Changelog

15 October 2026UpdatedWhat AI Actually Costs: Reference Cost Ranges for Enterprise AI
Price ranges rechecked against Anthropic and Google price lists on 15 October 2026. The small-model range is now $0.50 to $5 per million output tokens, after Anthropic launched Haiku 5.5 at $0.50 on 7 October. Added a note on context-tier pricing, where long prompts are billed at five times the standard rate. Graphic updated. View
15 October 2026UpdatedWhy Cheaper AI Will Cost More
Anthropic's output price range corrected from $5 to $50 to $0.50 to $50 per million tokens. The position is unchanged and the ledger entry is reassessed. View
15 October 2026UpdatedInference cost estimator and self-hosting break-even calculator
Hints updated for current price lists. The uplift slider now reaches 400%, to cover long-context price tiers. No change to the calculations. View
15 October 2026UpdatedAutomation Share Is Not Value
Added Deloitte's Q3 2026 UK CFO Survey: 52% expect AI to reduce headcount within 12 months, set against a US survey in which 3% are cutting staff. View
13 October 2026CorrectedAI TCO Worksheet: The Seven-Sheet Model
The worked example deducted the AI cost twice, and priced rework at about £11 an hour instead of the £45 review rate. Recalculated: Sheet 3 is £6,688 a month, fully loaded cost £3.03 per ticket (12.3 times the naive £0.25), saving £15.47 per ticket, and net value £37,122.50 a month, not £31,476. The Sheet 6 forecast, sensitivities and variance trigger now follow from the inputs, the unquantified pilot-to-production multiplier is now described qualitatively, and the graphic is updated. View
12 October 2026UpdatedHomepage
The third headline figure now comes from a Gartner survey of finance leaders (57% of finance AI use cases with unclear returns), replacing BCG's 2025 figure that 5% of companies achieve AI value at scale. BCG's 2026 index reports a much higher share on a self-reported basis, so the 2025 figure was out of date. View
12 October 2026UpdatedEvidence register
Six entries added from sources published in late September and early October 2026: BCG Applied AI Index 2026, a Gartner survey of finance leaders, Datarails and Pigment CFO surveys (both vendor research), Microsoft's default usage-based billing for Copilot Business, and the State of AI Report 2026. View
12 October 2026UpdatedThe AI Value Gap
Added BCG's 2026 Applied AI Index and a Gartner survey showing positive portfolio returns alongside mostly unclear use-case returns. View
12 October 2026UpdatedPricing and budgeting articles
Added Microsoft's move to default usage-based billing for Copilot Business to the software seat, SaaS token opacity and manager spend articles, and finance-leader overrun data to How to Budget for AI. View
12 October 2026UpdatedVerification, portfolio and autonomy articles
Added survey evidence on time spent checking AI output, use-case level returns, the gap between planned agent autonomy and controls in place, and market price and revenue trends. View
12 October 2026UpdatedBrowser tools
All five tools now use sliders with exact-entry boxes, a side-by-side results panel and clearer charts. Their calculations were rechecked against each article's worked example, and edge cases such as zero capacity, zero hurdle or falling demand now give correct results and wording. View
12 October 2026AddedCost per successful outcome calculator
Browser-only calculator comparing up to four model setups on cost per successful outcome, including human fixes. Defaults reproduce the worked example in The Economics of Model Selection. View
12 October 2026AddedFully loaded inference cost estimator
Browser-only estimator that builds a monthly inference cost from the rate card up, adding output ratio, caching, uplifts and fees, following The Real Cost of AI Inference. View
12 October 2026AddedAI budget stress test
Browser-only tool testing whether an AI case still meets its hurdle under higher prices, higher demand, or both. View
12 October 2026AddedSelf-hosting break-even calculator
Browser-only calculator comparing pay-per-token pricing with running an open-weight model on dedicated GPUs, including utilisation and people costs. Stores and sends nothing. View
12 October 2026AddedHow Inference Choices Change AI Cost
A guide for budget owners on batching, serverless versus dedicated hosting, open-weight models, serving engines, quantisation and when self-hosting pays. View
8 October 2026UpdatedAuthor disclosure
