Skip to content
How the ledger works

Every claim this publication has made appears here as a row. Each row carries an ID (AM-NNN, OPS-NNN, or AM-CANON-NNN), a verdict (Holding, Partial, or Not holding), and a next-review date. The cadence is 30–90 days. When the evidence moves, the verdict moves with it — and the correction appends, dated, never quietly. The original claim text is preserved forever; retractions are events, not silent edits. The full operating model is at the accountability architecture for AI-written publications (AM-CANON-001).

How verdicts are assigned

Every article asserts exactly one primary claim, created at publish with its own ID and review dates. Enterprise claims (AM) are re-reviewed every 30–90 days; operators claims (OPS) every 30–45 days, because SMB tooling and pricing shift faster. A re-review checks the claim against the primary sources it rests on — verify first, then judge.

Three verdicts exist. Holding: the claim is still supported; the review date updates and a new next-review date is set. Partial: one substantive part has been revised; a dated correction appends. Not holding: the claim has been falsified or overtaken; the article stays up with its correction log at the foot. The claim text itself is never edited — corrections append, status changes, nothing is quietly removed. Each row below shows the published, last-reviewed, and next-review dates.

The full editorial charter is at /standards/ and the production model at /how-its-written/.

Vigil·last review 44h ago·next review cycle 27 Jun 2026

Every claim this publication has made, and whether it still holds.

The point of writing about enterprise AI is to be right for longer than a news cycle. This page tracks every argument this publication has made, reviewed on a 30–90 day rhythm. If something stops holding, it's marked and the piece is annotated. Nothing is quietly removed. Claims made by others — vendors, analysts, regulators — are tracked separately at /archive/.

231holding
24partial
07not holding
Segment
Status
Holding-up index — each row is a tracked claim with its current status and next scheduled review date.
StatusClaimNext review
Holding

OPS-099 · pub 9 Jun 2026 · rev 9 Jun 2026

Anchored on Notion's Developer Platform release notes (13 May 2026, notion.com/releases/2026-05-13 — Workers launch, Sam Lambert CEO PlanetScale customer quote) and Notion's 'Understand pricing for Workers' help page (free during beta on Business/Enterprise; billing starts 11 Aug 2026; $0.0023 per run via credits ≈ 4,348 runs per $10). Business plan $15/user/month annual (consistent with OPS-092's verified figure). VERIFIED 2026-06-09 via direct fetch of both Notion pages. PRECISION: monthly cost figures in the body's table (daily ≈$0.07, hourly ≈$1.66, 5-min ≈$19.87) are arithmetic on the published per-run rate, labeled as such. Differentiation from OPS-092 is editorially load-bearing and stated up front: Custom Agents = conversational AI at $10/1,000 credits metered from 4 May; Workers = code runtime per-run metered from 11 Aug — different product, meter, and date. Operator-register advisory; the claim is the build-and-measure-in-the-free-window stance. 30-day cadence timed to land one re-review before the 11 Aug billing start. Triggers: (1) Notion shifts the billing date or the $0.0023 rate; (2) a bundled run allowance lands in the Business plan; (3) beta reliability limits make Workers unfit for scheduled glue work. Siblings: OPS-092 (Custom Agents credit pricing), OPS-098 (Zapier MCP two-task billing — co-published), OPS-083 (bill-rising), the bootstrapped-SaaS cost-discipline read.

Read article →

+27dnext review
Holding

OPS-098 · pub 9 Jun 2026 · rev 9 Jun 2026

Anchored on Zapier Help Center 'How Zapier MCP usage works' (article 45645738385805, updated 4 Jun 2026 — fetched directly): two tasks per successful tool call, fixed rate, no separate billing, no session cap, failed calls don't bill; and zapier.com/pricing (Professional from $19.99/month annual billing, 750 tasks). VERIFIED 2026-06-09. The capacity figures (375 actions; 100 calls/day exhausts in under 4 days; 20 calls/day = 1,200 tasks/month) are arithmetic on the cited rates, labeled as such in the body's table caption. Quote is institutional (Zapier documentation), attributed as such — no named individual exists for it; no person invented. Operator-register advisory; the claim is the budget-at-double + actions-not-monitoring discipline, not a recommendation against Zapier MCP. 30-day cadence, automation-platform billing moves fast. Triggers: (1) Zapier changes the two-task rate, caps or itemises MCP usage; (2) an MCP-specific plan/pool decouples agent calls from tasks; (3) a competing platform prices MCP calls at task-parity, changing the new-build comparison. Siblings: OPS-092 (Notion Custom Agents credit pricing), OPS-083 (ai-cheaper-but-your-bill-rising), OPS-099 (Notion Workers free window — co-published), the bootstrapped-SaaS cost-discipline read.

Read article →

+27dnext review
Holding

OPS-097 · pub 9 Jun 2026 · rev 9 Jun 2026

Anchored on Google's fraud and scams advisory of 8 Jun 2026 (blog.google, Laurie Richardson, VP Trust and Safety — fetched directly): names calendar phishing (malicious links via event invites) and ClickFix (fake fix-it pages soliciting terminal paste) as active patterns; cites ~$580B estimated global fraud losses for 2025; carries the verbatim QR-code line quoted in the body. VERIFIED 2026-06-09 via direct fetch of the advisory. The AI-tooling connection (trust habits raised by schedulers/notetakers) is the publication's own analytical frame, not a Google claim — framed as observation in the body. Calendar-settings guidance described functionally (restrict auto-add, report phishing) rather than menu-path-exact, because Google moves UI paths; the body says 'Settings, Event settings' at the functional level only. Operator-register advisory. 30-day cadence, scam waves shift fast. Triggers: (1) Google ships default platform-wide protections closing calendar auto-add abuse; (2) the named patterns are supplanted by a materially different wave; (3) AI-scheduling tools add mitigations making manual settings unnecessary. Siblings: OPS-086 (meeting notetakers), OPS-088 (coding-CLI security), OPS-078 (agent kill-switch), the operators shadow-AI capability read.

