicly-shown (if scrambled) sequence is more re-derivable than a private method.

Open questions:

  • Does the scramble strategy actually convert (view→lead→sale), or do sophisticated viewers re-sequence and self-serve? No evidence in-source.
  • For eugene: is a content engine faster than the in-person channels favored in 2026-07-17-best-method-first-client, or only viable at audience scale? Could content be the trust layer under outbound (Tony's model) rather than a standalone source?
  • ICP-band mismatch: the source anchors to $500K–$2M-revenue buyers, larger than the vault's usual SMB framing — which buyers does the engine actually fit?
  • Adversarial/failure-case source and a first lint remain the two highest-value next operations (12 sources, multiple tentative/contradiction flags).

2026-07-18 — lint (sources)

Operation: First lint (Workflow C), scoped to the 12 wiki/sources/* pages per user request ("run a lint on sources"). Report → 2026-07-18-lint.

Checks run: raw↔summary coverage · page-type tags · template-section completeness · outbound-link validity · inbound linking / orphans · cross-source contradiction coverage · stale/superseded claims · metadata completeness · evidence provenance. Inbound/outbound link health verified against a full-vault [[…]] link inventory.

Verdict: source layer healthy. Complete 12-raw→12-summary coverage, no orphans, no broken links, all #source tags correct and first, all template sections present, provenance chain intact (distillation demotion + struck-through resolved question preserved as history).

Findings (5; all Low/Info):

  • F1 (fixed). Content-marketing-timing contradiction was recorded two-way (Tony vs AB Analytics) but is now three-way — the 07-18 source makes content the primary engine. Upgraded the Contradictions bullet in client-acquisition-channels (rule #5).
  • F2 (fixed). 2026-06-15-linkedin-mail-template's Change Impact claimed a link from sales-discipline that did not exist; template was inbound-linked from only one content page. Added the reciprocal link from sales-discipline — claim now true, weak link strengthened.
  • F3 (flagged). 2026-06-15-meta-analysis-selling-dev-in-ai-era is the most weakly inbound-linked source (only overview among content pages). Acceptable for a second-order map; optional future strengthening.
  • F4 (flagged). Filename date-prefix convention is inconsistent for undated videos (ingestion date vs batch date vs file date). All documented in-page; suggest codifying "undated → ingestion date." No renames (link churn).
  • F5 (flagged, standing). Corpus is 11/12 one school; all stats attributable-but-unverified. Adversarial source remains the top-value ingest.

Changes: edited client-acquisition-channels (F1) and sales-discipline (F2); created 2026-07-18-lint; updated index (new Lint Reports row, page count 37→38, status line). No source-page content changed — findings were about link/contradiction coverage, not claims.

Open questions / next:

  • Full-wiki lint is the natural follow-up (this pass covered sources only) — concept/entity pages, plus high-mention concepts lacking pages (e.g. referrals).
  • Adversarial/failure-case source (F5) still the highest-value next ingest.
  • Optional tidy: F3, F4.

2026-07-18 — ingest (first adversarial source)

Operation: Ran deep-research (Workflow tool, user-requested) to locate and verify empirical AI-productivity evidence, then ingested it (Workflow A) as the vault's first adversarial source. Research run: 106 agents, 6 angles, 23 sources fetched, 104 claims → 25 adversarially verified (23 confirmed, 2 refuted). Task ww05pcpxq.

Raw created: raw/sources/AI Coding Productivity - Empirical Evidence (deep research 2026-07-18).md — a new immutable synthesis file (no existing raw modified). It is second-order but built from peer-reviewed papers / RCTs / large surveys, so higher-provenance than any prior source.

Source ingested: → 2026-07-18-ai-productivity-adversarial-evidence.

Wiki changes:

  • Created source page + 1 concept: ai-productivity-evidence (the durable home for the empirical counter-narrative). Both single-source → Status: tentative.
  • Added explicitly-marked counter-evidence to ai-market-shift (substitute→complement), seniority-and-ai (novices gain most; now Status: contested), future-of-engineering-work ("coding cost → 0" qualified; DORA team-stability). Prior claims preserved, not overwritten (Update Policy); contradictions recorded explicitly (rule #5).
  • overview: source count 12→13; added the foundational "AI capability — hype vs. measured evidence" contradiction; flipped the standing "no adversarial source" thread; added Branch-B caveat; new highest-signal item (perception ≠ measurement); navigation + concept list.
  • index: +1 source row (flagged adversarial), +1 concept row, page count 38→40, status line, Known-Gaps rewrite, quantitative-claims exception noted.
  • No entity pages for METR / Brynjolfsson / DORA / Stack Overflow — cited as evidence, not tracked subjects (consistent with the Buffett/Munger call on mohnish-pabrai).

Findings / what changed in understanding:

  • The premise beneath both branches is now contested, not settled. Strongest developer-specific RCT (METR) found experts 19% slower with early-2025 AI; best skill-distribution study (Brynjolfsson, QJE 2025) found novices gain most (+34%), inverting "seniors up / juniors out." Perception-vs-reality gap (~39 pts) shows self-reported gains are unreliable.
  • Honest framing: qualifies, doesn't demolish. Average effects are positive (14–15%); METR n=16 and "historical"; the strongest skill result is customer-support, not devs; all figures early-2025-scoped. Held to the same standard as the promotional sources.
  • It confirms a doubt the vault already held — ai-market-shift's "$200/mo substitute is asserted, not demonstrated" open question — rather than only contradicting.
  • The gap narrowed but didn't vanish: this tests the AI-capability premise, not the productization/sell-dear tactics. A failure-case source for the selling model itself is now the top gap; a pro-thesis RCT-grade source would balance further.
  • Verification caught a real error now avoided vault-wide: the widely-repeated "66% spend more time fixing AI code" conflates two SO-survey items; the correct debugging-time figure is 45%.

Open questions:

  • Does METR's −19% persist/reverse on mid-2026 agentic tools? (The source is explicitly historical.)
  • Does "novices gain most" hold for real, complex codebases, or invert where juniors can't catch AI errors?
  • Should dmitry-rodenko's "$200/mo substitute / hourly dev is dead" now carry Status: contested?
  • Next ingest candidates: (a) productization failure-case source; (b) pro-thesis rigorous source. Plus the still-pending full-wiki lint.

2026-07-18 — lint (full wiki)

Operation: Full-wiki lint (Workflow C) over all 40 pages — the follow-up to the sources-only lint. Every concept/entity/comparison/query page read in full; link health checked against a deduped inventory of all [[…]] targets. Report → 2026-07-18-lint-full-wiki.

Verdict: structurally healthy. Tags 100% (all 40 lead with the correct folder-matching type tag); templates complete (all 17 concepts + 6 entities carry the six required sections; query matches its template); no broken links (every target resolves; two [[…]] hits are code-span examples, one path-style [[wiki/overview]] normalized); no orphans (every page ≥7 inbound; new ai-productivity-evidence already at 26); contradiction coverage thorough (no unrecorded tension found).

Findings (6):

Note (not a defect): today's adversarial ingest was correctly surgical — it added counter-evidence only to the three pages it truly contradicts and left relationships-as-moat / product-ownership alone, because "novices gain most" corroborates their "AI levels skill" premise rather than contradicting it. The vault records that contradiction but not the corroboration (optional future touch).

