This is the full edition, with the evidence behind every claim. Read the short edition.
Part one: what goes away
Most AI data will sit there like microns in an ocean droplet. Trillions of data points, generated, scraped, summarised and stored, each one technically available and practically useless: no source anyone can open, no method anyone can read, no record of ever being right, and no one accountable for it. Most of it will be made only to be sold to another AI, and most of the time it will fail even at that, because the buyer can generate the same summary for nothing. It yields nothing to the economy, nothing to the people who own it, and nothing to the data centre it inhabits except a bill for the power, the cooling and the disks that keep it alive. It answers everything and helps no one decide anything. A brand cannot build a program on it, a banker cannot put it in a model, and an assistant cannot cite it, because there is nothing behind it to cite. It is not an asset. It is a cost with a storage address.
That changes what an answer is worth. When every assistant can produce one in a second, an answer on its own is worth nothing, and the value moves to the line that can be checked: the fact with its source, its date and a link anyone can open. The same pressure falls on every method kept behind a curtain. A proprietary score no outsider can test was a moat when the data was hard to get; it becomes a liability the day a published method, run on the same public filings, gives the same answer for free. Google's own guidance already tells publishers that the way into its AI answers is to be indexed, plain and quotable, and the largest research houses now list, in their own annual reports, the risk that third parties will use their data with AI to make their own insights.
Reach goes the same way. For twenty years brands bought traffic, impressions and "people reached", counted by the platform that sold them. Most counted traffic is now automated, and the house's own counters show it plainly: about one in four counted desk views is a reader at a home or an office. Once buyers can see that, they stop paying for numbers the seller grades itself. And when the numbers can be seen, so can the seams in the old way of serving a brand: the research firm that does not build, the agency that does not measure, the broker that never meets the brand and the event producer hired separately, each handoff leaking and each markup now visible. Personal data, the fuel of much of that machine, is under steady legal pressure, and measurement is moving from what people can be tracked doing to what companies actually report.
The proof, in their own numbers
None of this rests on opinion. The companies, the regulators and the grid operators have already written it down.
Follow the money first. Amazon spent $215.2 billion on capital in the twelve months to September 2026, against $120.1 billion a year earlier. Alphabet raised $80 billion of new equity in June 2026 to pay for "AI compute infrastructure" and expects to spend $180 to $190 billion this year. Meta spent $72.22 billion in 2025 and now expects $135 to $145 billion for 2026. In the annual reports filed with the SEC, the phrase "artificial intelligence" appeared in 558 documents in 2020 and in 3,914 so far in 2026; "generative AI" went from 69 in 2023 to 1,207. Inside the companies, the money comes out of the old lines: Klarna cut its external agency spend by a quarter and credited AI with 37% of its marketing savings.
The agencies' own accounts show the other side. WPP ended 2025 with 98,655 people against 108,044 a year earlier, revenue down 5.4% like for like and a £641 million impairment, £393 million of it on Ogilvy, and a plan to fold itself into a "single company". Dentsu wrote off ¥396.1 billion of goodwill, lost ¥327.6 billion and paid no dividend. Omnicom bought Interpublic and is consolidating to survive, on track for $900 million of savings this year. The giants that are still growing are growing on AI, data and technology, not on the old agency trade. Publicis grew 5% this quarter by combining "data, technology and talent" and expects the industry to settle around "three global players"; Omnicom's growth runs through its Omni platform, built on "comprehensive identity and data infrastructure", the very asset this piece shows turning into a liability, while it exits real estate; and Estée Lauder cut 31 agencies to one. The traditional agency is not being saved by this. It is being absorbed into three data companies, or closed.
The tangible brands still growing tell the same story from the other side: they are the ones people consume, wear or carry out of a store the same day. Coca-Cola's volume rose 5% and McDonald's plans about 2,500 openings this year; Tapestry and Ralph Lauren raised their outlooks on store sales, and Estée Lauder now calls the store its media. What is going away is the tangible that sold nothing on its own: the printed piece, the mailed flyer, the branded giveaway that could never show what it earned.
The data glut has a bill, and it is landing on the grid. Data centres used 4.7% of US electricity in 2024 and are projected to use up to 11.8% by 2030, by the Lawrence Berkeley National Laboratory's report for the Department of Energy. In Virginia, Dominion Energy's contracted data-centre capacity went from about 16.5 gigawatts in July 2023 to about 53.8 gigawatts in July 2026, and on the PJM grid nearly all of the rise in forecast peak demand now comes from data centres. Every data point stored and never used is paid for here, and much of it is paid for with borrowed money: Moody's says its fastest-growing ratings line this quarter came from "AI-related financing and hyperscaler investments", Meta has created a new class of stock to raise capital, and Snowflake sold $3.5 billion of convertible notes. What is borrowed has to be earned back.