Read article →

+27dnext review
Holding

AM-210 · pub 9 Jun 2026 · rev 9 Jun 2026

Anchored on three Gartner artifacts. (1) Hype Cycle for Agentic AI, published 2 Apr 2026 (Kandaswamy, Ramos, Olliffe, Coshow, den Hamer, Brethenoux; gartner.com/en/articles/hype-cycle-for-agentic-ai + document 7671861): agent-washing definition quoted verbatim; 17% deployed; 42% within 12 months; +22% the following year (the body states 'more than 80% within two years' per the fact-check correction — the scout's original '60%+' UNDERSTATED Gartner's own data); ~130 of thousands of vendors assessed as delivering real capability. (2) Standing prediction, 25 Jun 2025 (Anushree Verma, Sr Director Analyst): 'Over 40% of agentic AI projects will be canceled by the end of 2027, due to escalating costs, unclear business value or inadequate risk controls' — deliberately cited as a JUNE 2025 prediction, not new 2026 research, per fact-check. (3) Supply-chain agent-washing release, 20 May 2026 (Barcelona Symposium): Jan Snoeckx (Senior Director Analyst) two-sentence quote verbatim-confirmed; the 'end-to-end autonomous planning before 2027 = overstating' line presented as the RELEASE'S finding, not a Snoeckx verbatim quote, per fact-check (secondaries carry it as paraphrase). Gartner pages 403 to crawlers; figures corroborated via 2+ named secondaries each; canonical newsroom/article URLs verified via search. The four-property capability test (goal-directed sequencing, tool action, state, deviation handling) is the publication's operationalisation of Gartner's described capability bar — framed as ours, not as a Gartner framework. VERIFIED 2026-06-09 by hostile fact-check. 90-day cadence. Triggers: (1) a subsequent Gartner assessment materially revising the ~130 count or intent figures; (2) certification/standards making capability claims independently verifiable; (3) market data showing label-bought deployments succeed at the same rate as test-bought ones. Siblings: AM-187 (car-wash test — model maturity counterpart), AM-041 (procurement playbook), AM-113 (vendor contract gotchas), AM-192 (ISO 42001 checkpoint), the McKinsey 23% read.

Read article →

+87dnext review
Holding

AM-209 · pub 9 Jun 2026 · rev 9 Jun 2026

Anchored on the FIS press release (4 May 2026, fisglobal.com canonical URL verified): Financial Crimes AI Agent built with Anthropic; AML alert/case investigations days-or-hours to minutes; BMO + Amalgamated Bank named in active development; GA H2 2026; carries the industry-estimated $35-40B US AML operations spend and the UN ~$2T illicit-flows figure (both present in the release, attributed there); full two-sentence Stephanie Ferris (CEO and President, FIS) quote used UNTRUNCATED per fact-check guidance. Second anchor: Microsoft UK story (4 Jun 2026, ukstories.microsoft.com — Lloyds rolls out M365 Frontier Suite/E7): 40,000 Copilot licences, 97% of licensed colleagues active, more than 10,000 engineers on GitHub Copilot (June 2026 figures only; the older ~5,000-engineer Oct 2025 figure and the separate 21M-mobile-customers figure deliberately NOT used per fact-check precision warnings). BCG retail-banking figures from scouting deliberately omitted (root-verify confidence only). Carries an explicit production-model disclosure line (the piece analyses an Anthropic deployment; Claude writes the publication). VERIFIED 2026-06-09 by hostile fact-check. 90-day cadence, set to land after the FIS agent's planned H2 2026 GA begins. Triggers: (1) the FIS agent misses GA or the named banks step back; (2) a regulator objects to the decision-preserving configuration; (3) a documented production deployment moves the filing decision itself to the agent, falsifying the pattern half. Siblings: AM-202 (Microsoft 365 E7 — Lloyds runs it), AM-185 (frontier labs as integrators), the Wall Street agents cross-industry read, the McKinsey 23% scaling-gap read.

Read article →

+87dnext review
Holding

AM-208 · pub 9 Jun 2026 · rev 9 Jun 2026

Anchored on the SpaceX S-1 (filed 20 May 2026, SEC reg 333-296070; amendments 1 Jun and 3 Jun; $135/share, ~$75B raise, ~$1.75T target, first trade targeted 12 Jun 2026 on Nasdaq) and the 5 Jun 2026 Rule 433 free writing prospectus. AI-segment figures (PRECISION: the segment combines xAI + X + AI data centers post the Feb 2026 merger — stated as such in the body, NOT pure-xAI): 2025 revenue $3.201B / operating loss $6.355B; Q1 2026 revenue $818M / operating loss $2.469B. Leases (DIRECTION VERIFIED — money flows INTO xAI): Anthropic ~$1.25B/month for ~325,000 NVIDIA GPUs (Colossus, through May 2029, 90-day cancellation; disclosed in the S-1; TechCrunch + Axios 20 May 2026); Google ~$920M/month for ~110,000 GPUs (Oct 2026–Jun 2029 with ramp through Sep 2026, 90-day cancellation active after 31 Dec 2026; disclosed via FWP; TechCrunch + CNBC 5 Jun 2026, with the verbatim Google Cloud spokesperson Gemini-Enterprise bridge-capacity quote). The ~$26B/year total is arithmetic on the two monthly figures, labeled as such. VERIFIED 2026-06-09 by hostile fact-check (the original scout framing had the payment direction INVERTED — corrected before drafting; logged here per no-silent-fixes). 45-day cadence: IPO completes 12 Jun and the Google ramp dates land in the window. Triggers: (1) either lease cancelled, renegotiated or materially expanded; (2) post-IPO disclosures contradicting the S-1 segment picture; (3) capacity loosening (falling lease pricing, idle-capacity disclosures) falsifying the supply-constraint reading. Siblings: AM-203 (Anthropic valuation), AM-191 (Big Four concentration), the Maia-chip infrastructure read, the AI energy read.

Read article →

+42dnext review
Holding

OPS-096 · pub 8 Jun 2026 · rev 8 Jun 2026

Anchored on Google's I/O 2026 subscriptions announcement (19 May 2026, blog.google/products-and-platforms/products/google-one/google-ai-subscriptions/, fetched directly): 'We're launching a $100/month AI Ultra plan' and 'reducing the monthly price of our top-tier AI Ultra plan from $250 to $200'; both Ultra tiers include full YouTube Premium individual, AI Pro includes YouTube Premium Lite; models Gemini Omni + Gemini 3.5 Flash. Pichai quote from blog.google/innovation-and-ai/sundar-pichai-io-2026/ (19 May 2026). Gemini 3.5 Flash API $1.50 input / $9.00 output per M tokens (Google AI pricing docs, corroborated devtk.ai verified 24 May 2026). VERIFIED 2026-06-08 via direct WebFetch of the blog.google subscriptions page. PRECISION: the canonical Google blog states '$100' and '$200/$250' (NOT '$99.99/$199.99/$249.99' — those rounded forms come from secondaries); the AI Pro exact monthly price is NOT stated on the subscriptions page, so the body deliberately does NOT assert a Pro dollar figure (the widely-cited '$19.99' is secondary-only) and refers to 'the cheaper AI Pro tier'. Operator-register advisory; claim is the most-operators-should-stay-cheaper stance, NOT a recommendation to buy Ultra. 30-day cadence, consumer AI pricing moves fast. Triggers: (1) Google changing AI Pro pricing/limits in a way that narrows the gap; (2) the $100 Ultra's usage limits or model access changing materially; (3) the Gemini API price moving. Siblings: OPS-003 (Claude Pro vs ChatGPT Plus), the anthropic-vs-openai-vs-gemini SMB read, the bootstrapped-SaaS AI cost read.