Changes: edited productized-service, 2026-07-17-best-method-first-client, index (link + Lint Reports table + counts 40→41); created 2026-07-18-lint-full-wiki. No source or concept claim altered beyond the staleness corrections.

Open questions / next:

  • Create the referrals concept page (F4) — top structural gap.
  • Productization-failure-case source remains the top content gap; a pro-thesis rigorous AI source would balance further.
  • Optional: F5, F6, and the corroboration note.

2026-07-18 — create concept (referrals)

Operation: Created the referrals concept page, resolving finding F4 from the full-wiki lint (2026-07-18-lint-full-wiki). A synthesis/refactor of material already in the vault (no new source ingested).

Why: referrals appeared as a distinct, high-value channel with its own statistics and mechanics across 2026-06-15-17-ways-first-client, sales-discipline, client-acquisition-channels, relationships-as-moat, and 2026-07-17-best-method-first-client, but had no dedicated page — the clearest structural gap in the lint.

Wiki changes:

  • Created referrals (#concept #sales #outbound), full template. Key content: the 91%-would / 11%-asked gap; ask-immediately / be-specific / frictionless discipline; the two engines (transactional ask vs. relational emergence); the cold-start constraint (referrals can't bootstrap client #1); referral partners as the one referral route not subject to cold-start.
  • Wired inbound links (no orphan): client-acquisition-channels (taxonomy row + Related), sales-discipline (the ask bullet + Related), relationships-as-moat (the trust→referral chain + Related), 2026-07-17-best-method-first-client (step 4).
  • overview: added referrals to the selling-chain step 8 (flywheel/cold-start note) and the concept navigation list.
  • index: +1 concept row, page count 41→42, concept count 17→18, status line; marked the lint's F4 gap ✅ resolved.

Findings / notes:

  • Surfaced a mild method tension now recorded on the page: AB Analytics prescribes an explicit post-delivery ask; sebastian's model has referrals emerge from repeated in-person trust. Treated as complementary (ask captures goodwill the relationship created), unreconciled by any source.
  • The 91%/11% figures remain promotional/uncited (AB Analytics) — Status: tentative on the numbers, directionally corroborated by the relationship sources.

Open questions / next:

  • Productization-failure-case source is now the top remaining gap (F5/F6 from the lint are optional).
  • Candidate durable query: eugene's concrete 90-day plan (the existing first-client query stops at method).

2026-07-19 — ingest (Your Company Can't Outgrow Your Team)

Operation: Ingested raw/sources/Your Company Can't Outgrow Your Team.md (US leadership video, 4:35, undated, speaker unnamed) — the only raw source not yet in the wiki (14/14 now ingested).

Wiki changes:

  • Created 2026-07-19-your-company-cant-outgrow-your-team (#source #leadership #team), full template.
  • Created team-growth-ceiling (#concept #leadership #team, Status: tentative — single anecdotal source). Carries: growth-math thesis (company grows only as fast as its people), Business Athlete framework (2 of 7 practices captured), public scoreboard / private criticism, Netflix Keeper Test, values hire→inspire→fire ("values are what you tolerate", "complexity fails, simple scales").
  • overview: branches restructured two → three — added Branch C: running the team (explicitly thin/tentative); source count 13→14; navigation updated.
  • future-of-engineering-work: Related-Pages link to team-growth-ceiling (team collapse to 2–3 makes per-person growth a larger share of the ceiling).
  • index: +1 source row (anchors table), new "Running the team" concept group, counts 42→44, status line.

Findings / notes:

  • This is the vault's first internal-management source — it doesn't touch selling (Branch A) or the AI labor shift (Branch B), so it opens a branch rather than reinforcing one. No contradictions with existing pages; recorded a tension: the Keeper Test presumes replaceability, which sits awkwardly with 2–3-person AI-era teams where each member is a single point of failure.
  • Evidence quality is low (anecdotal talk, no data, speaker unnamed, framework only partially captured) — everything marked tentative; quotes attributed to the clip only.
  • No new entities: the speaker is unnamed; Netflix is only the origin of the Keeper Test, not an actor here.

Open questions / next:

  • Identify the speaker; recover the other 5 Business Athlete practices and the three values.
  • Branch C is dormant for a solo operator — if eugene has no team yet, it matters only at first hire; note for the eventual 90-day-plan query.
  • Standing gaps unchanged: productization-failure-case source (top), pro-thesis rigorous AI source.

2026-07-20 — ingest (Referrals Will Sink Your Business)

Operation: Ingested raw/sources/Referrals Will Sink Your Business.md (US coaching video, 6:47, undated; dan-martell coaching a founder stalled at ~$1.5M). 15/15 raw sources now ingested.

Wiki changes:

  • Created 2026-07-20-referrals-will-sink-your-business (#source #marketing #content-marketing), full template.
  • Created marketing-system (#concept #marketing #content-marketing, Status: tentative) — money-in-at-top→money-out-at-bottom definition; the three levers (content / paid / partnerships), pick one × 90 days; "the new paid is organic" (organic is the prerequisite for paid, not an alternative); reps-not-views; the ~6-month lag.
  • Created technical-founder-trap (#concept #positioning #marketing, Status: tentative) — solving ≠ explaining; explanation as the unlock.
  • Created dan-martell (#entity #person).
  • referrals: added a Counter-position subsection (dependency as a ceiling; referral flow has no throttle; "warning sign, not a badge") + Evidence/Related/Contradictions/Next-Questions revisions. Struck the page's standing "no adversarial source" bullet as partially resolved.
  • client-acquisition-channels: added the three-lever taxonomy with its scope analysis, a 5th row to the content-position table, the pick-1-vs-pick-3 conflict, Evidence + Related + 2 Contradictions.
  • sales-discipline: added reps-not-views as independent corroboration of consistency-over-intensity + the six-month patience budget.
  • information-vs-implementation: Summary re-scoped (mechanism still single-source; the bet now has two voices), "what the second source adds" (publishing as practice; organic→paid bridge; patience budget) and where the two differ (scramble trick is unique to source 1).
  • eugene: recorded the rival diagnosis of his networking blocker (articulation vs. relationships) + 2 Related links.
  • 2026-07-19-your-company-cant-outgrow-your-team: recorded the tentative speaker-attribution hypothesis.
  • overview: source count 14→15; selling chain extended to a 9th step (marketing-system); 2 new highest-signal items; new Open Thread (first channel counter-position, with its evidence caveat); 3 new/revised Contradictions; navigation + single-source list updated.
  • index: +1 source row, +1 entity row, +2 concept rows, counts 44→48, status line, Known Gaps.