And success is already measured in moments. Walmart's advertising grew 38% in its latest quarter, Walmart Connect 43%; Amazon took $19.6 billion of advertising in a single quarter, at the shelf, in the moment of purchase. The checkout itself is moving to agents: Stripe and OpenAI put Instant Checkout inside ChatGPT on the open Agentic Commerce Protocol, Google built its Universal Commerce Protocol with Shopify, Etsy, Wayfair, Target and Walmart, and Visa says agent-initiated payments will reach "mainstream adoption in 2026". Underneath it, nine federal agencies have named the Legal Entity Identifier the standard for who a company is, from 1 October 2026.
Who falls from the top
Palantir said it plainest, in its letter to shareholders this quarter. Every industry, from banks and insurers to hospitals and the old law firms, will be "remade or set briskly aside". The outcomes will be binary, winners and losers with nothing in between. The middle becomes a wasteland of former leaders who did not listen, and the bottom fills with those who once had an insight or an edge and let it go, their fate already sealed. Palantir can say it that harshly because its own numbers show it happening: in the third quarter of 2026 its US commercial revenue reached $984 million, up 148% on a year earlier, and passed its US government revenue for the first time.
Follow that same pressure through the industries this note touches and the verdict is the same. A business is remade around what can be checked and delivered, or it is set aside. The middle empties first, because it sold the layers a buyer no longer needs, and the bottom fills with those still selling the cheapest copy of what a machine now does for nothing.
The first to feel it are the businesses that sold access to public facts. Seat-licensed research and data terminals were built when filings were hard to read at scale; now the facts are public, the arithmetic is published, and an assistant will not cite a paywall. S&P Global, FactSet, Morningstar and Gartner each name AI, and third parties using their data, as competitive risks in their latest 10-Ks. Next to them sits analyst opinion sold as a ranking: a placement decided by opinion loses ground to a placement decided by filed numbers anyone can check, and Gartner's own report expects "more competition with increased adoption of AI services".
Content built only to catch traffic goes next. Assistants answer the question before the click, and a page that only rephrases what is already known is never the one cited; Anthropic's citations, for instance, return the exact passage that supports a claim, so the page with the passage wins and the page with the paraphrase disappears.
Then the traditional agency, in every form it took. That is long gone, and none of the quality workers want anything to do with agencies that gouged clients an hourly rate for "services" that never resulted in anything. The list sellers are already gone: their rented names fed the advertising mail that has fallen by almost half since 2007, and their trade in personal data ran into the privacy laws state by state. The marketing-services shop that executed campaigns by hand is going the same way, its work now drafted, scheduled and reported by a model in an afternoon. So is the design studio that sold layouts, decks and templates a model now produces in seconds, in a hundred variations, for a few cents, and the media buyer whose margin lived in the gap between what it paid and what it billed, a gap now closed by automated buying and by clients who can see the auction. So are the expensive CGI firms and advertising sleuths billing hundreds of dollars an hour for a thirty-second clip that a model now renders in minutes. Above them the holding companies are consolidating to survive: Omnicom completed its merger with Interpublic in November 2025, and it measures itself by its own revenue growth and expenses, not by what happens to its clients' filed results. A few will be remade around measured results and accountable delivery. The rest are closing, or already closed.
Success is now measured in microns of moments, within trillions of data points engaging with each other for perceived success. Traditional methods are gone.
Commerce moves too. Gifting and promotional platforms, social listening tools and AI platforms all grew in the gray area of the budget, the line nobody had to defend. That gray area is gone completely. Gifting and promotional platforms that grew by holding the buyer's data now meet a new kind of buyer: an AI agent acting under a signed mandate, which wants a structured feed and a quote, not a marketplace that keeps its customer. And the data they held has turned from an asset into a liability. Anyone holding a buyer's names, addresses, orders and gift lists now carries the breach, the lawsuit and the headline that come with them, and the old agencies and networks that passed that data from hand to hand, with no one checking who held it next, are exactly where the damage starts. The house's own papers on the House Edge have traced it: the small supplier who sinks a billion-dollar mark over a weekend, the vendor who becomes the client's headline, the agencies that bill the handoffs, and the chain of custody most houses cannot even name. That is dangerous, and buyers now know it. Social listening sold as brand success measures what people say rather than what the brand and its peers actually earned, and that gap is now visible beside the filed numbers. And AI platforms sold without proof of results are the most exposed of all; C3.ai's latest 10-K, with revenue down, a restructuring and a change of chief executive, shows what happens when a platform has no public record to point to.