Read article →

+26dnext review
Holding

OPS-095 · pub 8 Jun 2026 · rev 8 Jun 2026

Anchored on OpenAI's 'Codex for every role, tool, and workflow' announcement (2 Jun 2026, openai.com/index/codex-for-every-role-tool-workflow/): six plugins (Sales, Data Analytics, Creative Production, Product Design, Public Equity Investing, Investment Banking); Codex on Free(limited)/Go/Plus $20/Pro/Business/Enterprise; role plugins + Codex Sites rolling out to Business and Enterprise first; Denise Dresser (CRO, OpenAI) quote via TechCrunch 2 Jun 2026. VERIFIED 2026-06-08: openai.com 403s to crawlers, slug confirmed via TechCrunch + 9to5Mac; plugin names + Plus $20 + Business/Enterprise-first gating + Sales (Salesforce/HubSpot/Slack) + Creative (Figma/Canva/Shutterstock/Picsart) corroborated; Dresser quote confirmed in TechCrunch (fetched). PRECISION: the '62 apps / 110 skills' figure is OpenAI's own (originates in the 403'd blog body, one secondary corroboration) — attributed as 'per OpenAI' in the body, not asserted as independently verified; additional Sales/Creative integrations (Outreach/Clay/Rox/Actively/Fal) deliberately left out as M-confidence. Operator-register advisory; claim is the use-Codex-now/watch-the-plugins stance for Plus teams, NOT an endorsement of upgrading to Business. 30-day cadence, tier availability still moving. Triggers: (1) the role plugins reaching Plus; (2) OpenAI changing Codex tier availability or Plus pricing; (3) a competitor shipping comparable role bundles at the $20 tier. Siblings: OPS (AI cold-sales), the anthropic-vs-openai-vs-gemini SMB read, AI client proposals.

Read article →

+26dnext review
Holding

OPS-094 · pub 8 Jun 2026 · rev 8 Jun 2026

Anchored on the Canva + Perplexity joint announcement (4 Jun 2026, canva.com/newsroom/news/perplexity/): Perplexity Computer builds editable Canva assets from research; available to Perplexity Pro/Max/Enterprise Pro/Enterprise Max; 11 languages; Emily Jorgens (Head of Business Development & Partnerships, Perplexity) quote. Perplexity Pro price $20/month from perplexity.ai/pricing. VERIFIED 2026-06-08: Canva newsroom URL 403s to crawlers, slug confirmed via 9to5Mac + Storyboard18 reproductions; tiers + 11 languages + Jorgens quote corroborated across multiple syndications; Perplexity Pro $20/mo confirmed. PRECISION: Canva has NOT specified a plan requirement on its side — the body says so explicitly and does not assert one (one non-authoritative secondary inferred a Canva Pro/Teams requirement; not in the official announcement). Operator-register advisory; claim is the workflow-shortcut-for-Pro-subscribers stance + the starting-draft caveat, not an endorsement over ChatGPT/Claude/Gemini. 30-day cadence, SMB tooling moves fast. Triggers: (1) Perplexity changes Pro pricing or which tiers get the connector; (2) Canva adds a plan gate; (3) a competing assistant ships a comparable one-prompt research-to-design path. Siblings: OPS-003 (Claude Pro vs ChatGPT Plus), the solopreneur stack-consolidation read.

Read article →

+26dnext review
Holding

AM-207 · pub 8 Jun 2026 · rev 8 Jun 2026

Anchored on the White House executive order 'Promoting Advanced Artificial Intelligence Innovation and Security' (signed 2 Jun 2026; whitehouse.gov presidential-actions + fact sheet + Federal Register 2026-11415 published 5 Jun 2026): CISA binding operational directives within 30 days; Treasury with NSA and CISA AI cybersecurity clearinghouse within 30 days; voluntary frontier-model pre-release access up to 30 days; OPM cyber-specialist hiring within 60 days. Named quote: Nick Andersen, Acting Director CISA, AFCEA TechNet Cyber 4 Jun 2026 (Cybersecurity Dive). EU AI Act contrast (fines up to 7% of global turnover) cited via artificialintelligenceact.eu. VERIFIED 2026-06-08 via whitehouse.gov EO + fact sheet (fetched). PRECISION: the frontier-model 30-day pre-release access is VOLUNTARY ('may'), not a submission mandate — stated as such throughout the body. Distinct from AM-197 (us-ai-regulation-federal-state-standoff, the landscape) — this piece is the specific 2 Jun EO's operational requirements + the de-facto-baseline argument. 45-day cadence (the 30/60-day deadlines land within the window). Triggers: (1) CISA's binding operational directives arriving materially narrower or broader than the order implies; (2) the labs declining the voluntary review window, neutralising that mechanism; (3) subsequent federal action converting a voluntary element into a mandate. Siblings: AM-197 (US regulation standoff), AM-184 (EU AI Act Digital Omnibus), AM-195 (AI coding agents attack surface).

Read article →

+41dnext review
Holding

AM-206 · pub 8 Jun 2026 · rev 8 Jun 2026

Anchored on three primary moves: Zscaler intent to acquire Symmetry Systems (21 May 2026, access graph mapping which identities access which data; Jay Chaudhry Chairman/CEO quote; zscaler.com/press/ai-announcement + ir.zscaler.com + GlobeNewswire); Snowflake intent to acquire Natoma (27 May 2026, enterprise MCP gateway enforcing identity/policy/audit at the tool-call level; Sridhar Ramaswamy CEO quote; BusinessWire 20260527677399); Microsoft Agent 365 GA (1 May 2026, $15/user/month agent control plane with unified registry + shadow-agent detection; microsoft.com/en-us/security/blog/2026/05/01/...). VERIFIED 2026-06-08 via Zscaler press page, Snowflake/BusinessWire release, Microsoft Security Blog. FRAMING NOTE: Agent 365 is a governance/control-plane product that includes security capabilities, framed as such not as a pure security product; no named Microsoft exec quote on the GA was located, so the piece does not attribute one. Deal terms for both acquisitions undisclosed; both are intent-to-acquire (not closed). 90-day cadence, market dynamics. Triggers: (1) the announced deals fail to close or fail to integrate the access-graph capability; (2) a competing layer (e.g. runtime network controls) proves to be what incidents/audits turn on; (3) the access graph becomes a commodity feature, weakening it as a deciding factor. Siblings: AM-204 (NHI governance vacuum), AM-203 (Anthropic valuation / vendor lock-in), AM-176 (Okta vs specialist NHI vendors), AM-191 (Big Four model concentration).