The site now names its author, Ed Epstein, and his role at IBM in the Apptio portfolio, with IBM's standard disclaimer that the postings are his own and do not necessarily represent IBM's positions. The disclosure also notes that IBM sells products in areas the site covers and is a member of the Tokenomics Foundation. View
8 October 2026UpdatedVendor Map
All vendor ratings and maturity labels have been removed. Because the author works for IBM, the map now describes vendors from public sources and does not rate, score or rank them. IBM-owned products remain marked. View
8 October 2026UpdatedGlossary
Hardware, model and prompt terms rewritten around their economics: utilisation and commitment for accelerators, where training cost really falls, when fine-tuning pays, what RAG costs in input tokens, and why frontier models stay dear. TPU and ASIC are now part of GPU and AI Accelerators; Prompt Engineering is part of Prompt Efficiency. Old links redirect. View
8 October 2026CorrectedToken Economics
Corrected the billing mechanics: models bill tokens sent, not context-window size; output costs commonly four to six times input; cached reads cost 2.5% to 10% of input on one major list, with a write premium. Added reasoning tokens, context re-sent on every call, tokeniser differences and the Tokenomics Foundation's 4 August 2026 launch. The worked example now compares cost per query including escalation, not token counts. View
8 October 2026CorrectedToken (glossary)
The entry called the token the atomic unit of AI value, contradicting the site's own position. It now defines the token as a unit of consumption and pricing, not value. A Reasoning Tokens entry was added. View
8 October 2026AddedAutomation Share Is Not Value
New essay, version 0.1: a comment on the Tokenomics Foundation's proposed first value measure, the share of work done without human involvement, and the four measures it needs alongside it. View
8 October 2026UpdatedWhen to Stop an AI Initiative
Retitled from The AI Initiative Autopsy. The opening now reflects S&P Global's finding that 42% of companies abandoned most of their AI initiatives: stopping happens, but by attrition rather than by rule. An unsupported claim that Starbucks based its exit on a formal ROI analysis was removed. View
8 October 2026UpdatedMerged pieces
The Inference Cost Crisis is now part of Why Cheaper AI Will Cost More, restated as a risk to the structure of prices rather than a forecast of higher list prices. The First 30 Days is now the week-by-week first month of The First 100 Days. The distinction ladder is now part of How to Read an AI Metric. Old links redirect. View
8 October 2026UpdatedRetitled essays
FinOps for AI Is Necessary, Not Sufficient (was AI Value Management Is Not FinOps for AI); The Seat Is No Longer the Only Meter (was The End of the Software Seat); Proportionate AI Governance (was The Cost of AI Governance); What Agentic AI Changes in AI Economics; Where Power Sits in the AI Value Chain; The AI Economics Map (was The AI Economics System, to avoid a clash with the four-phase system). Arguments are unchanged; added evidence from EY, PwC, Gartner and DORA 2025. View
8 October 2026UpdatedSite navigation
Start Here is now a single link. The Debate and The Frontier are one Understand section, Open questions. About and Methodology moved to the footer and Start Here. The frameworks index shows which door each model sits in. View
8 October 2026UpdatedPosition ledger
Added P3, the token is the meter, not the value, assessed as holding and now close to consensus. View
8 October 2026UpdatedSite navigation
The site now has two doors. Understand holds the argument, the evidence and the models. Implement holds the system, toolkit, decision cases and operating model. Each has its own front page, header menu and section bar, and every page links to its counterpart in the other door. The header has four groups instead of six. No URL has changed. View
8 October 2026AddedThe AI Portfolio and Cost Register
New proposal, version 0.5: a register of initiatives, tools and agents with owners, and a cost register in the seven layers of the AI TCO Framework, as an Excel template. View
8 October 2026AddedThe AI Supplier Economics Review
New proposal, version 0.5: a review of a supplier contract on pricing risk, lock-in, consolidation exposure and governance fit, with a twelve-term checklist and an Excel template. View
8 October 2026AddedThe First 100 Days of AI Value Management
New proposal, version 0.5: a sequenced plan that takes the toolkit from an empty register to a first portfolio review, board pack and operating model in 100 days. View
8 October 2026UpdatedToolkit versions