Part two: what the future keeps
What survives all of this is what can be checked and what can be done.
It keeps the checkable line: one fact, its source, its date and a link anyone can open, the line an assistant cites, a banker pastes into a model and a brand builds on. It keeps published methods, the measures the top investment houses and VCs already trust, run in the open on the actual audited numbers so that nobody has to trust a secret. It keeps a record that cannot be back-dated, forecasts sealed before the outcome and graded after it, misses included, because time is the one input no competitor and no model can buy.
It keeps success found where it actually happens. The filings say what worked; the launches, campaigns, deals, retail and social moves, events and hires behind the brands that win say how it worked, and the house reads both every day and merges them, so the record and the story check each other. It keeps identity and signatures as well: the Legal Entity Identifier becoming the identifier of record, verifiable credentials on filed facts and signed ledgers, so that who said what, and when, can be proved.
The leaders' own outlooks this season say the same thing in their own words. The brands are cutting their own layers, not only their agencies': Procter & Gamble is removing up to 7,000 overhead roles and Nike expects about $2.5 billion of savings, mostly in people. A client with fewer people to manage vendors needs work that arrives finished, measured and easy to check. Every outlook now carries a tariff line, Tapestry builds a tariff rate into its guidance and PepsiCo books refunds, so a program has to be planned on the landed cost, not last year's price. And the shopper is stretched: McDonald's credits an "intervention on value" for its recovery and Berkshire's businesses report "sluggish customer demand", so what a program earns has to be proved at the price the shopper actually pays, not in how many people it reached.
It keeps trust that can be shown rather than promised. Visa now calls trust "the next growth vector" and is paying $2.4 billion for fraud intelligence. A successful house shows its trust: a named chain of custody, a dated record, and an answer when something goes wrong.
It keeps the client's data where it belongs, with the client. Microsoft says it helps organizations "build and own their own intelligence", and WPP's clean room lets partners work together without moving the data. The rising house works inside the client's data and leaves it there; it never copies it out, keeps it or claims it.
It keeps the agents that are built on real data. Every company is now building AI agents to read, decide and buy for it, and an agent is only as good as what it reads. An agent fed on summaries acts on guesses, at machine speed, in the company's name. The agents that last read the primary record itself, the filing, the price, the order, the signed mandate, cite the line they acted on, know exactly who the counterparty is from its legal identifier, and are measured against what actually happened afterwards. Anthropic's own platform returns the exact passage behind every claim, the Model Context Protocol lets an agent call a source directly instead of scraping a copy, and the payment networks now require a signed mandate before an agent can spend. Amazon now ships rules to "prevent autonomous agents from causing unintended harm", and Microsoft has made agents a reporting segment of their own. Building agents that work with real data, and that can show their working, is now the job, and the limits come first: what the agent may read, buy and sign, and who answers when it errs. It is the job any successful house has to do for its clients.
Above all it keeps accountable execution. An AI can tell you who is winning; it cannot run your event, make your pieces or answer for the result. The future belongs to the house that turns what works into the brand's own program, makes it, and measures it again when the next filings land, and that can be hired by a person or by an agent: a feed an agent can read and a quote it can accept under a signed mandate. The payment networks are building the rails for it: Mastercard paid $1.4 billion for stablecoin infrastructure and Visa runs stablecoin settlement of its own. A successful house takes the agent's order, settles on whatever rail the signed mandate allows, and keeps the record.
The short version: the future keeps what can be checked and what can be done, and it discards what can only be summarised.
Why the old ways are now dangerous, and how fast they sink a brand
Everything above says the old ways are going. The harder truth, and the subject of the house's next piece, is that while they last they are dangerous. The old networks were built to pass things along: a buyer's list to a broker, a brief to a subcontractor, a file to a factory, a customer record to a vendor's vendor. Nobody checked who held it next, and nobody was asked to. That pass-through was the business model, and it made every brand in the chain complicit in whatever the weakest link did.
The record already shows where that ends. In the SEC filings the house reads every day, companies now report the damage themselves. On 8 September 2026 Veradigm told the SEC that "one of its third-party vendors experienced a cybersecurity incident that impacted certain data associated with a small number of the Company's customers"; six days later CenterPoint Energy Resources reported that an unauthorized third party had obtained personal information of a portion of its customers. The house's own Breach Clock and cyber-disclosure boards hold every one of these, filing by filing, each linked to its source. The regulators have gone after the data trade directly: the Federal Trade Commission has finalized orders barring X-Mode and its successor Outlogic, Gravy Analytics and Venntel, and Mobilewalla from selling sensitive location data, and proposed a ban on InMarket selling precise location data. And the House Edge papers have traced what it does to a brand: the small supplier who sinks a billion-dollar mark, the vendor who becomes the client's headline, the publicist and producer whose conduct a search engine cannot tell from the brand's own.