Read article →

+86dnext review
Holding

AM-205 · pub 8 Jun 2026 · rev 8 Jun 2026

Anchored on the CSA + Token Security survey 'Autonomous but Not Controlled: AI Agent Incidents Now Common in Enterprises' (published 21 Apr 2026, n=418 IT/security professionals, fielded Jan 2026): 82% found at least one unknown AI agent in the past year, 68% believe they have strong visibility, 65% had an AI agent security incident, of those incidents 61% involved data exposure, 21% run a formal decommissioning process; Itamar Apelblat (CEO/Co-Founder, Token Security) quote. Scale anchored on CrowdStrike press release (23 Mar 2026): 1,800+ distinct AI applications across ~160 million unique instances, Shadow AI Discovery for Endpoint capability; Michael Sentonas (President) quote. VERIFIED 2026-06-08 via cloudsecurityalliance.org press release/artifact and crowdstrike.com press release. PRECISION: the 61% data-exposure figure is of the incident group (65%), not all 418 respondents — stated as such in the body. Distinct from AM-204 (NHI governance vacuum — machine-identity scale/lifecycle) and AM-168/shadow-ai-discovery-playbook (broad shadow AI): this piece's core is the believed-vs-actual visibility gap and discovery-as-first-control. 90-day cadence. Triggers: (1) a later large-sample survey showing find-rate and believed-visibility converging; (2) discovery tooling becoming a default in major endpoint/cloud platforms; (3) incident data showing policy maturity, not discovery, separates breached from unbreached. Siblings: AM-204 (NHI governance vacuum), the shadow-AI discovery playbook, approved-tool-unapproved-capability.

Read article →

+86dnext review
Holding

OPS-061 · pub 7 May 2026 · rev 10 Jun 2026

Operators register pillar piece. Anchor for the entire register — every operator-section piece can link 'see the delegation framework here.' Cadence 45-day (operators tooling cadence + framework refinement). Trigger conditions: significant capability shift in any of the six 'AI works' classes (e.g. multi-step reliability moving above 60%); regulatory clarification on disclosure obligations under EU AI Act Article 50; emerging consensus across operator-cohort surveys on the six 'AI fails' classes. Sister claims: OPS-024 (when not to use AI for small business — inverse), OPS-009 (picking first AI agent), OPS-018 (1-page AI policy).

Read article →

+43dnext review
Holding

OPS-060 · pub 7 May 2026 · rev 10 Jun 2026

Dutch SMB e-commerce piece. NL-language hook (Dutch metaDescription with English body) targeting AVG-aware searchers. Cadence 45-day (EU AI Act Article 50 enforcement window approaching; bol.com policy shifts move quarterly). Trigger conditions: Bol.com seller policy changes on AI imagery or listing-quality scoring; Shopify Magic feature shifts; AVG enforcement actions by Autoriteit Persoonsgegevens against Dutch SMBs deploying AI; new Dutch hosting providers reaching production-scale data-residency posture. Sister claims: OPS-019 (NL bookkeeping Moneybird/eBoekhouden/Exact), OPS-052 (ZZP AI displacement). External sources: bol.com/verkopen, shopify.com/magic, autoriteitpersoonsgegevens.nl, eur-lex EU AI Act Article 50.

Read article →

+43dnext review
Holding

OPS-059 · pub 7 May 2026 · rev 10 Jun 2026

SMB-specific procurement piece. Lead-magnet pair with RES-005 (AI MSA Red-Team Checklist downloadable). Cadence 45-day (vendor MSA template revisions + EU AI Act enforcement guidance both move at this cadence). Trigger conditions: major AI vendor MSA template revision (Microsoft, OpenAI, Anthropic, Google all publish public MSAs); new EU AI Act Article 16 implementation guidance affecting deployer obligations; SMB-tier consumer protection rulings on auto-renewal / data-portability. Sister claims: AM-145 (enterprise-tier exit clauses), OPS-014 (vendor due diligence small business), RES-005 (MSA red-team checklist).

Read article →

+43dnext review
Holding

OPS-058 · pub 7 May 2026 · rev 10 Jun 2026

Voice-agent emerging-category piece. Low competition relative to query-volume growth in 2025-2026 (Twilio + LLM streaming, sub-500ms latency). Cadence 45-day (regulatory volatility + emerging-vendor space). Trigger conditions: FCC TCPA enforcement guidance on AI-voice; new platform entrants reaching production maturity (e.g. ElevenLabs Conversational AI scale); pricing model shifts (currently per-minute usage); voice-cloning IP litigation outcomes. Sister claims: OPS-007 (solo founder customer-service stack), OPS-029 (local service business appointment-driven). External sources: vapi.ai pricing, bland.ai pricing, retellai.com, fcc.gov TCPA, eur-lex GDPR Article 22.

Read article →

+43dnext review
Holding

OPS-057 · pub 7 May 2026 · rev 10 Jun 2026

Etsy SMB segment piece. Largest single underserved cohort in the marketplace cluster. Cadence 45-day (operators register default + tooling-volatility signal). Trigger conditions: Etsy policy changes on AI imagery or seller categorisation; new pricing tiers from Claude Pro / ChatGPT Plus / Midjourney; Etsy-specific tool consolidation (eRank-Sale Samurai-Alura competitive shifts); EU AI Act Article 50 enforcement guidance affecting marketplace sellers. Sister claims: OPS-022 (marketplace resellers Etsy/Marktplaats/Vinted), OPS-037 (Claude Pro vs ChatGPT Plus). External sources: etsy.com/legal/sellers, claude.ai/pricing, midjourney.com pricing, helpwithselling.com Etsy IR data.

Read article →

+43dnext review
Holding

AM-145 · pub 7 May 2026 · rev 7 May 2026

Procurement-anxiety lead-magnet piece, paired with RES-005 (AI MSA Red-Team Checklist downloadable). Cadence 60-day. Trigger conditions: further major AI vendor M&A activity changing change-of-control clauses; new EU AI Act enforcement guidance on Article 16 audit-evidence retention; first major published settlement under model-deprecation clauses; SaaS-industry MSA template revisions. Sister claims: AM-027 (vendor contract gotchas), AM-085 (RFP 60 questions), RES-005 (MSA red-team). External sources cited inline: ServiceNow Q1 2026 10-Q, Automation Anywhere Aisera press, Klarna Bloomberg reversal, OpenAI deprecation page, Anthropic model lifecycle, EU AI Act Articles 12 and 16.