Findings / notes:

  • First source in the vault to attack a tactic rather than a premise. The prior adversarial source (ai-productivity-evidence) contested the AI-capability premise; this one contests a channel every other source rated highest. Precision matters: it does not dispute that referrals convert best — it disputes dependency, on the grounds that referral flow has no throttle.
  • Kept the evidence grades separate. This is advocacy against advocacy — a coaching clip, no data, author sells the prescribed alternative — not evidence in the sense the adversarial dossier set. Recorded as such on referrals, overview, and index so the gap isn't logged as fully closed.
  • Its rhetorical high point is its logical low point. The Tones and I / Oliver Anthony anecdotes are survivorship selection and cannot support "everyone is one creative project away"; the narrower claim they do support (be shipping when it lands) was kept, the broad one flagged.
  • Reconciled by stage, and said so. The three levers exclude outbound and referrals entirely — they are all one-to-many motions aimed at a founder who already has clients. So the source succeeds rather than refutes the vault's first-client answer (2026-07-17-best-method-first-client). Flagged explicitly as the vault's inference, since that reading conveniently dissolves a conflict that may be real.
  • Recognised a merged claim rather than double-counting it. "Can't bootstrap client #1" and "can't scale past a ceiling" are one structural property (referrals are derived demand) observed at two ends — noted on referrals so the vault doesn't read them as two independent findings.
  • Corroboration weighed, not counted. Content-as-engine now has two voices (information-vs-implementation + this), but both are US coaching videos without conversion data — head-count moved, evidence didn't. Recorded on both pages and in overview's channel contradiction.
  • Possible hidden dependency surfaced: the unnamed speaker of the 07-19 leadership clip may be Dan Martell ("Business Athlete" framework, same genre). If true, two "independent" sources share an author. Logged as tentative inference in three places; unverified — no name appears in either raw note.

Open questions / next:

  • Verify the Dan Martell attribution for 2026-07-19-your-company-cant-outgrow-your-team before either source is treated as corroborating the other.
  • Resolve pick-1 vs pick-3 channels, and the articulation-vs-relationships diagnosis for eugene — both now blocking a credible 90-day plan.
  • Standing gaps unchanged: productization-failure-case source (top), pro-thesis rigorous AI source. New: nothing in the vault explains the partnerships lever.

2026-07-20 — attribution fix (Dan Martell ↔ "Your Company Can't Outgrow Your Team")

Operation: Resolved the open speaker question raised during the same-day ingest. The vault owner confirmed that 2026-07-19-your-company-cant-outgrow-your-team — previously logged as "speaker unnamed" — is also dan-martell. Not a new ingest; a provenance correction with downstream consequences.

Wiki changes:

  • 2026-07-19-your-company-cant-outgrow-your-team: added a Speaker field to Source Metadata (with provenance: curator confirmation, not a documentary citation — the clip still names no one); resolved the Open Question; added dan-martell + marketing-system to Connections.
  • team-growth-ceiling: status line now records the shared authorship; added a cross-branch comparison table (stall → misdiagnosis → real cause → mechanism, internal vs. external); Evidence re-sourced with the attribution; 2 new Contradictions; speaker question closed in Next Questions.
  • dan-martell: extended from one source to two — added his Branch C positions, the through-line section, Evidence split by branch, sharpened the incentive caveat, 2 new Next Questions. Now the vault's only entity spanning two branches.
  • 2026-07-20-referrals-will-sink-your-business: Open Question closed ✅; Change Impact updated to record the same-day confirmation and what was applied.
  • overview: the "possible hidden dependency" contradiction → confirmed, rewritten with the worldview/engine analysis; Branch C paragraph now names the author and flags the non-independence.
  • index: status line, 07-19 source row, dan-martell one-liner, team-growth-ceiling row; replaced the "unverified attribution" gap with a source-independence gap.

Findings / notes:

  • The real cost is independence, and the real gain is coherence. Branch C stops being an anonymous leadership talk (traceability up) but also stops being a voice independent of Branch A (corroboration down). Both recorded; the second matters more, because the vault's whole confidence method is convergence between unconnected traditions.
  • The attribution surfaced a synthesis that was invisible while the author was unknown. Both clips diagnose a stalled founder and relocate the constraint inward — the ceiling is never the market — and both prescribe the same engine, deliberate practice at volume: "practice until we can't get it wrong" (people) and "the only thing you control is the volume of the reps" (marketing). Recorded as a table on team-growth-ceiling and a through-line on dan-martell.
  • Flagged the convenience of that frame. A coach whose diagnosis is always "the constraint is internal and fixable" is describing a world in which coaching is by construction the answer — and "have a coach" is one of his own seven Business Athlete practices. Noted on both pages; not treated as disqualifying.
  • Kept the provenance grade honest. This is curator testimony, not an external citation. Strong enough to act on, recorded as what it is.
  • Recovering the full Business Athlete framework (5 of 7 practices + 3 values missing) is now a findable task rather than a dead end — the highest-value cheap win for Branch C.

Open questions / next:

  • Recover the full Business Athlete framework from Martell's published material.
  • Does "the ceiling is never the market" ever fail? Neither clip admits a genuinely small or shrinking market exists.
  • Unchanged: productization-failure-case source (top gap); pick-1-vs-pick-3; articulation-vs-relationships for eugene; the unexplained partnerships lever.

2026-07-22 — ingest (Stop Cold Calling, Do This Instead)

Operation: Ingested raw/sources/Stop Cold Calling, Do This Instead.md (US sales video, 4:38, undated; third dan-martell source, named in the raw note itself). 16/16 raw sources now ingested. This one closes the standing "unexplained partnerships lever" gap.

Wiki changes:

  • Created 2026-07-22-stop-cold-calling-do-this-instead (#source #sales #partnerships), full template.
  • Created partnerships (#concept #sales #marketing, Status: tentative) — borrowed credibility as the enterprise entry mechanism; the six reverse-engineering questions; the partner-archetype recruiting playbook; the throttle distinction vs. client referrals (recruited vs. derived demand); "ten good partners can replace an outbound sales team"; scope ~$10K+ ACV.
  • marketing-system: Summary re-attributed (single author, now two sources); lever table row updated; new "third lever, filled in" paragraph; one-to-many framing qualified (partnerships is one-to-few-to-many); Evidence; Related; the partnerships Next Question closed ✅.
  • client-acquisition-channels: three-lever bullet qualified; new counter-position on cold outbound paragraph (enterprise-scoped; Martell↔Sebastian convergence on trust-mediated entry); 2 new Contradictions; Evidence + Related.
  • referrals: referral-partners paragraph now points to partnerships; the "referral variant with a throttle?" Next Question closed ✅ — including the resolution of Martell's apparent self-contradiction (anti-referral-dependency yet pro-partner-intros: control of the input differs, not the shape of the output).
  • relationships-as-moat: new "convergence from the opposite pole" paragraph — borrowed vs. built trust; borrowed is faster but rented; Evidence + Related.
  • dan-martell: Summary 2→3 sources; third-source positions block; through-line extended (systematize what worked by luck — every clip converts an accident into a machine); Contradictions updated (3 of 16 sources, largest single-voice share); partnerships Next Question closed ✅.
  • eugene: networking blocker gains a target — recruit a partner archetype (for CV/embedded: industrial-equipment vendors / machine-builder integrators, flagged as untested inference); noted partner intros compose with both rival diagnoses (the room and the vouching).
  • overview: source count 15→16; step-9 partnerships mechanism; new highest-signal item (borrowed credibility / throttle); new ✅ Open Thread; online-vs-in-person contradiction gains the enterprise-convergence nuance; Martell dependency updated to 3 sources; navigation + single-source list.
  • index: +1 source row, +1 concept row, counts 48→50, status line, dan-martell one-liner, 2 Known-Gaps updates.