What has changed is the speed. A brand used to have days to answer for a bad link: a news cycle, a statement, a quiet change of supplier. There is no room any more for the questionable old media monger, angry and posting vitriol, while accessing your company's records at the same time; they are now filed below the bottom feeders. Now the first sentence anyone reads about a brand is written by a machine, from whatever stood closest to its name, the moment someone asks. A breach in a vendor's vendor, a supplier's scandal, a partner's post: each is read, joined to the brand and repeated in every answer at once. Brands no longer sink over days. They sink in microns of a second, and the old networks are where the hole is.
That is why the house works the other way: a checked network small on purpose, a chain of custody it can name, blind shipping so suppliers never hold the buyer, no personal data in any score, and a record that shows what it said before the outcome. The next piece is about exactly that: why the older ways and approaches are dangerous and no good. The creepy pass-throughs that handed a buyer's data on and never asked who held it next, the complicity of every network that looked away, the lack of action when it went wrong, and how all of it now sinks a brand in microns of a second, shown with first-hand proof and the real record.
---
The evidence
Research note, 8 November 2026. Primary sources first (regulators, standards bodies, the companies' own filings and announcements, protocol specifications), reputable reporting second. Every claim carries its URL in the text or in the sources list, with the source type in brackets. Anything not read on a primary source today is marked UNVERIFIED or attributed to the reporting that carried it. Quotes are verbatim from pages read on 7–8 November 2026.
Who wrote this, and why it can. This piece comes from a house that works the way it describes. Every day it scours brand success wherever it happens, in the launches, campaigns, deals, retail and social moves, events and hires behind the brands that win, through ten editorial desks, and merges that with the record of every company the SEC releases: about 2,800 companies scored against their field with the published measures the top investment houses already use, Piotroski, Altman, Beneish and Sloan among them, plus the Trillion Score and Fall Risk. All of it is free to read, cite and download, for people and for machines alike, beside a forecast ledger sealed before each outcome and graded in public, misses included. Then the house turns what works into a brand's own program, event and pieces, through 1,400 vetted American makers and no plant of its own. Brand success, proven daily.
1. Where company data is going (5, 10, 20 years)
The United States has set the floor. On 21 May 2026 the SEC and eight other agencies adopted the Financial Data Transparency Act joint data standards (Release 33-11420), published in the Federal Register on 25 June 2026 and effective 1 October 2026 (regulator). The SEC's page is careful: "At the effective date, the joint rule will not change any reporting requirements without further action by the agencies." The content is what matters. The legal-entity identifier is ISO 17442, the LEI; data transmission and schema formats must, where practicable, be searchable and machine-readable, carry documented metadata in machine-readable taxonomy or ontology models, consistently flag which data satisfies a regulatory requirement, and be nonproprietary or openly licensed; and agency-specific rules "must take effect not later than two years after the final joint rule is promulgated" (standards body summary). That puts agency rules on the books by mid-2028.
The SEC already serves every standard XBRL fact as open JSON, with a hole the house can fill. The EDGAR APIs on data.sec.gov require no key, update "throughout the day as filings are disseminated", and expose submissions, companyconcept, companyfacts and frames (regulator). The aggregated APIs "include only facts that use non-custom taxonomies, such as us-gaap, ifrs-full, dei, or srt"; company-specific extension facts are excluded. A house that reads every filing can publish the normalised extension facts, point-in-time, that the SEC's own frames leave out. Filer access itself moved to EDGAR Next in 2025: Login.gov with multifactor authentication, "Legacy login methods have been discontinued", optional filing APIs in beta (regulator).
Less frequent filings are on the table. On 5 May 2026 the SEC proposed optional semiannual reporting on a new Form 10-S (Release 33-11414), with Chair Atkins calling it the "first step" and staff evaluating broader Regulation S-K amendments (reporting on the proposing release 1, reporting on the proposing release 2). If adopted, the gap between filed financials widens for electing companies, which raises the value of 8-K event reading, of forecast ranges that bridge the gap, and of a publisher that scores those ranges in public.