Read article →

+24dnext review
Holding

AM-144 · pub 7 May 2026 · rev 7 May 2026

Microsoft-stack reader piece. No prior dedicated coverage of M365 Copilot Agent Mode in the corpus despite ~60% of target audience running Microsoft-heavy stacks. Cadence 60-day. Trigger conditions: Microsoft expanding Foundry Agent Service capabilities materially; pricing changes to Copilot for M365 (currently $30/user/month); BAA/data-residency posture shifts; new MCP-server integration that closes the multi-vendor gap. Sister claims: AM-141 (Agent Mode pillar), AM-003 (GPT-5 Pro), AM-125 (ServiceNow Now Assist). Inline sources: Microsoft 365 Copilot pricing page, MS Learn agents documentation, EU Data Boundary.

Read article →

+24dnext review
Holding

AM-143 · pub 7 May 2026 · rev 7 May 2026

Procurement-compliance pillar. Captures 'ai bom' / 'aibom' search volume currently at GSC positions 80-95 (~21 imp/month combined). Cadence 60-day. Trigger conditions: EU AI Act 2 August 2026 enforcement window opening; NIST AI RMF v2 publication; CycloneDX-AI specification milestones; first major enterprise AI vendor publishing an audit-ready AI BOM that resets vendor expectations. Sister claims: AM-018 (EU AI Act compliance scope), AM-038 (Article 12 audit evidence), AM-064 (supply-chain disclosure existing piece). External sources cited inline: artificialintelligenceact.eu, EO 14028 archives, NIST AI RMF site, cyclonedx.org, Datenschutzkonferenz Muss-Liste.

Read article →

+24dnext review
Holding

AM-142 · pub 7 May 2026 · rev 7 May 2026

Site-wide pillar piece. Captures the 'ai agent vs ai assistant' query family (currently 16+4 imp/month at GSC positions 57-85). Cadence 60-day. Trigger conditions: foundation-model providers reframing categories (e.g. OpenAI redefining 'agent' more narrowly or broadly); new benchmark families that change the multi-step-reliability ceiling cited inline (CRMArena-Pro 35%, TheAgentCompany 30%); regulatory categorisations under EU AI Act guidance that distinguish or merge these classes. Sister claims: AM-141 (What is Agent Mode), AM-029 (Not holding since 10 Jun 2026 — its Stanford 12/88 figure failed primary-source verification), AM-122 (agent eval frameworks).

Read article →

+24dnext review
Holding

AM-141 · pub 7 May 2026 · rev 7 May 2026

Brand-collision pillar piece. Captures the ~358 imp/month 'agent mode' query family currently spread across Google positions 5-69 (per GSC 2026-05-06 28-day pull). Cadence 60-day. Trigger conditions: any of the four vendors materially changing what 'Agent Mode' includes in their product; new entrant adopting the same brand label; pricing-model shift on M365 Copilot or Cursor that changes the procurement read. Sister claims: AM-003 (GPT-5 Pro pricing-tier signal), AM-125 (ServiceNow Now Assist procurement). Sources cited inline include Microsoft 365 Copilot agents page, Cursor.com pricing, GitHub Copilot features, OpenAI Agents page.

Read article →

+24dnext review
Holding

OPS-056 · pub 5 May 2026 · rev 5 May 2026

Cohort: bootstrapped SaaS founder under €30K MRR with AI features in production. Cadence 60-day. Trigger conditions: foundation-model provider releasing tier that materially changes cost-per-user math; aggregate published data on bootstrapped SaaS AI cost-per-user that benchmarks the 30-40% threshold; regulatory or platform-economic changes altering vendor cost structure. Indicative pricing figures marked source:our-estimate. Sister claims: AM-136 (foundation-model uptime + pricing), OPS-014 (AI vendor due diligence). Note: published as OPS-056 (not OPS-051 as originally drafted) because OPS-051 was already taken by an unrelated AI client-proposals claim.

Read article →

+22dnext review
Holding

OPS-055 · pub 5 May 2026 · rev 5 May 2026

DE-specific sister to OPS-045 NL. Cohort: Kleinunternehmer, Einzelunternehmer, GmbH-Geschäftsführer, freiberuflich. Cadence 60-day. Trigger conditions: vendor product changes at DATEV, sevDesk, or Lexware affecting the AI-bookkeeping integration; ELSTER-side changes affecting USt-Voranmeldung filing flows; Bundesfinanzministerium guidance on AI-assisted bookkeeping under GoBD. Sister claims: OPS-045 (NL bookkeeping), OPS-031 (jurisdiction-neutral DIY case), OPS-049 (Mittelstand BetrVG/DSGVO deployment), OPS-037 (VAT). Note: published as OPS-055 (not OPS-050 as originally drafted) because OPS-050 was already taken by an unrelated local-SEO claim.

Read article →

+22dnext review
Holding

OPS-054 · pub 5 May 2026 · rev 5 May 2026

Reframed from saturated Cursor-vs-Copilot comparison to the EU client-code-residency angle that the saturated category misses. Cohort: solo developer / freelance contractor doing EU client work, particularly with regulated-sector clients (financial services under DORA, healthcare under EHDS, public sector, legal). Cadence 60-day. Trigger conditions: AI Office or national supervisory authority publication of specific guidance on AI coding tools and EU residency; landmark enforcement action establishing GDPR Article 28 precedent for AI-tool processing chains; product-tier changes at GitHub Copilot, Cursor, or Claude Code that materially change residency-configuration landscape. Sister claims: OPS-014 (vendor due diligence), OPS-052 (NL solo legal — parallel professional-services case), OPS-056 (bootstrapped SaaS cost discipline).

Read article →

+22dnext review
Holding

OPS-053 · pub 5 May 2026 · rev 5 May 2026

Image-specific sister to OPS-046 (listing-copy workflow). Marktplaats-led ordering (NL/EU underserved cohort) per competitive-landscape review. Cohort: marketplace resellers across Etsy crafts, Marktplaats secondhand, Vinted clothing. Cadence 60-day. Trigger conditions: platform-level policy change at Marktplaats, Vinted, or Etsy on AI imagery; EU AI Act Article 50 transparency obligations imposing watermarking requirements on marketplace listings; industry-standards convergence on AI-image provenance signals (C2PA Content Credentials adoption at platform level). Sister claims: OPS-046 (listing-copy workflow), OPS-041 (broader platform-algorithm-AI-content piece).

Read article →

+22dnext review
Holding

OPS-052 · pub 5 May 2026 · rev 5 May 2026

NL-only solo legal piece (narrowed from tri-jurisdiction NL/DE/FR scope per buyer-journey expert review — DE/FR become future sister pieces). In Dutch (register-aware scanner pass). Cadence 60-day pegged to EU AI Act Artikel 50 enforcement. Trigger conditions: NOvA-Hof van Discipline uitspraak markerend specifieke AI-werkstroom als wel/niet toegestaan; EU AI Act Artikel 50 enforcement-action tegen NL advocatenkantoor; published NL-tuchtrechtelijke klacht over AI-gegenereerde citaten zonder verificatie (Mata v. Avianca-equivalent). Sister claims: OPS-014 (vendor due diligence), OPS-035 (when not to use AI), RES-002 (DPIA template), AM-135 (Article 50).

