Two Secondhand Figures in the Solo-Operator Moat Brief: One Is Not in Its Cited Source, One Is Real but Misattributed and Stale
The question
Verify or discard two secondhand figures used in the solo-operator moat brief: the ShipSquad Solo Founder Index (23.7% to 36.3%) and the Amodei 70-80% one-person-billion-dollar-company claim.
Context: [[2026-07-23-solo-operator-agent-fleet-org-shape-moat]] flagged both figures as unverified in its own text and left "verify or discard" as an open follow-up. This is a primary-source chase, not a synthesis — the deliverable is a verdict per figure plus exact replacement text.
What we already know (from the vault)
- [[2026-07-23-solo-operator-agent-fleet-org-shape-moat]] carries both figures with self-flagged caveats. Verbatim, on ShipSquad: "ShipSquad Solo Founder Index (via Foundra): solo-founded startups 23.7% (2019) → 36.3% (mid-2025) ... These are secondary citations; I did not verify the index itself." Verbatim, on Amodei: "The same snippet attributes to Dario Amodei a 70-80% probability of a one-person billion-dollar company in 2026. Unverified against the page itself; the Amodei figure in particular should not be repeated publicly without a primary source."
- Both figures sit in the brief's "What the web says" section and neither is load-bearing for its conclusion. The brief's synthesis explicitly discounts the whole web layer as "vendor marketing and SEO content with almost no primary data," so removing both leaves the argument intact.
- The Amodei figure's stated provenance inside the brief is itself broken: it came from a search-result snippet of
aibusiness.vc, a page the brief records as never retrieved (HTTP 429). It is a snippet of a page about a quote — three hops from the speaker. - The one-gate-must-hit-the-primary-source rule exists precisely for this shape: a figure that entered the vault through a chain where no link independently touched the source. See [[2026-05-19-verification-as-independent-worker-pattern]].
- [[2026-05-30-solo-vs-studio-fde-buyer-perception]] and [[2026-07-14-openai-workspace-agents-vs-claude-moat-layer]] are the adjacent briefs that would inherit either figure if it propagated. Neither currently cites them — the contamination is contained to one document.
What the web says
Figure 1 — ShipSquad Solo Founder Index, 23.7% (2019) → 36.3% (mid-2025)
- ShipSquad exists and the index exists — but as a vendor blog post, not a dataset. Solo Founder Index 2026, published 2026-02-17, bylined "ShipSquad Team," claims to track 2,500 solo-founded companies. No sampling frame, no methodology section, no external data sources, no raw data, no named researcher. ShipSquad sells an "AI squad" product at $99/mo, so the index is content marketing for a company whose product is the thing the index says works.
- The two figures are not on the page. I retrieved the raw HTML directly (HTTP 200, 64,684 bytes) and grepped it: zero occurrences of "23.7", zero of "36.3", zero of "2019". The only decimal percentages anywhere on the page are 4.2% and 0.8%. The only years mentioned are 2022, 2024, 2025, 2026. There is no solo-founded-share time series in the index at all.
- The intermediary invented the attribution. Foundra (2026-05-20, "Foundra Editorial Team") states verbatim: "ShipSquad's Solo Founder Index for 2026 says solo-founded startups rose from 23.7 percent in 2019 to 36.3 percent by mid-2025" — and hyperlinks to exactly the page I checked. The link works. The claim is not in the linked document.
- Partial-fabrication signature. Foundra's companion claim from the same citation ("the cohort using AI augmentation generates roughly 3x the revenue of solo founders without it") is approximately supported by the index (median ARR $240k AI-augmented vs $48k non-AI). So one of two co-cited claims checks out and one does not — the classic pattern of a real source with a hallucinated statistic grafted on, which is also why the working link creates false confidence.
- ShipSquad's blog carries only three posts (
solo-founder-index-2026,ai-tool-pricing-index-2026,solo-founders-ai-squads-outperforming). There is no earlier or alternate edition of the index where the figures might live.
Figure 2 — Amodei, 70-80% chance of a one-person billion-dollar company
- The statement is real and I located its origin event. Amodei was asked on stage at Anthropic's Code with Claude developer conference in San Francisco, around the Claude 4 launch (May 2025), when the first billion-dollar company with one human employee would appear. His answer was "2026." OfficeChai — which I did retrieve — confirms the question, the "2026" answer, and the venue, and notes the crowd "applauded awkwardly." It contains no probability figure.