Identity is becoming verifiable, not just assigned. Active LEIs passed 3.02 million in Q1 2026, with 99% of registrants reporting parent information; EU DORA and a March 2026 Reserve Bank of India mandate were the drivers (standards body). ISO 17442-3:2024 standardised the verifiable LEI: "digitally signed, tamper-resistant credentials" built on ACDC containers and the KERI protocol with GLEIF as root of trust, covering organisations and the people who act in official roles for them (standards body 1, standards body 2). GLEIF has signed its own Inline XBRL annual report with a vLEI, and XBRL International's digital-signatures working group is defining how signatures attach to an XBRL document (standards body). Within ten years a filed number will carry a cryptographic chain from the filer's role-holder to the fact. A publisher whose every number already cites the accession on sec.gov is one step from signing its own outputs the same way.
Sustainability is being tagged, after a delay. The ISSB taxonomy (April 2024) is in its first update, consultation closed 28 September 2026 (standards body). Omnibus I entered into force on 18 March 2026 and narrowed CSRD to undertakings with more than 1,000 employees and EUR 450 million turnover, wave 2 reporting from financial years beginning 1 January 2027 (reporting, law firm). EFRAG's revised draft ESRS XBRL taxonomy (17 September 2026) is in consultation to 11 November 2026, with final advice due "by the end of 2026", after which ESMA drafts tagging RTS and the Commission amends the ESEF regulation (standards body). ESMA began collecting data for the European Single Access Point on 10 July 2026; ESAP opens to the public by July 2027 (regulator).
What becomes machine-readable next, and what the house can be first to publish.
Five years (to 2031): FDTA agency rules in force; LEI on US regulatory collections; ESAP live; ESRS tagging adopted; vLEI pilots in filings. First to publish: an LEI on every company page; extension-fact normalisation the frames API does not do; the first open, scored daily read of ESAP filings.
Ten years (to 2036): signed facts become the audit trail; semiannual reporting thins the interim data set for electing companies. First to publish: a signed forecast ledger and signed scores; an events-between-filings board that fills the 10-S gap.
Twenty years (to 2046): the filing is a dataset first and a document second, and the question moves from "what did the company file" to "who read it correctly in advance". First to publish: the longest continuous, hash-sealed, publicly scored record of company-health forecasts in the field.
2. Where AI retrieval and citation are going
The platforms say the same thing: be indexed, be quotable, be plain. Google's guidance for AI Overviews and AI Mode: "There are no additional requirements to appear in AI Overviews or AI Mode, nor other special optimizations necessary"; "You don't need to create new machine readable files, AI text files, or markup to appear in these features"; a page must be indexed and "eligible to be shown in Google Search with a snippet"; both features may use "query fan-out" across subtopics and data sources (company documentation). OpenAI's ChatGPT search launch (31 October 2024) said any website can choose to appear, with sources in a sidebar, alongside named partners (company announcement). Anthropic's Citations returns the exact passages supporting each claim from documents, PDFs and search results; for web search "Citations are always enabled", each carrying `url`, `title` and `cited_text` of "Up to 150 characters", and developers showing outputs "must" include citations to the original source (company documentation 1, company documentation 2). The 150-character window is a design constraint: every fact should be a complete sentence with the number, the period and the source in it.
Payment for citation is becoming a line item, and deals do not buy the answer. Perplexity's Comet Plus (25 August 2025) pays publishers for three kinds of traffic: human visits, search citations and "agent actions" completed on publisher sites (company announcement). Cloudflare moved from pay-per-crawl (July 2025) and robots.txt Content Signals (September 2025) to a 2026 pay-per-use model that pays when content appears in results (reporting). Ahrefs' November 2026 count found OpenAI partner domains' share of ChatGPT citations among the top 200 at 7.5% in March 2026, 15.8% in July and 12.5% in late September (reporting, vendor study). A free, CC BY, primary-sourced page competes on equal footing.
The machine-readable conventions that exist. llms.txt (Jeremy Howard, 3 September 2024) is an inference-time site index that Google says it does not need (specification). schema.org `Dataset` carries `distribution`, `license`, `measurementTechnique`, `variableMeasured`, `temporalCoverage`, `isBasedOn` and `citation` (standards body); `ClaimReview` carries `claimReviewed`, `itemReviewed` and a `reviewRating` that can be numeric and textual (standards body). A forecast hashed before the outcome and graded after it is, structurally, a ClaimReview of the house's own claim; publishing it in that vocabulary is cheap and unusual.