Findings / notes:

  • The gap this closes was flagged in three places (marketing-system, dan-martell, referrals Next Questions) — all three now cite the answer, and all three answers carry the same caveat: the mechanism is single-author and anecdote-grade ($95K contracts, 7 pharma intros — war stories, no cohort).
  • The best finding is the throttle distinction. Partner intros and client referrals produce the same output (a warm, trust-carrying intro), but partner acquisition is a controllable input while client goodwill is not. That is a structural claim that stands without any of the source's numbers, and it cleanly resolves what would otherwise be Martell contradicting himself across his own clips.
  • A genuine cross-voice convergence, from opposed poles. Martell (content maximalist) and sebastian (content = "Big zero") independently agree that enterprise doors open on trust, not cold outreach — differing only on whose trust (borrowed vs. built). Recorded on relationships-as-moat and in overview's channel contradiction as the strongest support yet for the audience-dependent reconciliation. This is the one place where the third Martell source adds weight despite the shared authorship, because the convergence is with a non-Martell voice.
  • Corrected an earlier vault characterization. The 07-20 ingest described all three levers as "one-to-many motions" — the partnerships playbook (events, win the individual) shows that was overstated; the exclusion of the first-client toolkit is softer than it looked. Qualified on marketing-system and client-acquisition-channels rather than silently rewritten.
  • Named the title's overclaim. "Stop cold calling" prescribes… approaching strangers at events — outbound redirected at a higher-leverage audience, not abolished. Recorded on the source page, partnerships, and client-acquisition-channels so the anti-outbound position isn't over-read against Tony/AB Analytics, whose SMB segment this source never addresses.
  • Source-independence continues to thin: Martell now authors 3 of 16 sources; team-growth-ceiling, marketing-system, and partnerships are one worldview. The restated three-lever taxonomy was logged as framework stability, not corroboration.

Open questions / next:

  • The entry bar is the new blocking question: would a Tata-class (or any consequential) partner meet a no-name solo operator? Decides whether the lever is available to eugene at all. Related: what smaller-scale partners get paid, and the SMB partner archetype for dev services.
  • The 90-day-plan query for eugene now has three unresolved inputs: pick-1-vs-pick-3, articulation-vs-relationships, and now content-vs-partnerships as his lever.
  • Standing gaps unchanged: productization-failure-case source (top), pro-thesis rigorous AI source. Cheap win still open: recover the full Business Athlete framework.

2026-07-23 — ingest (Make My First $100K in Month)

Operation: Ingested raw/sources/Make My First $100K in Month.md (US blueprint video, 22:54, undated; fourth confirmed dan-martell source, named in the raw note). 17/17 raw sources ingested. His first material addressed to a $0 operator — and the ingest that exposed the vault's single-voice problem as larger than counted.

Wiki changes:

  • Created 2026-07-23-make-my-first-100k-in-month (#source #offer-design #sales #demand-generation), full template.
  • dan-martell: 3→4 confirmed sources; fourth-source positions block ($0 protocol); through-line extended (sequence over effort); Contradictions — suspected authorship of 07-18 (near-certain, verbatim mechanism reuse) and 07-17 (plausible, "buy back 10 hours" = his book's frame), possibly 6 of 17; three within-author tensions logged (cold-call vs. stop-cold-calling; parallel engines vs. pick-one; $10K-dislike vs. $10K+ partnerships); pre-first-client Next Question closed ✅.
  • information-vs-implementation: "independent second voice" claim withdrawn (superseded note — the founding anonymous source is near-certainly Martell too); Flowtown 350K→50K added as first (self-reported) outcome figure; scramble/prompt restatement recorded as stability, not corroboration.
  • offer-ladder: second ladder ($500/$1K/$10K) with explicit decoy rationale; Summary re-scoped ("two examples, possibly one voice"); ratio instability (5× vs 10× top) and sellable-flanks vs. decoy-flanks disagreement recorded.
  • productized-service: offer-first camp gains a source (marketing-before-building, Maritime-vs-Flowtown); format triage added ("customer cash funds the eventual product"); $50 paid-waitlist pre-sell — cross-tradition convergence with RU validate-before-build; when-to-productize weighting re-stated (3:2 by sources, maybe 3:1 by voices).
  • pricing-from-value: money map added ($1K floor restated, sub-$10K ceiling, "investment" language); floor's "second assertion may be the same speaker" caveat.
  • outcome-based-selling: time/money/status triad; five offer elements (outcome/deliverable/investment/risk-reversal/urgency); "features tell, outcomes sell."
  • client-acquisition-channels: parallel-engines paragraph (outbound present at $0 → supports stage reading; within-author tension); phone-mining ladder; enterprise-scoping of "stop cold calling" confirmed from inside the corpus (he cold-calls SMB).
  • sales-discipline: closing playbook (chat DM flow, cold-call qualify-and-book); objections-vs-obstacles (first concrete principle for the meta-analysis's objection-script gap); volume rules (100 no's/day, first-5 throwaways, 30-second callback, AI transcript review, spend-nothing-until-paid); calls-vs-chat contradiction softened (one source prescribes both).
  • referrals: third engine added — "ask past the person" network mining (referral-shaped intros with no past client — dodges the cold-start constraint); Eugene next-question updated.
  • marketing-system: scope note upgraded (same-author $0 protocol confirms the staged reading); pick-one contradiction gains the parallel-engines complication.
  • partnerships: stage tension logged ($10K-dislike at $0 vs. $10K+ scope at scale).
  • niche-selection: business-owners-at-$1K+ buyer default; Ikigai as first supply-side filter.
  • eugene: two additions — a concrete day-one move (phone-mining/ask-past-the-person) and a stage/format mismatch (his CV/embedded deals are naturally few×$10K+, the shape Martell's blueprint avoids).
  • Source pages 2026-07-17-design-the-perfect-offer and 2026-07-18-information-is-free-implementation-is-paid: speaker open-questions updated with the attribution evidence.
  • overview: source count 16→17; convergence caveat (US pole is substantially one corpus); 3 new highest-signal items (ask-past-the-person, obstacles-not-objections, money map); new Open Thread ($0 blueprint + single-voice audit); hidden-dependency contradiction expanded (4 confirmed + 2 suspected); calls-vs-chat and channel-count updates; statistics list extended.
  • index: +1 source row, counts 50→51, status line, Martell one-liner, Known Gaps rewrite (source independence), one conflict resolved ✅ (cold-call scoping).