The new toolkit assets are now labelled version 0.5, in line with the rule that the site's proposals stay below 1.0 until they have survived outside challenge. The initiative autopsy template was renamed the Stopped Initiative Review to avoid a clash with the essay When to Stop: The AI Initiative Autopsy. View
8 October 2026AddedThe AI Benefits Register
New proposal, version 0.5: a register that keeps realised, absorbed, released and avoided benefit apart, with an Excel template and an invented example that reconciles to the sample portfolio. View
8 October 2026AddedThe AI Investment Decision Memorandum
New proposal, version 0.5: a seven-part memorandum for a material funding request, as a Word template. View
8 October 2026AddedThe Stopped Initiative Review
New proposal, version 0.5: a blameless record of a stopped initiative, as a Word template. View
8 October 2026UpdatedAI Economics Maturity Model
Added a section scoring ten dimensions against the existing five levels, twelve readiness questions and an Excel assessment workbook. The five levels are unchanged. View
8 October 2026CorrectedThe AI Value Ledger
The template's illustrative row showed claimable value below total cost yet a Scale decision. Version 1.1 clarifies the total cost and net outcome columns, adds a column comparing claimable value with total cost, and changes the example so its decision follows from its figures. Version 1.0 remains available. View
8 October 2026AddedDecision cases
New /cases page and four illustrative cases, one for each decision: contract review assistant (scale), workforce assistant (hold), contact-centre summarisation (redesign) and customer-service agent pilot (stop). All figures are invented and reconcile with the sample portfolio in the portfolio review pack. Linked from the Library menu. View
8 October 2026AddedToolkit
New /toolkit page listing six working assets by phase of the system: the AI TCO worksheet, attribution coverage calculator, AI Value Ledger, and three new assets below. Added to the main menu and footer. View
8 October 2026AddedThe AI Portfolio Review Pack
New proposal, version 0.5: a review agenda, decision tests and an Excel workbook with an invented five-initiative portfolio showing claimed against realised value. View
8 October 2026AddedThe AI Board Reporting Pack
New proposal, version 0.5: an eight-page board pack, one-page dashboard, audit committee supplement, decision memorandum and question bank, as a Word template. View
8 October 2026AddedThe AI Value Office Charter
New proposal, version 0.5: a charter template covering mandate, scope, roles, decision rights and a monthly and quarterly cycle, as a Word template. View
8 October 2026AddedThe AI Economics System
New /system section: an overview and four phase pages (Measure, Prove, Decide, Operate) that organise the site's existing frameworks, tools and articles by the question each answers. The System menu replaces the Frameworks menu; all frameworks remain on the Frameworks page. The homepage gains a four-phase band. No existing URL has changed. View
8 October 2026UpdatedThe AI Value Ledger
Added a fully worked, illustrative entry: full monthly cost, claimable value after attribution and persistence, the gap between released and realised value, and the resulting Hold decision with a board commentary line. Article moves to version 1.1; the downloadable template is unchanged at version 1.0. View
8 October 2026AddedThe AI Value Lifecycle
New proposal, version 0.1: seven stages from funding to a scale, hold, redesign or stop decision, each with an owner and a gate, with a separate path for bought AI and evidence roles linked to vendor map categories rather than named products. View
8 October 2026UpdatedVendor map
Added ten entries: Planview, SAP AI Agent Hub, Microsoft Copilot Analytics, OpenAI Codex, IBM Bob, Promptfoo, TruLens, Ragas, and Celonis and UiPath Maestro in a new Process Outcomes & Benefits Realisation category. Renamed the IBM Apptio value entry to IBM Apptio AI Value & ROI and noted its public preview status. Arize now distinguishes Phoenix from Arize AX, Helicone shows its gateway role, and OpenAI covers ChatGPT workforce analytics and spend controls. Introduced parent companies, secondary categories, capability tags, spend control depth and value evidence, and a panel on enabling technologies such as semantic and prefix caching. Qualified the claim that value management lacks products. View
7 October 2026CorrectedWhat CFOs Should Ask of AI ROI Claims