MCP is the plumbing, and it is now neutral ground. The 2025-11-25 specification added OpenID Connect discovery, icons for tools, URL-mode elicitation, long-running tasks, "guidance on tool names" and formal governance with working groups and SDK tiers (specification). The official registry is still in preview (specification 1, specification 2). On 9 December 2025 the Linux Foundation formed the Agentic AI Foundation with MCP, goose and AGENTS.md as founding projects and AWS, Anthropic, Block, Bloomberg, Cloudflare, Google, Microsoft and OpenAI as Platinum members (standards body); OpenAI's Apps SDK for ChatGPT is built on MCP (reporting). The incumbents are already there: FactSet lists "model context protocol ("MCP") servers" among its delivery methods (company filing) and Morningstar "launched the Morningstar MCP server, providing turnkey AI-connectivity of Morningstar data" (company filing). Theirs are paid and gated. The house's is free and keyless; that is the difference an agent sees.
Agentic commerce protocols have converged on signed intent against a product feed. The Agentic Commerce Protocol (OpenAI and Stripe, 29 September 2025) specifies a product feed, agentic checkout and delegated payment (company announcement). Google's AP2 chains a Checkout Mandate and a Payment Mandate into "an audit trail for both human-present and human-not-present transactions", as an extension of A2A and MCP, with standardisation continuing in FIDO Alliance working groups (specification); Google's Universal Commerce Protocol followed in January 2026 (reporting). Visa's Trusted Agent Protocol (14 October 2025, with Cloudflare) signs agent intent with HTTP Message Signatures (company announcement); by 18 December 2025 Visa reported "hundreds of secure, agent-initiated transactions" with more than 100 partners, including B2B pilots with Ramp (company announcement). Mastercard's Agent Pay (29 April 2025) issues Agentic Tokens to registered, verified agents (company announcement). A brand program with a price, a spec and a maker list is a product feed.
What a source must be to be the one agents call and cite, and what the house has. Indexed and snippet-eligible (has). Stable URLs with one sentence per fact and the source inside the sentence (has in part; worth a style pass). Open licence in machine-readable form (CC BY 4.0: has; `Dataset` markup: to add). A keyless MCP server with tool names, descriptions and icons to the November 2025 guidance (has the server; metadata is a small job). A record verifiable without trusting the publisher: the hashed ledger (has). A product feed and quote endpoint so agents can price and commission programs (`price_program` exists; an ACP/UCP-shaped feed to add). A signing chain (vLEI: not yet; nor has anyone in the field).
3. Where the leaders in corporate intelligence are going, in their own filings
Nearly all of them are putting AI on top of data they charge for, and the research houses name the same risk in their own annual reports: that AI, and other people using their data, will erode what they sell. None of them builds in the open or keeps a public record of its own calls, and the few that deliver the work do it behind closed doors.
Ten leaders, in their own words. S&P Global, FactSet, Morningstar, Gartner, Palantir, Zeta, Sprinklr, Omnicom, Snowflake and C3.ai.
S&P Global, FY2025 10-K. What it says about AI: "AI continues to be an increasingly important part of our business and industry"; a "Company-wide AI strategy to drive our approach to data protection, licensing and AI integration"; Kensho LLM-ready APIs (company IR). How it holds its data: Risk factor: "Third parties could also use our data with AI tools to create their own insights"; "Our competitors are deploying AI in ways that could materially reduce demand". What it is not building: No open data; no public record of its own calls; does not build or deliver.
FactSet, FY2026 10-K. What it says about AI: "help our clients re-design their core workflows with AI"; MCP servers as a delivery method; 9,258 clients, 260,671 users, ASV retention above 95%. How it holds its data: Paid, contracted; 650+ clients on MCP contracts or trials per the Q4 call (reporting). What it is not building: No open data; no public record of its own calls; does not build or deliver.
Morningstar, FY2025 10-K. What it says about AI: "Make Morningstar AI-native, rethinking capabilities and processes"; "Become the trusted universal language for public and private markets"; MCP server launched. How it holds its data: Licensed data; DBRS ratings paid by issuers. What it is not building: No open licence; no public record of its own calls; does not build or deliver.
Gartner, FY2025 10-K. What it says about AI: AskGartner launched August 2025; "independent, objective insights ... through subscription services"; "We anticipate encountering more competition with increased adoption of AI services". How it holds its data: Subscription, twelve-month minimum. What it is not building: No open data; no scored predictions; does not build or deliver.
Palantir, FY2025 10-K. What it says about AI: Ontology "serving as the heart of our platforms"; AIP "a development toolchain for building AI-powered agents and automations"; bootcamps "deliver real workflows on actual customer data in days"; $4.5 billion revenue, 954 customers. How it holds its data: Customer data inside customer deployments. What it is not building: No public data product; no public forecasts; delivers only inside the client's walls.