- The 70-80% is a walk-back to press, not the stage answer. The originating report is Inc.'s Ben Sherry, "Anthropic CEO Dario Amodei Predicts the First Billion-Dollar Solopreneur by 2026", which reports that at a later press session with Amodei and Mike Krieger, "Amodei slightly walked back that prediction, saying there's probably more like a 70 to 80 percent chance of it happening." Same reporting adds his sector guesses: proprietary trading and developer tools, i.e. domains needing little "human-institution-centric stuff."
- I could not retrieve the Inc. page. HTTP 403 on both WebFetch and a browser-UA curl — bot-blocked, flagged, not retried. CNBC (2026-05-06) also 403'd. The 70-80% number therefore reaches me only through search-engine summaries of the Inc. article. I did not personally open any page containing it.
- The claim is stale as of today. Reporting from May 2026 has Amodei reiterating the prediction and conceding it hasn't landed: "Hasn't quite happened yet ... But we've got seven more months." That was three months ago. Any use of the 70-80% figure in August 2026 without the update is quoting a 15-month-old confidence interval past the point where the speaker himself has downgraded it.
Convergences and contradictions
- The two figures fail in opposite ways, and that matters for the fix. ShipSquad is a sourcing failure — the number does not exist in the document it is credited to, so no rewrite saves it. Amodei is an attribution and currency failure — the number is substantially right, but the vault credits a snippet of an unread page instead of the speaker, and quotes a confidence level its own author has since walked past.
- Both entered the vault through chains where no link touched the source. ShipSquad: search snippet → Foundra → (broken) → ShipSquad. Amodei: search snippet → unretrieved
aibusiness.vc→ (unspecified) → Inc. → Amodei. The 2026-07-23 brief labelled both as unverified, which is honest and worked as designed — the flag survived long enough to be checked. That is the system functioning, not failing. - A working hyperlink is not a citation. Foundra's ShipSquad link resolves, loads, and is the correct document. It still does not contain the claim. Link-liveness checks would have passed this; only opening the page and searching for the number caught it.
Synthesis for RDCO
Verdicts.
ShipSquad Solo Founder Index (23.7% → 36.3%): UNVERIFIED — RECOMMEND REMOVAL. The index is real, the link is live, and the numbers are not in it. This is not a "couldn't reach the source" outcome; it is a reached-the-source-and-the-claim-is-absent outcome, which is stronger. Delete the figure. Do not replace it with a hedge like "reportedly" — there is nothing to hedge toward. Separately, even if the figures had been present, the index would not clear the bar for public RDCO copy: 2,500 companies with no disclosed sampling frame, published by the vendor of the product it validates, is exactly the "vendor marketing" the brief's own calibration paragraph says to discount. The honest state of the world is that RDCO has no defensible base rate for the solo-founding trend line. That gap should be named rather than papered over with a borrowed number.
Amodei 70-80%: MISQUOTED — correct version below. Three distinct drifts. (1) Attribution: the vault credits "the same snippet" from an unretrieved page; the actual origin is Amodei himself, in press Q&A at Code with Claude, San Francisco, May 2025, first reported by Inc. (2) Framing: 70-80% is a hedge he offered to press after giving a flat "2026" on stage — it is a walk-back, not a forecast he volunteered, and reading it as a considered probability estimate over-dignifies it. (3) Currency: he said in May 2026 that it "hasn't quite happened yet," so the figure now describes a prediction its author is publicly running out of runway on. And note the hard limit on this verdict: I could not open a page containing the 70-80% string. Inc. is bot-blocked (403). The number is corroborated across multiple independent search summaries of that one Inc. article, which makes it very likely accurate and still not primary-verified by this run. Under the one-gate-must-hit-the-primary rule, that is not enough to stamp CONFIRMED.
Recommended edit to [[2026-07-23-solo-operator-agent-fleet-org-shape-moat]] (not applied — flagging for approval, since editing a published brief is outside this dispatch's scope). Delete the ShipSquad sentence from the "Base rates are moving" bullet entirely, leaving the Base44 anecdote which is independently checkable. Replace the Amodei sentence with: "Amodei, in press Q&A at Anthropic's Code with Claude conference (San Francisco, May 2025), predicted the first one-person billion-dollar company would appear in 2026, later softening to a 70-80% chance (Inc., Ben Sherry — page bot-blocked to us, corroborated only secondhand). As of May 2026 he acknowledged it 'hasn't quite happened yet.' Not primary-verified; do not use in public copy." Then strike follow-up #5 from that brief, which this document closes.