Findings / notes:

  • The biggest finding is not in the source — it's about the corpus. The video reuses the scramble trick, the "10 nuanced and observable problems" prompt, and the "describe their pain better than they can" line verbatim from the anonymous 07-18 content source — in a video that names Martell. That makes him its near-certain author, which retroactively deletes the "independent second voice" the vault granted content-as-engine on 07-20. The 07-17 anchor is also suspect ("buy back 10 hours of your time per week" — Buy Back Your Time is Martell's book; $997 core; chat-first close). If both hold: 6 of 17 sources are one coach, including one of the two "independent anchors," and every intra-US-pole convergence (the $1K floor stated "twice," the three-tier ladder appearing "twice") is self-agreement. The RU side (Rodenko, AB Analytics, Tony) now carries nearly all of the vault's independence weight.
  • Martell contradicts Martell, three times, and the scope lines are never drawn. Cold calling is "the hardest path" (07-22) yet a prescribed channel (07-23); pick one lever (07-20) yet run two engines in parallel (07-23); he dislikes 10×$10K (07-23) yet his partnerships play is $10K+ ACV (07-22). All three plausibly resolve as stage/segment scoping ($0-SMB start vs. $1.5M+ scale/enterprise) — and the parallel-engines protocol actually supports the vault's earlier staged-reading inference — but no clip states the boundary. Recorded without silent reconciliation.
  • One genuine cross-tradition gain: the $50 paid-waitlist pre-sell (money as the validation signal) independently matches the RU "validate before writing code" rule — different tradition, same discipline, and the paid version is the stronger test.
  • First outcome number for content-as-engine, for what it's worth: Flowtown 350K visitors → 50K customers — self-reported, blog-era, different business model. Until now the concept had zero numbers; now it has one bad one.
  • Most portable new tools: objections-vs-obstacles (surface concerns pre-offer), and "ask past the person" (the one referral-shaped motion available with no past clients — directly relevant to eugene's cold start).

Open questions / next:

  • Owner action, cheap and high-value: confirm or refute the two suspected attributions (is the 07-18 content video Martell? is the 07-17 offer-design video Martell?). One answer settles whether the single-voice share is 4, 5, or 6 of 17 — and whether the vault still has two independent anchors or one.
  • Does the 100×$1K sweet spot survive delivery load for a solo operator (100 concurrent customers vs. 10 heavier deals)? The video never costs out support.
  • Where is the seam between Martell's start protocol (both engines, cold calls, SMB) and his scale protocol (one lever, partners, no cold calls)? Revenue, headcount, segment — unstated.
  • Standing gaps unchanged: productization-failure-case source (top), pro-thesis rigorous AI source, Business Athlete framework recovery.
  • The 90-day-plan query for eugene now has a fourth input: the $0 blueprint's sequence (offer → pre-sell → build; phone-mining first) fits him; its format (SMB × $1K chat/cold-call) does not — his domain implies few × $10K+, which pushes toward partnerships and the in-person route instead.

2026-07-26 — ingest (Как Построить МИЛЛИАРДНУЮ КОМПАНИЮ в 2027)

Operation: Ingested raw/sources/Как Построить МИЛЛИАРДНУЮ КОМПАНИЮ в 2027.md (RU YouTube video, 34:52, undated; speaker Oskar Hartmann — KupiVIP/ShoppingLive founder, Fab.com shareholder, identified by context in the raw note). 18/18 raw sources ingested. The vault's first third-tradition voice — VC/product-startup world, independent of both the Martell corpus and the RU dev-sales cluster.

Wiki changes:

  • Created 2026-07-26-how-to-build-a-billion-dollar-company-2027 (#source #gtm #strategy #pricing), full template.
  • Created oskar-hartmann (entity; identification tentative — by raw-note context, not on-camera statement).
  • Created sales-channel-as-moat (#concept) — moat candidate #3: the repeatable, scalable channel with predictable CAC (Pediant vs FlatPay; FlatPay founder on business #6 through the same channel; LTV×⅓-CAC entry-barrier mechanism).
  • Created unit-economics (#concept) — full-cost pricing (forgotten lines: distribution, repeat acquisition, amortization, write-offs; "cash gaps are real losses"), demand→cost→price sequence, buffer/best-year rule, $1,000 credit-card CAC. Partially fills the CAC/LTV gap flagged by the meta-analysis since day one — general principles present, B2B-services version still missing.
  • Created venture-fit (#concept) — ~1% rule, venture stretch kills (Fab.com $1.5B round → bankruptcy), family-owned/slow-growth alternatives, ShoppingLive case, AI capital suction.
  • partnerships: first independent second voice on the page — adversarial on partner-as-savior (46th-priority problem, gatekeeper margin capture, 100–200× shortfalls; his own $20M one-partner deal filed as boost-not-channel), convergent with Martell on many-partners-never-one; scope split recorded (referring partner vs distribution partner — the disappointment mechanism bites the latter harder); status upgraded from "single author" to "mechanism single-author, warning two-voice."
  • pricing-from-value: pricing power as the PMF test (raise prices, watch the flow — the cheapest test in the vault; "simulation of entrepreneurship" for price-takers); "most sell below full cost" joins sell-dear from the cost side; convergence note upgraded — first out-of-school voice.
  • niche-selection: SOM > TAM (HVAC-agent example, Manifest immigration-law unicorn); narrow-first now converged by three independent traditions — the vault's best-supported claim after the core thesis; beachhead nuance added.
  • client-acquisition-channels: attention-economics hierarchy (search → banner → offline events → TV; "average loud beats better silent"; AI noise — 2,000 LinkedIn messages); whale-client counter-position (Pediant — org churn outlives long integrations); third position on channel count (1 repeatable → then multichannel — closest any source comes to stating the vault's staged reconciliation).
  • marketing-system: the machine-vs-arriving-demand core gains an independent second voice (repeatable channel with predictable economics as the company-defining asset); three-lever framework remains Martell-only; pick-one contradiction gains Hartmann's both-ends position.
  • sales-discipline: founder-sells now stated by three traditions — including as Hartmann's explicit investment criterion ("I don't invest where the founder doesn't sell").
  • technical-founder-trap: feature-#26 syndrome — the product-startup form of the trap (builds instead of sells vs can't explain); Hartmann's blunt behavioral fix vs Martell's skill-building one.
  • ai-market-shift: vibe-coding product glut ("10–15 great products in Cloud Code"); attention as the scarcest resource; the ~$4B Anthropic/OpenAI PE land grab; AI labs sucking venture capital dry.
  • methodology-as-moat: two-moat framing updated to three.
  • eugene: two cheap self-tests (SOM question; pricing-power probe on the next quote); venture-fit seconding of his bootstrapped default; many-small-partner-bets caveat.
  • overview: source count 17→18; third-tradition paragraph; step 10 added to Branch A (channel/unit-economics/venture layer); 3 new highest-signal items (pricing-power probe, channel-beats-technology, "46th priority"); three-moats contradiction; channel-count contradiction updated; statistics list extended; unit-economics gap marked partially filled; new ✅ Open Thread.
  • index: +1 source row, +3 concept rows, +1 entity row, counts 51→56, status line, Known Gaps updates (independence improved; adversarial gap; unit-economics partial fill; 6-of-18 renumbering).