Zeta Global, FY2025 10-K. What it says about AI: "Athena is the interface to Zeta's AI-native infrastructure layer"; data on "more than 245 million individuals in the U.S. and more than 535 million individuals globally". How it holds its data: Personal-data identity graph with consent management. What it is not building: Built on personal data, the asset now turning into a liability; does not measure filed results.
Sprinklr, FY2026 10-K. What it says about AI: "AI-native platform"; "ingests, processes and analyzes unstructured conversational data at massive scale"; "Competitive Insights & Benchmarking" to "benchmark their social performance against competition". How it holds its data: Platform data, per client. What it is not building: Benchmarks social chatter, not filed results; no open data; does not build or deliver.
Omnicom, FY2025 10-K. What it says about AI: IPG merger completed 26 November 2025; "our next generation of Omni, our proprietary marketing intelligence platform"; "generative AI and agentic AI have, and will continue to have, a significant impact". How it holds its data: Acxiom identity data; proprietary. What it is not building: Measures itself by "revenue growth and variability of operating expenses", not clients' filed results; delivers, but nothing it measures is open.
Snowflake, FY2026 10-K. What it says about AI: "build and deploy AI agents"; Snowflake Intelligence and Cortex Agents generally available in FY2026; Marketplace of "hundreds of live, ready-to-query third-party data sets". How it holds its data: Distribution rail for others' data, Iceberg-open. What it is not building: Not a publisher and holds no view; a distribution rail for other people's data.
C3.ai, FY2026 10-K. What it says about AI: Partner ecosystem "serves as a primary channel"; "rapidly move down-market"; restructuring for about $135 million of savings; Siebel back as CEO 8 May 2026 (company announcement). How it holds its data: Enterprise platform. What it is not building: Not a data publisher; what happens when a platform is sold without proof.
The pattern: AI is a delivery layer over proprietary, paid data; the moat words are "trusted", "independent", "objective"; the risk-factor words are that off-the-shelf models and third parties using their data will erode demand. None publishes under an open licence, none keeps a public, hash-sealed record of its own forecasts graded after the fact, and none closes the loop from measurement to the brand program meant to move the measure. Sprinklr benchmarks social chatter, Omnicom measures itself by its own revenue, and Zeta measures people. What none of them builds is what the future keeps: the checkable line, the open method, the record that cannot be back-dated, and delivery measured against what the client actually filed. Palantir, the one growing fastest, is growing on exactly that promise of real data and real outcomes, with US commercial revenue up 148% in a year, but inside each client's walls rather than in the open.
4. Where brand measurement and marketing procurement are going
Measurement is moving back to aggregated, model-based methods. Google's Meridian is an open-source (Apache-2.0) Bayesian marketing-mix model on geo-level data that can be calibrated with experiments (company documentation 1, company documentation 2). The reason is durable: on 22 April 2025 Google chose "to maintain our current approach to offering users third-party cookie choice in Chrome" (company announcement), while the statutory floor keeps rising: Indiana, Kentucky and Rhode Island privacy laws took effect 1 January 2026 and Maryland's became operative 1 April 2026, for 19 or 20 states depending on the tracker (industry body). Retail media got its first common yardstick in the IAB/MRC Retail Media Measurement Guidelines (January 2024), whose incrementality chapter treats controlled experiments as the base method and demands transparency on method and assumptions (industry body).
Procurement is asking for receipts. The ANA's Programmatic Media Supply Chain Transparency Study found that at most 36 cents of each programmatic dollar reaches the publisher, over $20 billion of efficiency available, and that only 21 of 67 willing advertisers could obtain log-level data (industry body). The 2025 Trends in Agency Compensation report is pitched at marketing, finance and procurement together (industry body), and the ANA's May 2026 study reports planned compensation changes "amid concerns over transparency, rebates, and artificial intelligence-driven labor" (industry body).