Bain's 44% had been described as funding AI from savings that had not yet materialised. Bain says 44% fund it from savings of earlier automation programmes, a wave that mostly fell short of target; the finding was published in June, not April. The UK government Copilot experiment had been described as rigorous baseline measurement; it relied on self-reported savings with no control group, and is now described as such. IBM Institute for Business Value and APQC finance figures (8%, 18%, six to eight months) were removed because they could not be found at source. View
7 October 2026CorrectedWhat AI Actually Costs
Removed findings attributed to AWS, BCG and the FinOps Foundation that could not be traced, an unsourced example of Copilot subscription cancellations and an unsupported 10 to 25% hyperscaler platform margin. Rewrote the Starbucks example from press reports: the company retired an AI inventory counting tool after miscounts, not an agent overridden by managers. Marked the Uber per-engineer cost range as single-source. Replaced unlinked references with sourced ones. View
7 October 2026UpdatedThe AI Value Ledger
Decision options aligned with the downloadable template: scale, hold, redesign or stop. View
7 October 2026UpdatedThe AI Abundance Paradox
Pressure Index figure caption now credits the publication rather than an unpublished working document. View
7 October 2026ReviewedThe AI Abundance Paradox
House framework, v1.0. Historical citations (Simon, Goldratt, Jevons, Bainbridge) added to the evidence register; primary texts not re-read. Pressure Index caption corrected to remove an unpublished source. View
7 October 2026ReviewedWhat CFOs Should Ask of AI ROI Claims
Bain re-read at primary and wording corrected. UK government Copilot report read at primary and description corrected. IBM IBV and APQC finance figures removed: not found at source. McKinsey and Deloitte aligned to the register. WSJ cited by headline only; article body not read. View
7 October 2026ReviewedWhat AI Actually Costs
Full review. Unsourced attributions to AWS, BCG and the FinOps Foundation removed. Copilot cancellation example and hyperscaler margin figure removed. Starbucks example rewritten from secondary reports. Uber per-engineer range marked single-source. Ranges other than list prices are house estimates and labelled as such. View
7 October 2026ReviewedThe AI Value Ledger
Tokenomics Foundation mapping checked against the register (RC2). Decision set aligned with the downloadable template (scale, hold, redesign, stop). Template opened and checked. View
7 October 2026ReviewedWhat the Engineering Evidence Says About AI Coding Value
Peng, METR (both posts) and DORA 2024 figures re-read at primary, including the DORA adoption and METR selection figures not previously registered. View
7 October 2026ReviewedEvidence Review, Q3 2026
Every figure checked against the evidence register. FOCUS 1.4 commitment detail re-read at primary. Bain publication date corrected in the register to 1 June 2026. View
7 October 2026UpdatedVendor map
Rebuilt the vendor map around reader problems: a problem-first chooser, card and comparison-table views, vendor profiles with logos, homepage screenshots and sources, a dated timeline of market moves and six market signals for buyers. Logged four further ownership changes: Portkey (Palo Alto Networks), Cursor (SpaceX), Helicone (Mintlify) and Langfuse (ClickHouse). View
7 October 2026AddedLibrary
Seven new pieces: The AI Economics System (a single map from compute to capital allocation), How to Measure AI Value (the Tokenomics Foundation chain plus the steps to a decision, with three worked examples), Who Owns AI Economics (a federated operating model with a proposed RACI), How to Budget for AI, When Every Manager Can Create AI Spend, The Economics of Model Selection, and the AI value instrumentation blueprint. View
7 October 2026UpdatedFrameworks
Every model and tool now carries a status (hub proposal, adapted concept, reference explainer or tool), the job it does, what it is best for and not for, and its lineage. Hub proposals carry a version below 1.0 until they have survived outside challenge. The index is grouped by job. View
7 October 2026CorrectedProductivity J-curve
The page attributed the J-curve to a 2023 study of customer-support agents, which found productivity gains rather than a dip. It now credits Brynjolfsson, Rock and Syverson (2021) for the concept and describes the 2023 study accurately. View
7 October 2026CorrectedAI Value Gap and AI Economics in 5 Minutes
Removed figures that read as benchmarks but were house assumptions: a 10 to 20% starting attribution coverage, a 50 to 60% one-year target, and a claim that about twenty pence of each pound of time saved reaches the P&L. Replaced with guidance to measure the starting point and set a target trajectory. View