Read article →

+53dnext review
Holding

AM-139 · pub 5 May 2026 · rev 5 May 2026

Bridge piece between AM-* enterprise register and OPS-* operator register. Replaces the earlier abstract asymmetric-instrument framing that the four-expert review cut. Concrete proof points already in corpus: AM-128 (MIT 95% misread), AM-130 (2024-2025 retrospective with four classes of evidence), AM-138 (post-enforcement MSA carrying the asymmetric-instrument insert), OPS-051 (cancellation-trigger discipline), OPS-052 (solo-legal cross-cohort pattern), OPS-014 (vendor due diligence). Cadence 60-day. Trigger conditions: industry-wide convention on case-study formatting with operational-substrate disclosure; procurement-buyer industry-wide convention on case-study verification protocols; regulatory development imposing case-study substantiation requirements on AI vendors.

Read article →

+53dnext review
Holding

AM-138 · pub 5 May 2026 · rev 5 May 2026

Companion to RES-005 update v1.1 scheduled August-September 2026 after the first 30-60 days of enforcement reveal which clause additions are operationally load-bearing. Cadence 60-day with first review immediately after enforcement window opens. Carries the asymmetric-instrument bridge insert that was originally a standalone bridge piece. Sister claims: AM-135 (Article 50), AM-136 (uptime), AM-026 (60-question RFP), AM-046 (Article 12 audit substrate), RES-005 (MSA red-team checklist), AM-139 (vendor case-study citation chain bridge).

Read article →

+53dnext review
Holding

AM-137 · pub 5 May 2026 · rev 5 May 2026

Cluster-gap piece between AM-122 (eval-tooling decision) and the MTTD-for-Agents framework. Cadence 60-day. Trigger conditions: foundation-model provider releasing a model that materially changes the eval-set noise floor; OpenTelemetry GenAI semantic conventions extending to evaluation events with industry-standards-grade adoption; landmark customer incident attributable to evaluation-discipline failure with published learning. Sister claims: AM-122 (eval procurement), AM-123 (observability), AM-126 (red-team), MTTD-for-Agents framework.

Read article →

+22dnext review
Holding

AM-136 · pub 5 May 2026 · rev 30 May 2026

30-day Holding-up cadence — the operational record ages monthly, not quarterly. Trigger conditions: a foundation-model provider publishing sustained 99.99% across 12 consecutive months; regulatory development requiring multi-provider provisioning for high-risk deployments; landmark vendor outage with material customer harm and follow-on litigation. Indicative procurement-fee figures in the body marked source:our-estimate. Sister claims: AM-026 (60-question RFP), AM-138 (post-enforcement MSA), AM-123 (observability), RES-005 (MSA red-team checklist).

Read article →

+17dnext review
Holding

AM-135 · pub 5 May 2026 · rev 5 May 2026

Deadline-anchored claim pegged to 2 August 2026 enforcement window. Cadence 60-day with the first review immediately after enforcement opens (4 Aug 2026). Trigger conditions: AI Office detailed implementing guidance on Article 50 with named UX patterns endorsed or rejected; national supervisory authority enforcement actions in the first 12 months; industry-standards convergence on Article 50(2) marking for text-based generative AI. Sister claims: AM-046 (Article 12 audit substrate), AM-138 (post-enforcement MSA red-team), OPS-052 (NL solo legal Article 50 application).

Read article →

+53dnext review
Holding

AM-134 · pub 5 May 2026 · rev 5 May 2026

Implementation extension of AM-029 (the conceptual case for NHI for AI agents). Decision-tree piece mapping deployment topology to control plane. Cadence 60-day. Trigger conditions: material vendor product release across the six options; EU AI Act / NIST AI RMF / ISO 42001 regulatory development imposing specific NHI requirements; industry-standards convergence on cross-platform NHI federation. Sister claims: AM-029, AM-126 (red-team), AM-043 (OWASP Agentic Top 10), AM-046 (Article 12 audit substrate).

Read article →

+22dnext review
Partial

AM-132 · pub 5 May 2026 · rev 10 Jun 2026

Re-review 10 Jun 2026: the Stanford DEL leg failed primary-source verification — the playbook contains no 12/88 ROI distribution (see correction below and AM-029). The Gartner Q1 2026 (28% fully paying off), McKinsey State of AI 2025 (23% scaling, 17% EBIT-attribution, n=1,993), and MIT NANDA (95% no measurable P&L impact) legs verify and support a small high-performing tail with a large struggling body; no verified source documents a two-peak bimodal distribution. History: URL-equity restoration of /why-73-of-agentic-ai-projects-fail-and-how-the-27-generate-312-roi/ — previously retired with claim AM-014 status:down on 28 Apr 2026 because the original WordPress-era body used composite case studies that did not survive editorial-standard scrutiny; restored 5 May 2026 per Peter's Option A decision 2026-05-04, slug warning (clickbait metric '312-roi') accepted as the AI-citation preservation trade-off. Original AM-014 claim remains status:down per Holding-up rule that retracted claims do not return. Sister claims: AM-029 (Not holding 10 Jun 2026 — Stanford 12/88 attribution failed verification), AM-053 (McKinsey 17%), AM-128 (MIT 95%), AM-129 (mid-market ROI), AM-130 (2024-2025 retrospective), AM-131 (AI Training Lead). Follow-up review after the article body is restated on the three verified legs. Trigger conditions: new Stanford DEL update with refreshed deployment cohort; Gartner I&O Q3/Q4 2026 update; McKinsey State of AI 2026 mid-year refresh; MIT NANDA follow-up.

Read article →

+33dnext review
Holding

AM-131 · pub 5 May 2026 · rev 5 May 2026

URL-equity restoration of /from-it-pro-to-ai-training-lead-the-180k-career-path-nobodys-talking-about/ — previously retired, but Bing Webmaster AI Performance data 2026-04-21 → 2026-05-02 showed continued AI-citation activity on the URL across the GSC follow-up window. The retraction broke the citation chain for the 'AI hiring playbook for CIOs' query family. The piece had already been rewritten 27 Apr 2026 from the original careers/personal-development register to a CIO hiring/budget playbook anchored to Stanford HAI 2026 AI Index, WEF Future of Jobs Report 2025, and BLS occupational data, but never moved out of content/archived/. New editorial-standard piece at the original slug preserves the URL while replacing the original $180K-career-path framing with stat-anchored hiring guidance. Slug warning (clickbait specificity '180k career path nobodys talking about') is accepted as the intentional AI-citation preservation trade-off per Peter's Option A decision 2026-05-04. Sister claims: AM-129 (mid-market ROI), AM-130 (2024-2025 retrospective). Cadence 60-day. Trigger conditions: published Stanford AI Index, McKinsey State of AI, or WEF Future of Jobs update with explicit breakouts on the AI evaluation/training role; published case study from a named enterprise comparing outcomes between domain-expert vs pure-ML evaluation leads; vendor (Anthropic, OpenAI, Microsoft, Databricks) shipping evaluation-as-a-service that changes the build-vs-buy calculus on the role; EU AI Act or comparable regulatory development specifying qualifications for human oversight of high-risk agent deployments.