Findings / notes:

  • The headline is independence. After the 07-23 single-voice audit shrank the US pole, this ingest materially rebuilds the vault's confidence base: a third tradition independently restates the spine — selling beats building, founder sells first, narrow beats broad, sell dear. Narrow-first (SOM>TAM) and founder-sells are now the vault's most broadly converged claims after the core thesis.
  • The partnerships page finally has a second voice, and it's a warning. Not a flat contradiction: Hartmann's failure mode (single gatekeeper, partner-as-distribution-channel) and Martell's playbook (many recruited referring partners) overlap only partially, and both state the many-never-one rule. The vault's reconciliation — an intro costs the partner nothing, shelf space costs plenty — is recorded as tentative inference.
  • Pediant vs FlatPay is the source's best structural argument: channel beats technology because the channel is reusable across products (business #6) while a product is not. It founds moat candidate #3 and independently corroborates marketing-system's machine-vs-arriving-demand distinction.
  • Scope mismatch flagged throughout: the source addresses product startups choosing venture vs bootstrapped growth; the vault's domain is dev services. Transfer judgments (which claims carry over) are marked per page as the vault's own inference.
  • Grade caution: everything is stage-war-story grade — 100–200×, $1,000 CAC, $4B JVs, 12h/40GB, Apple-45× are uncited; the speaker sells a paid "unicorn creator" program (same incentive caution as Martell); the Tim Cook claim is factually contestable and recorded as a framing device.

Open questions / next:

  • What is the services analogue of FlatPay's door-to-door — the one repeatable channel a dev-services operator could run to predictable closes? (Connects directly to the 90-day-plan query for eugene.)
  • Does the LTV×⅓-CAC rule mean anything for lumpy B2B-services LTV? The flagged unit-economics gap is narrowed, not closed — a B2B-services-specific economics source would close it.
  • Do referring partners (Martell's kind) actually escape Hartmann's disappointment mechanism, or does the 46th-priority problem bite intros too once the novelty fades?
  • For eugene, two cheap owner actions now queued: the pricing-power probe on his next quote, and naming his SOM in one sentence. Standing owner action unchanged: confirm/refute the two suspected Martell attributions (decides 4/5/6 of 18).
  • Standing gaps: productization-failure-case source (still top), pro-thesis rigorous AI source, Business Athlete framework recovery.

2026-07-26 — query follow-up (create concept: TAM / SAM / SOM)

Operation: User asked "what are SOM and TAM?" after the Hartmann ingest, then requested the definitions be turned into concepts. Created one canonical page rather than three stubs (merge-over-duplicates rule): tam-sam-som.

Wiki changes:

  • Created tam-sam-som (#concept #strategy #positioning) — the three nested circles (table), the SOM > TAM directional rule, the beachhead reading (TAM stays behind the SOM as the growth story), and a terminology caveat.
  • Attribution honesty: TAM/SOM usage and all examples trace to 2026-07-26-how-to-build-a-billion-dollar-company-2027; SAM is standard industry vocabulary from general knowledge, present in no ingested source — marked as such in Evidence and Contradictions, with a rule that a future ingested source's definition wins.
  • Linked from niche-selection (SOM>TAM paragraph), venture-fit (Related), the source page (Connections), overview (navigation).
  • index: +1 concept row, counts 56→57, status line.

Findings / notes:

  • The page's one non-obvious contribution: the beachhead vs fortress distinction — a SOM chosen as an entry point to a bigger market (Hartmann/venture logic) and a niche chosen for durable narrowness (Rodenko's AI-weak methodologies) are different strategies wearing the same circle. No source distinguishes them; flagged as an open contradiction.

Open questions / next:

  • For eugene: is industrial-equipment CV/embedded a beachhead or a fortress? The answer changes what "winning the SOM" should lead to.
  • No honest sizing method in the vault yet (all circles in pitch practice are constructed top-down) — worth capturing if a source supplies a bottom-up one.

2026-07-26 — ingest (Главный Принцип УСПЕШНОГО БИЗНЕСА)

Operation: Ingested raw/sources/Главный Принцип УСПЕШНОГО БИЗНЕСА.md (RU YouTube video, 17:52; second oskar-hartmann source, named in the raw note — which also effectively confirms the context-based attribution of the first). 19/19 raw sources ingested. Central thesis: "Сначала продавай, потом строй" — sell first, then build; only money votes.

Wiki changes:

  • Created 2026-07-26-main-principle-of-successful-business (#source #validation #strategy), full template.
  • Created sell-before-build (#concept) — signal hierarchy (click < waitlist < payment < pre-payment); cheap-experiment toolkit as a cost-ordered table (AI-mockup blast, landing→payment screen, fake payment screen, fake-door buttons, Wizard-of-Oz, paid pre-order); acute-pain bridge; duration-without-revenue red flag. Rule is three-tradition (RU validate-first + Martell $50 paid waitlist + Hartmann toolkit); mechanics single-source tentative.
  • productized-service: pre-sell paragraph upgraded to three traditions; sharpened bar recorded (unpaid waitlist doesn't count — makes Martell's paid slot the load-bearing detail); when-to-productize dispute updated — offer-first camp gains its first genuinely independent voice, but with the scope caveat that the camps may answer different questions (validate-before-build vs standardize-after-repetition; vault synthesis, would dissolve the dispute if right).
  • pain-discovery: acute-pain test added ("good business = everything is bad and people still pay"; hospitals) — revealed preference at its purest; smallest-group-by-name targeting (a SOM statement → tam-sam-som).
  • cloning-over-originality: first independent second voice — app studios that "forbid themselves from inventing," Samwer's fake-button eBay clone built in click-count order, robot-data founder copying 10 LLM-data playbooks; adds the clone-then-measure answer to the page's unobservable-details worry; status line updated.
  • sales-discipline: duration-without-revenue red flag recorded as cross-tradition convergence with Rodenko's "год перепаковки — для тех, кто боится взять трубку" (same claim, seller's coach vs investor screening pitches); launch-ugly / endure-reality-early.
  • unit-economics: Zavent post-payment trap ("a bank financing its clients" — payment timing as a unit-economics variable); $80–100 CAC-burn barrier restated (within-author consistency, flagged as such); status line now "single author, two sources."
  • sales-channel-as-moat: CAC-burn restatement noted (stability, not corroboration).
  • oskar-hartmann: second source; framework-stability block (prove-demand-first, money-as-signal, CAC-burn all restated consistently); one within-author tension logged (founder-sells-personally vs automated payment-screen tests — plausibly segment-scoped, unstated); attribution upgraded to confirmed; credibility note (two self-reported failures — coworking, Zavent — unusual among the vault's promotional voices).
  • eugene: runnable pre-build checklist (10 named industrial prospects; deposit-backed audit/discovery sprint as the services analogue of a paid waitlist; semi-manual first CV-inspection delivery); composes with phone-mining.
  • overview: source count 18→19; third-tradition thread expanded (two sources, framework stability, cloning corroboration, deception caveat); step 5 gains sell-before-build; 2 new highest-signal items (only-money-votes / coworking test; acute-pain test); when-to-productize contradiction updated; statistics list extended.
  • index: +1 source row, +1 concept row, counts 57→59, status line, Hartmann one-liner.