7 October 2026UpdatedAI Value Gap
Added a worked £10m portfolio showing where the value gap opens, test by test, and how the choice of denominator changes attribution coverage. View
7 October 2026UpdatedAbout and Methodology
About now says plainly that one author writes and edits the site, how it is funded, and how vendor and employer-owned products are treated. Methodology adds framework status rules and six dimensions for reading a claim. View
7 October 2026UpdatedHomepage, Start Here and 5 Minutes
Homepage rebuilt around what the site is for and where to start. Start Here adds nine task-based reading paths. The 5-minute guide adds the eight-step logic and how the site relates to external standards. View
7 October 2026UpdatedOperating Model
Reframed the landing page around a federated operating model, crediting the Tokenomics Foundation's production, consumption and value stages. View
7 October 2026UpdatedThe Frontier
Each Frontier page is now labelled a speculative thesis, and each essay ends with what is already happening, what is projected and what would prove it wrong. View
7 October 2026RetiredAI Economics Stack
Superseded by The AI Economics System, which credits the Tokenomics Foundation's stages. The page stays online so existing links work. View
6 October 2026UpdatedVendor map
Re-verified every vendor entry against its own site and added sources. Added 27 vendors and a new AI unit economics and attribution category. Removed Productiv (ceased operations) and Zluri (now an identity governance product). Updated Spot (now Flexera), Devin Desktop (formerly Windsurf), Umbrella (formerly Anodot cloud cost), IBM Kubecost and Ascerta (formerly Pay-i). Marked Tokenomics Foundation members. View
6 October 2026AddedSite
Added the changelog, corrections log, methodology page, review dates and RSS feed. Pages show Last reviewed only where a review has been logged. View
6 October 2026UpdatedFive Levels maturity model
Removed the unsourced market distribution percentages from the chart. Bar lengths are now labelled an indicative hypothesis, not measured data. View
6 October 2026UpdatedThe AI Value Gap
Added a section on evidence published since June 2026 (State of Tokenomics, State of FinOps 2026) with caveats. View
6 October 2026UpdatedWhat AI Actually Costs
Defined the denominator behind the inference share estimate, linked the Salesforce statement to its report, and noted that price ranges have not been re-verified since publication. View
6 October 2026UpdatedStart Here
Added a four-step path for CFOs. View
6 October 2026UpdatedVendor Map
Removed drafting notes from vendor entries, marked IBM-owned products, replaced unsourced 'market leader' and 'reference-grade' phrasing with neutral descriptions, replaced the CloudHealth opinion with a scope statement, and published the rating method. No ratings were changed. View
6 October 2026AddedVendor Map rating method
Published the checklist behind the four ratings. Existing ratings have not yet been re-scored against it. View
6 October 2026AddedAttribution coverage calculator
Browser-only calculator for attribution coverage, using the method in the AI Value Gap essay. Stores and sends nothing. View
6 October 2026AddedAI TCO worksheet
Added a downloadable Excel version (v1.0) of the seven-sheet worksheet described in the article. Blank template, no example data. View
6 October 2026AddedThe AI Abundance Paradox
Flagship essay on how cheap generation moves the constraint to review and absorption, with the AI Pressure Index, evidence levels E0 to E4 and return on scarce capacity. Framework proposal v1.0. View
6 October 2026AddedThe AI Value Ledger
Framework proposal v1.0 with a mapping to the Tokenomics Foundation claim chain and a downloadable Excel template. View
6 October 2026AddedThe Cost of Verification
Essay proposing verification cost as a named sub-line across the seven TCO layers. View
6 October 2026AddedHow Much Proof for How Much Autonomy
Essay pairing an autonomy ladder with evidence levels. Proposed default, v1.0. View
6 October 2026AddedScale, Hold, Redesign, Stop
Decision rules combining return on scarce capacity with evidence strength. Proposal v1.0. View
6 October 2026AddedWhat the Engineering Evidence Says About AI Coding Value
Essay on four verified studies and what each measures. View
6 October 2026AddedEvidence Review, Q3 2026
First quarterly evidence review: new evidence, what it means for positions, and what we changed. View
6 October 2026AddedStandards and frameworks map
Neutral map of FOCUS, the FinOps Framework, the TBM Taxonomy and Tokenomics Foundation work, with the hub's proposals. View