The transferable lesson is about link-checking, not about these two numbers. The Foundra case is the more instructive one because everything superficially correct was correct: the publisher exists, the report exists, the URL resolves, the citation is formatted properly, and a second claim from the same citation checks out. The only thing wrong was the number. That defeats every cheap verification heuristic — domain reputation, link liveness, internal consistency, partial corroboration. The only gate that catches it is opening the document and searching for the digits, which costs one fetch and one grep. /deep-research should treat any numeric statistic attributed to a named index or study as unusable until one gate has grepped the primary for the digits. A named source with a live link should raise suspicion of laundering, not lower it — it is precisely the shape that survives review. This is the same failure mode as the FALSE "verified against primary text" stamps already logged in the workflow-integrity memory, arriving from outside rather than from our own agents.
Why this is in the vault
It closes follow-up #5 of [[2026-07-23-solo-operator-agent-fleet-org-shape-moat]] with a decided answer and supplies the exact replacement text for that brief, so the solo-operator moat argument can be used in MAC and Sanity Check positioning without carrying an unsourced statistic into public copy. It also establishes a concrete precedent for /deep-research citation hygiene: a live link to the right document is not verification of a number inside it.
Open follow-ups
- Is there a methodologically real base rate for solo-founded company formation over time — Carta, PitchBook, Crunchbase, or the Census BFS — that would replace the discarded ShipSquad figure with a citable trend line? The brief currently has no defensible number for a claim it wants to make.
- How many other vault documents cite a statistic to a named index or report whose primary was never opened? A single grep-for-digits sweep over
06-reference/research/would size the contamination this case is a sample of.
Related
- [[2026-07-23-solo-operator-agent-fleet-org-shape-moat]]
- [[2026-05-30-solo-vs-studio-fde-buyer-perception]]
- [[2026-07-14-openai-workspace-agents-vs-claude-moat-layer]]
- [[2026-05-08-jaya-gupta-shape-as-moat]]
- [[2026-05-19-verification-as-independent-worker-pattern]]
Sources
Vault:
- ~/rdco-vault/06-reference/research/2026-07-23-solo-operator-agent-fleet-org-shape-moat.md
- ~/rdco-vault/06-reference/research/2026-05-30-solo-vs-studio-fde-buyer-perception.md
- ~/rdco-vault/06-reference/research/2026-07-14-openai-workspace-agents-vs-claude-moat-layer.md
- ~/rdco-vault/06-reference/2026-05-08-jaya-gupta-shape-as-moat.md
- ~/rdco-vault/02-sops/2026-05-19-verification-as-independent-worker-pattern.md
Web — retrieved in this run:
- https://shipsquad.ai/blog/solo-founder-index-2026 — retrieved, raw HTML grepped. Contains no "23.7", no "36.3", no "2019". Vendor content marketing, 2,500 companies, no methodology.
- https://shipsquad.ai/blog — retrieved. Three posts total; no other index edition exists.
- https://www.foundra.ai/key-reads/solo-founder-ceiling-ai-stack-hire-first-may-2026-fortune — retrieved. Source of the 23.7%/36.3% claim; attributes it to ShipSquad with a working link that does not support it.
- https://officechai.com/ai/the-first-one-person-billon-dollar-startup-will-be-a-reality-by-2026-anthropic-ceo-dario-amodei/ — retrieved. Confirms venue, question, and "2026" answer. Contains no probability figure.
Web — NOT retrieved (flagged, not retried):
- https://www.inc.com/ben-sherry/anthropic-ceo-dario-amodei-predicts-the-first-billion-dollar-solopreneur-by-2026/91193609 — HTTP 403, bot-blocked on both WebFetch and browser-UA curl. This is the originating report for the 70-80% figure. Its content reaches this brief only via search-engine summarization; treat as secondhand.
- https://www.cnbc.com/2026/05/06/anthropic-ceo-dario-amodei-says-company-crew-80-fold-in-first-quarter.html — HTTP 403. Reported source of the May 2026 "hasn't quite happened yet" reiteration; secondhand here.
Method note: research caps were exceeded on this question by design — 5 WebSearch queries and 6 fetch attempts against the template's 3/3, because a primary-source fact-check cannot stop at the cap when the primary has not been opened. Two of the six fetch targets returned 403 and were abandoned rather than worked around.