Findings / notes:

  • Best structural gain: validate-before-build is now the vault's second three-tradition claim (after narrow-first/SOM). RU dev-sales, Martell, and Hartmann all state it independently, and Hartmann resolves the apparent Martell-vs-Hartmann waitlist discrepancy himself: unpaid lists don't count, paid ones do — the two prescriptions agree at the money bar.
  • cloning-over-originality finally has a second voice — it had been flagged since June as single-lineage. Hartmann's cases add the missing mechanism: where Pabrai clones judgment, Samwer clones the artifact and lets instrumentation pick priorities.
  • Ethics recorded, not endorsed: two toolkit rows (fake payment screen, fake-door buttons) collect purchase intent under false pretenses; caveat placed on both the source page and sell-before-build, with the honest variants (visible pre-order, paid waitlist, manual MVP) noted as carrying most of the same signal.
  • "Build.ai" flagged as almost certainly Builder.ai (general-knowledge note, marked as such): the company collapsed into insolvency in 2025 amid allegations its AI was largely manual work — which makes Hartmann's "great way to start" framing half the story; the same case is the cautionary tale for never graduating from Wizard-of-Oz.
  • One new tension for the articulation thread: "don't dictate what the customer should want" vs pain-discovery/outcome-based-selling's name-the-pain-better-than-they-can. Likely scoped (product demand vs service framing); logged on sell-before-build and the source page.

Open questions / next:

  • What is the smallest paid test for a dev-services offer? (Paid audit? Deposit-backed discovery sprint?) Feeds eugene's 90-day-plan query directly — arguably its missing first step.
  • Does the Zavent pre-pay rule survive B2B services reality, where net-60/90 post-payment is standard enterprise practice? If enterprise never prepays, is enterprise demand untestable by this method — or is a paid pilot the equivalent?
  • Hartmann's two sources are consistent; a third would test whether the stability holds across formats (talk vs interview).
  • Standing gaps: productization-failure-case source (still top — note the when-to-productize dispute may partially dissolve instead), pro-thesis rigorous AI source, Business Athlete recovery, the two Martell attribution checks.

2026-07-26 — query (Eugene's concrete 90-day plan)

Operation: Answered the standing next-query flagged on index, eugene, client-acquisition-channels and overview since 2026-07-17: turn the corpus into eugene's dated channel/niche plan, and resolve the live tension inside it (in-person first-client route vs. building a marketing-system early). Read-only over 12 existing pages; no source claim altered.

Wiki changes:

  • Created 2026-07-26-eugene-90-day-plan (#query #sales #planning) — assumptions table (A1–A5, all tentative), the stage-based resolution, channel commitment, week-by-week plan 2026-07-26 → 2026-10-24, Day-90 gate, evidence trail, follow-ups.
  • eugene: 90-day-plan Next Question closed ✅; new bullet on the two rival diagnoses now being cheaply testable; 2 Related Pages links.
  • sell-before-build: its "smallest paid test for a dev-services offer" question struck through with a proposed answer (fixed-scope paid diagnostic) + its falsifier.
  • client-acquisition-channels: the "which 3 channels for Eugene" question answered ✅, with the load-bearing assumption named.
  • 2026-07-17-best-method-first-client: successor callout — method vs. schedule; core recommendation explicitly preserved.
  • overview: "Application is unstarted" → "Application has started — an experiment awaiting data"; stale Timelines / Queries / Lint reports: empty navigation line corrected (it had listed 3 existing pages as empty).
  • index: +1 query row, counts 59→60, queries 1→2, status line, next-operation note.

Findings / notes:

  • The tension resolves by stage, not by compromise. The three-lever map comes from a clip aimed at a stalled-$1.5M founder and omits outbound/referrals/in-person entirely; the same author runs one-to-one at $0; oskar-hartmann nearly states the staged rule outright. So: one-to-one is the engine. But not to zero — marketing-system's ~6-month lag means "start it when you need it" starts it late, and publishing doubles as the articulation reps technical-founder-trap prescribes. Budgeted at ~2h/week, explicitly not measured on leads.
  • pick-1 vs pick-3 dissolves for this case under the units reading the vault had already inferred: one lever (relationship-mediated one-to-one), three venues. Both rules satisfied at once — the conflict is not settled in general, only routed around here.
  • Four vault-original contributions, all inference and marked as such: (1) the stage resolution with its non-zero-deposit exception; (2) the Week-2 phone mine as the discriminator between technical-founder-trap and relationships-as-moat — the vault has called those two diagnoses "testable against each other" since 07-20 without ever specifying the test; (3) the paid diagnostic as the smallest paid test for dev services; (4) a Day-90 gate scored on inputs, applying reps-not-views to the plan itself.
  • Honest weak joints, recorded in the plan rather than smoothed over: the niche (A2) and the enterprise-lite buyer read (A5) are vault inferences that carry the entire channel decision; the industry room is unnamed; A4 (prior paid work to mine) is unverified; every threshold in the gate is invented, its function being pre-commitment, not accuracy.
  • Evidence-quality caveat restated on the page: no failure-case source exists for the productized/sell-dear model, so nothing in the plan has been tested against a documented loss.

Open questions / next:

  • Owner-answerable, and they unblock the plan's weakest joints: which machine-builder/automation association or trade fair meets often enough for the 3×/6× mechanic, and will an industrial buyer pay for a diagnostic at all (if free feasibility is the norm, Month 2 needs redesign).
  • Standing owner actions unchanged: confirm/refute the two suspected dan-martell attributions (several rungs of this plan rest on his corpus — 4 vs 6 of 19 matters here); the pricing-power probe; naming the SOM in one sentence (now Week 1, step 1).
  • A lint is due. Last full lint was 2026-07-18 at 40 pages; the wiki is now 60, with sales-channel-as-moat, unit-economics, venture-fit all single-source tentative and unlinted, and this query adding a page whose inbound links are all fresh.
  • Standing gaps: productization-failure-case source (still top), pro-thesis rigorous AI source, Business Athlete framework recovery.

2026-07-29 — ingest (Start a business with Claude Code)

Operation: Ingested raw/sources/Start a business with Claude Code.md (conclusions note on a YouTube Short, 0:42, author unnamed, undated). 20/20 raw sources ingested. The vault's shortest and thinnest source: a $0→$1M playbook with claude-code as the full GTM stack — landing page + invented company + waitlist, scraped prospect list, cold-outbound script, close the sale, then build the product.