Read article →

+22dnext review
Holding

OPS-093 · pub 5 Jun 2026 · rev 5 Jun 2026

Anchored on Microsoft's 2 Dec 2025 Microsoft 365 blog (Copilot Business GA, $21/user/month annual commitment, up to 300 seats, Jared Spataro CMO-AI-at-Work quote naming Word/Excel/PowerPoint/Outlook/Teams) and Microsoft's Tech Community deadline notice (existing M365 Business customers hold a promotional $18/user/month through 30 Jun 2026, first year). VERIFIED 2026-06-05 via microsoft.com/en-us/microsoft-365/blog (2 Dec 2025 GA post, $21 standard, 300-seat cap, Spataro quote) and techcommunity.microsoft.com (30 Jun 2026 deadline notice). Precision caveat carried in the body: the $18 standalone add-on offer is distinct from an earlier, now-expired Dec 2025 to 31 Mar 2026 bundle promotion (do not conflate). The $36/year/seat figure is arithmetic on the cited $3/user/month difference, not a Microsoft statistic. Operator-register advisory; the claim is decide-before-the-deadline-but-on-usage, NOT a blanket endorsement of Copilot. 30-day cadence, set to land just after the deadline. Triggers: (1) Microsoft extends the $18 rate past 30 Jun 2026; (2) Microsoft changes the $21 standard price or the 300-seat cap; (3) a packaging change folds Copilot into the base Business plans (as E7 did for enterprise tiers). Siblings: OPS-003 (Claude Pro vs ChatGPT Plus for a solo founder); the solopreneur stack-consolidation read; AM-202 (Microsoft 365 E7 enterprise tier read).

Read article →

+23dnext review
Holding

OPS-092 · pub 5 Jun 2026 · rev 5 Jun 2026

Anchored on Notion's 4 May 2026 beta-wrap blog post (Custom Agents left free beta, usage-based pricing went live, Akshay Kothari over-1-million-agents quote), Notion's Custom Agents help page (credits $10 per 1,000 monthly, no rollover, Business or Enterprise plan required, no published per-action rate, billing starts each workspace's first billing date on or after 4 May 2026), and Notion's 24 Feb 2026 release notes (Custom Agents launch). Business plan $15/user/month figure is the annual-billing rate (monthly billing is higher; stated as annual in the body). VERIFIED 2026-06-05 via notion.com/help/custom-agent-pricing, notion.com/blog/what-we-learned-during-the-custom-agents-beta, notion.com/releases/2026-02-24, corroborated by TechCrunch 13 May 2026. Per-run dollar range deliberately omitted from the body: Notion does not publish a per-action rate, so the piece stays qualitative (run-frequency framing) rather than citing an analyst-derived range as if it were vendor-stated. 30-day cadence, SMB AI pricing moves fast. Triggers: (1) Notion changes the $10-per-1,000 rate or introduces rollover; (2) Notion publishes a clearer per-action cost making budgeting precise; (3) a bundled credit allowance lands inside the Business plan, changing break-even. Siblings: OPS-077 (no-code agent building), the Notion agents hub operator-upgrade read.

Read article →

+23dnext review
Holding

OPS-091 · pub 5 Jun 2026 · rev 5 Jun 2026

Anchored on HubSpot's 14 Apr 2026 company-news announcement (Customer Agent $0.50 per resolved conversation down from $1.00 per conversation; Prospecting Agent $1.00 per lead recommended for outreach; 28-day free trial; Pro and Enterprise plans) and HubSpot's Customer Agent knowledge-base article (the resolved definition: content source shared or action taken, no human handoff within 72 hours of last message, resolution locked at 72 hours, reopen after 72 hours resets the window). VERIFIED 2026-06-05 via hubspot.com/company-news (pricing, trial, plan requirement, Jon Dick CCO quote) and knowledge.hubspot.com/customer-agent/understand-the-customer-agent (resolved definition). Operator-register advisory; the claim is the trial-it-and-measure-resolution-rate stance plus the read-the-definition caveat, NOT an endorsement of HubSpot over alternatives. The 200-conversations example in the body is illustrative arithmetic on the cited unit price, not a HubSpot statistic. 30-day cadence, SMB AI pricing moves fast. Triggers: (1) HubSpot raises the $0.50 rate or narrows the 72-hour resolution window; (2) HubSpot extends the agents below the Pro plan, widening the audience; (3) a competing help desk ships a comparable outcome-priced agent a small team on that platform should weigh instead. Siblings: OPS-043 (solo-founder customer-service stack); the agentic-AI cost-governance enterprise read.

Read article →

+23dnext review
Holding

AM-204 · pub 5 Jun 2026 · rev 5 Jun 2026

Anchored on three May 2026 datasets: CSA AI Safety Initiative whitepaper 'The Non-Human Identity Governance Vacuum' (20 May 2026 — 45:1 average / 144:1 cloud-native NHI-to-human ratio, 78% no policy to create/retire AI identities, 51% no clear ownership, 20% formal API-key offboarding); Sophos 'State of Identity Security 2026' (12 May 2026, n=5,000 across 17 countries — 71% suffered an identity breach in the past year, $1.64M mean recovery cost, weak NHI management a factor in 41% of incidents, CISO Ross McKerchar quote); Gartner 'Six Steps to Manage AI Agent Sprawl' press release (28 Apr 2026 — over 150,000 agents per average Fortune 500 firm by 2028 vs <15 in 2025, only 13% believe they have adequate agent governance, six-step sequence starting with governance policy then centralised inventory). VERIFIED 2026-06-05: CSA labs page, Sophos press release, Gartner newsroom press release (Gartner page 403s to crawlers; figures and URL confirmed via WebSearch returning the canonical newsroom URL verbatim). Distinct from CSA's earlier 26 Jan 2026 'State of NHI and AI Security' survey — the 45:1/78%/20%/51% figures are exclusively from the 20 May 2026 whitepaper. 90-day cadence. Triggers: (1) a later large-sample dataset showing the ratio or policy-gap figures compressing materially; (2) a standards or platform shift (e.g. broadly adopted workload-identity attestation) making lifecycle governance a default; (3) breach data showing perimeter controls, not inventory, separate affected from unaffected. Siblings: AM-167 (NHI procurement clause gap), AM-176 (Okta vs specialist NHI vendors), the agent-identity IAM architecture read.