6 October 2026AddedPosition ledger
Two dated house positions with status, assessment and history. View
6 October 2026AddedIs Attention the New Scarce Resource?
New Debate instalment, both sides argued. View
6 October 2026AddedGlossary
15 new terms: Generate Verify Absorb, AI Pressure Index, counterfactual evidence level, AI value ledger, autonomy ladder, verification cost, absorption capacity, return on scarce capacity, risk tiers, net AI value, position ledger, value claim, quality floor, Observe Net Convert Book, total cost of AI. View
6 October 2026AddedAI value ledger template
Excel template v1.0 with an illustrative example row, evidence levels, decisions and instructions. View
6 October 2026UpdatedFrameworks
Added cards for the AI Abundance Paradox, the AI Value Ledger, decision rules and the standards map. View
6 October 2026UpdatedStart Here
CFO path extended to the Value Ledger and decision rules. View
6 October 2026UpdatedFinOps and AI, TBM and AI
Aligned to FOCUS 1.4 (ratified 4 June 2026), planned FOCUS 1.5, State of FinOps 2026 and TBM Taxonomy 5.0.1. View
6 October 2026UpdatedWhy Cheaper AI Will Cost More
Added a dated update box checking the position against current provider price lists. View
6 October 2026UpdatedThe Inference Cost Crisis
Added a dated update box pointing to the corrected treatment of one vendor's figures. View
6 October 2026CorrectedAI value-gap statistics (McKinsey)
The McKinsey State of AI 2025 finding had been described as 5 to 6% seeing clear financial ROI. The source says 39% report any EBIT impact and about 6% attribute 5% or more of EBIT to AI and report significant value. Corrected on the Five Levels page, the Value Management Layers page and diagram, and What CFOs Should Ask of AI ROI Claims. View
6 October 2026CorrectedExecutive confidence in measuring AI ROI
A 29% figure attributed to IBM could only be traced to a blog post. Replaced with IBM Institute for Business Value's 25% of executives who say their organisations consistently measure business value and AI ROI, labelled as an IBM-owned source. View
6 October 2026CorrectedInference forecast gap and price spread
A vendor's 30 to 50% gap between forecasts and bills had been presented as a finding, and a 9x spread had been described as fully loaded cost. The first is the vendor's own estimate with a worked example, and the second is a spread in list input prices for one model. Corrected in six articles and the waterfall graphic replaced with the vendor's worked example. View
6 October 2026CorrectedBain survey figures
Removed an unverifiable 4% over 30% figure and a revenue threshold. Restated the finding as in Bain's report: nearly 40% of those who measured landed at 0 to 10%, and 44% are funding the next wave from prior savings that fell short. Corrected on the homepage graphic and caption. View
6 October 2026CorrectedToken price ranges
Ranges were out of date. Replaced with current list prices for Anthropic and Google read on 6 October 2026, and the chart rebuilt. Self-hosted range remains an unverified estimate and is labelled as such. View
6 October 2026CorrectedAI billing APIs reference
The Anthropic usage endpoint was wrong, the GitHub Copilot usage endpoint had been replaced by usage metrics endpoints, and Cursor now documents an Admin API. Added a last verified table. Five providers are marked not re-verified. View
6 October 2026CorrectedIDC finding on FinOps in AI projects
Wording changed from 'meaningfully integrated' to 'embedding', matching the IDC statement as reported in April 2026. View
6 October 2026ReviewedThe AI Value Gap
Evidence added and checked at primary: State of Tokenomics, State of FinOps 2026, McKinsey, Bain. Uber and WSJ reports confirmed through secondary coverage. MIT NANDA, KPMG and UK government items not re-checked. View
6 October 2026ReviewedFive Levels