Wiki changes:

  • Created 2026-07-29-start-a-business-with-claude-code (#source #validation #gtm #ai), full template; evidence grade stated up front (42s, no cases, no figures, no named speaker — pure prescription).
  • Created claude-code (#entity #tool #ai) — the vault's first tool entity: three corpus roles (build engine/commoditizer per Hartmann's "10–15 products in Cloud Code"; full GTM stack per this source; domain tutor/leveler), harness-market adjacency, capability counter-evidence linked.
  • sell-before-build: fourth-voice echo noted in Summary (not counted toward independence until attributed); AI-era compressed variant paragraph — its unpaid waitlist fails the vault's money bar but its closed sale exceeds it; invented-company front filed with the deception-based toolkit rows; Evidence + Contradictions updated.
  • ai-market-shift: Leveler row added to the roles table (domain knowledge on demand collapses the entry barrier); synthesis paragraph noting the odd alliance — ai-productivity-evidence's novice-tilt cuts against expert-productivity hype but mildly for the leveler role, while still contesting the "then ask Claude to build it" step; claude-code added to Related Pages.
  • client-acquisition-channels: second $0-stage voice for cold outbound — AI-scraped list + AI-written scripts as the entire first channel; matches Martell's $0 blueprint shape (caveat: if the author is Martell, the match is self-agreement); further supports the $0/SMB-vs-enterprise scoping of cold outbound.
  • sales-channel-as-moat: slogan echo recorded ("distribution is the moat, not code") — second voice for the priority claim, zero corroboration of the repeatability/CAC mechanism; status explicitly unchanged.
  • technical-founder-trap: Evidence bullet — the source's stated audience ("developers who over-invest in building and under-invest in selling"; blocker psychological, not technical) as a slogan-grade restatement of the build-instead-of-sell form.
  • overview: source count 19→20; corpus paragraph and navigation updated; entities line gains a tools group.
  • index: +1 source row, +1 entity row, counts 60→62, status line, Known Gaps updated (third unattributed source joins the two suspected Martell items; denominators refreshed to /20), lint note 60→62.

Findings / notes:

  • Nothing in this source is new to the vault as a claim — that is the finding. Sell-first, distribution-over-product, cold outbound at $0, bias to action: each already held, with better support elsewhere. Its marginal contributions are (1) the Leveler role for AI (genuinely new to ai-market-shift's table), (2) Claude Code named as the executor of the entire pre-product stack, and (3) an ultra-compressed confirmation that this playbook has reached mass short-form content — the vault's spine is now ambient advice.
  • Attribution discipline applied up front this time. After the 07-23 audit (anonymous sources later unmasked as Martell), this source is held at echo-grade on every page it touches: it shifts no status lines, gains no independence credit, and each page notes the attribution condition explicitly.
  • The playbook passes the vault's own money bar — but not where it appears to. Its waitlist is unpaid (fails the Hartmann/Martell bar); its actual validation event is the closed sale in step 5, which sits above mere payment in the signal hierarchy. Recorded on sell-before-build.
  • Ethics/compliance recorded, not endorsed: the invented-company front is a fake business (kin to fake payment screens — same unmodeled reputational/legal cost), and scraped-list cold email carries GDPR/CAN-SPAM exposure. Neither cost appears in the source.
  • Contrast with the vault's own application: 2026-07-26-eugene-90-day-plan reaches the same sell-first conclusion but replaces the fake front with a paid diagnostic — the honest variant of the same discipline.

Open questions / next:

  • Who is the author? Third unattributed short-form source (after the two suspected Martell items). No verbatim marker links it to any known voice; cheap owner action to check.
  • Does "Claude will tell you how" survive regulated/physical domains (the owner's industrial CV niche), or only commodity digital services?
  • A lint is now overdue — 40 pages at last full lint (2026-07-18), 62 now; this ingest also added a tool entity, a page type the lint conventions haven't seen.
  • Standing gaps unchanged: productization-failure-case source (top), pro-thesis rigorous AI source, Business Athlete framework recovery, the two Martell attribution checks.

2026-07-29 — attribution update (Start a business with Claude Code = Dan Martell)

Operation: Vault owner confirmed the author of 2026-07-29-start-a-business-with-claude-code is dan-martell (the short names no speaker; same provenance class as the 07-19 confirmation). His 5th confirmed source of 20 — possibly 7 with the two still-suspected. Propagated the relabel from "unattributed echo" to "within-author self-agreement" across every page the same-day ingest touched.

Wiki changes:

  • 2026-07-29-start-a-business-with-claude-code: metadata updated (author, provenance); author open question closed ✅; Connections rewritten (each link now names its relation — restatement vs new claim); Change Impact extended.
  • dan-martell: 4→5 confirmed; fifth-source positions block (AI as the entire pre-product stack; the Leveler claim; distribution-is-the-moat slogan); two within-author drifts logged — the short's waitlist is unpaid against his own $50-paid-slot rule (07-23), and the invented-company front is deception-shaped with unmodeled cost; through-line updated to five sources; denominators refreshed (5 of 20, possibly 7); provenance note now covers both owner-confirmed clips; claude-code added to Related Pages.
  • sell-before-build: "fourth voice" withdrawn — within-author restatement (Martell already counts among the rule's three traditions); the unpaid-waitlist nuance upgraded from cross-source observation to within-author drift.
  • client-acquisition-channels: "independent match to the $0 blueprint" evaporated as predicted by the ingest's own conditional — relabeled framework stability; one genuine gain kept: the $0/SMB-vs-enterprise scoping of cold outbound now holds across every format he publishes.
  • sales-channel-as-moat: the slogan echo folds into the already-recorded Martell/marketing-system convergence — no second voice; status explicitly unchanged.
  • technical-founder-trap: Evidence bullet relabeled — same author as the page's founding source; restatement, not corroboration.
  • ai-market-shift: Leveler attribution updated — Martell's first claim in the roles table; still single-source, slogan-grade.
  • claude-code: roles table attributed; Contradictions sharpened — all promotional claims about the tool trace to exactly two voices (Martell, Hartmann).
  • overview: Martell cluster four→five confirmed; hidden-dependency contradiction updated (5 of 20, possibly 7); corpus paragraph and navigation relabeled.
  • index: source row, Martell one-liner, Known Gaps (single-voice problem worsened: 5 of 20), status line.

Findings / notes:

  • The echo-grade discipline paid for itself within a day. Every page had recorded the source with an explicit attribution condition and zero independence weight — so the confirmation changed labels, not conclusions. No claim needs re-weighing; this is the cheapest possible outcome of an attribution reveal, and the direct payoff of the 07-23 audit's lesson.
  • Martell's share is now the vault's largest by a wide margin: 25% confirmed, possibly 35%. The RU side (Rodenko, AB Analytics, Tony) and Hartmann now carry essentially all of the vault's independence weight. Practical consequence: any future US-coaching short-form ingest should be presumed Martell-adjacent until shown otherwise.
  • The two within-author drifts are the only new information. Unpaid waitlist (vs his own load-bearing $50 paid slot) and the fake-company front (vs his "spend nothing until customers pay" — which at least is honest about money) — both plausibly short-form lossiness, but they show his compressed formats shed exactly the details his fuller protocols mark as critical. A caution for weighting any 42-second source, his or not.
  • One thing the attribution strengthens rather than weakens: the enterprise-vs-$0/SMB scoping of cold outbound is now consistent across five Martell formats — within-author stability at that sample size is real evidence about his framework, if still none about the world.

Open questions / next:

  • The two remaining attribution checks (2026-07-18-information-is-free-implementation-is-paid near-certain, 2026-07-17-design-the-perfect-offer plausible) are the same cheap owner ask that just resolved this one — settling them decides 5, 6, or 7 of 20, and whether the vault has two independent anchors or one.
  • A lint is overdue (40 pages at last full lint, 62 now) — and it should now also check every page still describing Martell as "4 confirmed" or using stale /17–/19 denominators.
  • Standing gaps unchanged: productization-failure-case source (top), pro-thesis rigorous AI source, Business Athlete framework recovery.
Reference in New Issue View Git Blame Copy Permalink