Read article →

+83dnext review
Holding

AM-203 · pub 5 Jun 2026 · rev 5 Jun 2026

Anchored on Anthropic's 28 May 2026 Series H announcement ($65B raised, $965B post-money, run-rate revenue crossed $47B, leads Altimeter/Dragoneer/Greenoaks/Sequoia, $15B hyperscaler tranche incl $5B Amazon, CFO Krishna Rao quote) and OpenAI's most recent round ($122B raised, $852B post-money, closed 31 Mar 2026). VERIFIED 2026-06-05: anthropic.com/news/series-h (primary, all Anthropic figures + Rao quote); OpenAI valuation via OpenAI's own accelerating-the-next-phase-ai post ($122B) and CNBC 31 Mar 2026 ($852B post-money) — OpenAI does not publish valuations on-site, so the comparison rests on contemporaneous financial press. Both figures are private-market marks, not public valuations; framed as such in the body. The claim is the market-structure read (consolidation to a hyperscaler-backed top two) and its procurement consequence (negotiate exit, not benchmark), NOT a prediction of either vendor's IPO outcome. 90-day cadence, market dynamics. Triggers: (1) a credible third frontier vendor or open-weight option taking material enterprise share; (2) enterprise AI prices falling rather than holding, indicating pricing power did not materialise; (3) a vendor offering structurally cheaper enterprise terms that re-open price as the deciding variable. Siblings: AM-191 (Big Four model concentration), AM-185 (frontier labs as systems integrators), the Karpathy-to-Anthropic vendor-trajectory read.

Read article →

+83dnext review
Showing the 40 most recent. Full ledger also available as machine-readable JSON.

Each claim links to the piece it came from and the review cadence Peter set when publishing it. How this works →

Moving this week

5 claims have moved off Holding in the last 60 days. The full correction log is on the ledger.

See all corrections →
  1. AM-021PartialDMAIC for agentic AI deployment: why the 87% / 27% success gap reflects measurement discipline, not methodologyMoved 10 Jun 2026Read article →
  2. AM-024PartialThe unverified citation chain: where enterprise AI decisions actually come fromMoved 10 Jun 2026Read article →
  3. AM-029Not holdingWhy 88% of agentic AI deployments failMoved 10 Jun 2026Read article →
  4. AM-031PartialThe CMU 30.3%: the enterprise agent capability gapMoved 10 Jun 2026Read article →
  5. AM-035PartialThe EU AI Act and agentic AI: what August 2026 actually requiresMoved 10 Jun 2026Read article →
Audit trail · Retractions

What we've pulled

14 pieces retracted since launch. Pulled when the claim cannot be defended — fabricated sources, unfixable factual errors, claims that do not survive review. Nothing is quietly removed; every retraction stays on the public record below.

See all retractions →
  1. 28 Apr 2026
    Mid-market agentic AI: what the 50-500 employee bracket actually ships
    Reason: Slug carries an unsupportable 240% / 90-day specificity that the body could not defend without rewriting the headline. Body was rewritten on 27 Apr 2026 with primary-source sourcing (Stanford HAI, McKinsey State of AI, BCG generative-AI value research, Gartner Q1 2026 I&O), preserved in archived/. The slug itself is the structural problem and Google's quality algorithm independently flagged the URL as low-quality (crawled, not indexed) per the 28 Apr 2026 GSC drilldown. · Linked claim AM-058
  2. 28 Apr 2026
    Multi-agent document processing in pharma: the FDA-submission workflow that's actually shipping
    Reason: Slug attributes specific Moderna throughput numbers ("6 weeks to 6 hours", "50,000 records daily") that cannot be tied to any disclosed Moderna source. Body was rewritten on 27 Apr 2026 with FDA / Veeva / IQVIA / Moderna-OpenAI public-partnership citations (preserved in archived/) but the URL specificity is the structural problem under §6a, and Google has independently rejected the URL. · Linked claim AM-059
  3. 28 Apr 2026
    Why your agentic-AI deployment needs an AI Training Lead
    Reason: Slug carries clickbait salary specificity ("$180k career path nobody's talking about") that is off-thesis for the publication's CIO register. Body was reframed on 27 Apr 2026 as a CIO hiring playbook anchored to Stanford AI Index, WEF Future of Jobs, and BLS occupational data (preserved in archived/) but the slug cannot be reframed and Google has rejected the URL. · Linked claim AM-060
For external work

Cite the ledger

The ledger is structured for citation. Every claim has a stable ID, a permanent URL, and a public verdict log. Use any of the three forms below depending on how granular your reference needs to be.

  1. 1 · Inline reference to a single claim

    Use the claim ID with the as-of date. Reviewed claims keep their ID forever; the verdict line carries the time-sensitive part.

    Agent Mode AI, claim AM-024 (Holding as of 17 Apr 2026): https://agentmodeai.com/holding/?claim=AM-024
  2. 2 · Dataset citation (APA-shaped)

    Reference the ledger as a whole — for survey work, briefs, or academic citation of the corpus rather than a single argument.

    Agent Mode AI (2026). Holding-up: every published claim with verdict and review history [Dataset]. https://agentmodeai.com/holding/
  3. 3 · Permanent identifier proposed

    Publication-internal permanent identifier (not a real DOI; scheme proposed in the framework registry, final lock pending). Resolves to the ledger URL above.

    amai:holding-up:v1.0
Aggregate signal

How often do these claims still hold?

Holding rate, verdict distribution, cadence health, recent corrections, and upcoming reviews — the aggregate trust signal across every claim above.

View insights →
Affiliate firewall

Vendors we will and won't affiliate-link

The publication earns affiliate commission from a subset of vendors it covers. The rule: never affiliate-link a vendor whose tracked claim has been Partial or Not holding. The rule is enforced in code — the build fails if a blocked vendor link slips through. This panel is the public face of that firewall. See /disclosures/ for the full editorial framework.

Vendor affiliate-firewall status — each vendor with its current firewall verdict and the reason.
VendorStatusWhy
n8n
ai-tooling-subscription · operators
Eligible (no audited claim)No tracked claims about "n8n" yet.
Affiliate program not yet enrolled — no commission earned today.
Anthropic
ai-tooling-subscription · enterprise / operators
EligibleAll tracked claims about "Anthropic" are positive.
Affiliate program not yet enrolled — no commission earned today.