Supporting statistics re-verified (State of FinOps, IDC via CIO Dive, IBM IBV via TechInformed, McKinsey). Distribution removed. View
6 October 2026ReviewedHomepage statistics
Homepage statistics checked against McKinsey and Bain. View
6 October 2026ReviewedWhat AI Actually Costs
Price ranges checked against Anthropic and Google price lists. OpenAI not read. Self-hosted range unverified. View
6 October 2026ReviewedWhat Cloud Taught Us About the Real Cost of AI Inference
Checked against the Inferbase source post of 13 May 2026 and Anthropic's price list. Claims restated. View
6 October 2026ReviewedThe Inference Cost Crisis
Update box only. Body checked for the Inferbase figures. View
6 October 2026ReviewedWhy Cheaper AI Will Cost More
Position tested against current price lists. No invoice series verified. View
6 October 2026ReviewedProgrammatic Access to AI Costs
Partial: Anthropic, GitHub Copilot and Cursor checked against documentation. Five providers not re-verified. View
6 October 2026ReviewedFinOps and AI
Aligned to FOCUS 1.4, planned 1.5 and State of FinOps 2026. View
6 October 2026ReviewedTBM and AI
Aligned to TBM Taxonomy 5.0.1. View
6 October 2026ReviewedStandards and frameworks map
All rows read from primary sources. View
6 October 2026ReviewedVendor Map
Partial: four ownership statements verified (CoreWeave and Weights and Biases, Cognition and Windsurf, IBM and Apptio, Stripe and Metronome). Remaining entries and all ratings not re-scored. View
6 October 2026AddedKey numbers
Every statistic the site relies on, with source, sample, date, caveat and the pages that use it. The full evidence register is downloadable as CSV. View
6 October 2026UpdatedHomepage
Rebuilt around the evidence: three sourced figures drawn from the evidence register, six core pieces, reading paths by role, frameworks, reference pages and tools. View
6 October 2026UpdatedLibrary
Curated 'Start with these', topic and type filters, reading time, and the date each piece was last reviewed. Ten unreviewed general guides moved to an archive section. View
6 October 2026RetiredTen general guides
Mid-market, consulting partners, regulated industries, engineering leaders, CAIO first 100 days, cost basics, FinOps for inference-era workloads, vendor selection, business case guide and where TCO models fail are archived: not reviewed against sources, partly duplicated by maintained pieces, and containing unlabelled illustrative examples. Still reachable at their URLs with a banner, not indexed. View
6 October 2026CorrectedClaim labels
186 claims labelled Evidence cited no source and are relabelled Interpretation. Evidence is now reserved for claims that link to a source. View
6 October 2026CorrectedFlagship framework pages
Removed or relabelled figures that could not be traced: IDC 56% outside IT budgets, a 70/20/10 split attributed to McKinsey, Deloitte 16%, IBM and APQC finance benchmarks, Deloitte 2026 decision maturity, Apptio 35% and a BCG function-share figure. Verified figures from Deloitte, BCG, Kyndryl, McKinsey and Gartner now link to their sources. Layer cost ranges are labelled as this publication's estimates. View
6 October 2026CorrectedIndustry perspectives
Removed percentages attributed to Accenture, Microsoft and BCG that could not be traced to a source. View
6 October 2026CorrectedIn five minutes
Replaced an unverifiable 79% figure and corrected the Bain wording and quote. View
6 October 2026UpdatedGlossary
Merged five duplicate entries (FOCUS standard and specification, AI return on investment, retrieval-augmented generation, allocation) into their main entries, with redirects. View
6 October 2026AddedVendor Map
Added Surveil to Workforce AI Usage and SaaS AI Spend, rated against the published method. Scope read from the vendor's own website. View

See also the corrections log and the methodology.