The house's own channel is large, fragmented and under the heaviest pressure it has ever faced. Print is no more. It is no longer even the four cents of every advertising dollar it was in WPP Media's last count, $45.5 billion of $1.08 trillion worldwide in 2025, and it has stopped being considered or counted at all; Publicis no longer reports a print line at all, the presses are being removed and the plants are becoming apartments (company forecast). The house reported this in its own notes on 28 October 2026, and found where the money goes next: the word AI appears in 3,904 annual reports filed this year, AI governance in 226, and the next four cents go to AI tools, AI insurance and AI training, not to paper (house reporting). AI needs more than those four cents, so it is now coming for the physical products and events that cannot show what they returned. The money for AI comes out of the same budgets, and marketing has never been able to explain what a $50,000 event bought. Gone are the $50,000 events, and even the $5,000 ones. When a team can spend $50 on data and get a clearer read on who to invite, what to make, what to say and what moved afterwards, for the same event, the $50,000 has to answer for itself. What survives is spend that arrives with a reason and leaves with a receipt, which is the house's whole method: measure the field first, make only what the record says works, and measure again when the next filings land. The channel's own numbers still grow, slowly, and show the strain. ASI reports North American distributor sales of $27.7 billion in 2025, up 4.2%, and only 2.5% growth in Q2 2026 with inflation and delayed decisions weighing (industry body 1, industry body 2). PPAI puts US distributor sales at $27.1 billion in 2025, with distributors above $2.5 million taking 54%, and recorded $4.3 billion (16%) bought from non-industry suppliers in 2024 (industry body 1, industry body 2). Visa's B2B pilots and the ACP/UCP feed specifications describe how that $27 billion will start to be ordered: an agent with a mandate, a merchant with a feed.
Where did tangible media go? Tangible brands are going away fastest, and the money is moving to AI. The physical real estate these businesses owned is selling for pennies on the dollar, to make way for housing and better supermarkets.
Read 8 November 2026 from official and primary sources, plus the house's own published reporting.
A note on the job counts: the government files each employer under the industry the employer itself reports as its main activity. A packaging, billing or mail house can be counted as printing, a publisher's press can sit under printing or publishing, and some printers file under other codes. The counts are what the businesses say they are.
Read together: the jobs, the mail and the magazines that carried print are mostly gone, packaging is running to the minimum, a clear bag and a plain label where a custom box used to be, the agencies have stopped counting print, and marketing budgets are not growing while about 15 cents of each marketing dollar now goes to AI. That money comes from the lines that cannot show what they returned.
Where "measure brand success from filed results" fits, and why no one else does it. MMMs measure media against one company's sales; retail media against one retailer's till; social tools measure noise against noise. None is public, none is comparable across a field, none can be checked by a stranger. Filed results are public, audited, comparable, dated and machine-readable, and they are the number the CFO and the board already accept. No competitor builds on them for structural reasons: data vendors sell the raw material and would undercut themselves; agencies are paid on inputs and would be grading their own homework; platforms measure only what runs through them. A house that publishes the yardstick free and sells the program has no such conflict.
5. Ten positions to take now
1. The public scored forecast record as a standard. What: keep hashing ranges before the outcome and grading them after, in public, and publish the hit rate by sector and size. Defensible because time cannot be bought; every year the record exists is a year no competitor can replicate. Cost: near zero marginal, one methods document. First proof: a calibration page of 80%-range coverage by year, with ClaimReview markup.
2. Receipts for brand programs tied to filed results. What: every program carries the client's field score before and the filed results after, stored and cited. Defensible because the ANA's findings show buyers want log-level proof and nobody has a public, audited after-measure. Cost: client consent and a template. First proof: one anonymised receipt with the 10-K accession numbers behind it.
3. Free for citation, paid for execution. What: the data stays CC BY and keyless; revenue comes from programs, events and pieces. Defensible because the incumbents' 10-Ks name third parties using their data with AI as the risk; for the house that is the plan. Cost: forgone data-licence revenue. First proof: the model stated next to the licence.
4. Agent-callable procurement of brand programs. What: an ACP/UCP-shaped feed and quote endpoint so an agent holding an AP2 or Visa mandate can price and commission work from 1,400+ makers. Defensible because the protocols and card rails are live and no promotional-products distributor has done it. Cost: a feed schema and a quote API. First proof: one program priced by an agent through `price_program`, logged.
5. A brand score every CMO quotes. What: one number per company, against its field, from filed results, daily, free. Defensible because it is the only cross-company, public, checkable brand measure; safe only if it stays reporting, not a rating (section 6). Cost: a style guide and discipline. First proof: a field league table with the method on the same page.
6. The yardstick for AI-era company health. What: Trillion Score and Fall Risk, with Altman, Piotroski, Beneish and Sloan shown, as the open benchmark assistants cite for "is this company healthy". Defensible because agents cite indexed, quotable, licensed sources and the paid vendors are gated. Cost: the citation style pass. First proof: a monthly citation count by assistant.
7. Independent review of the score as a published standard. What: a university or standards body reviews the method; the review is published and the score versioned. Defensible because review turns a house measure into a standard and nobody in the table has submitted theirs. Cost: time and humility. First proof: a METHODS document with a DOI and a named reviewer.
8. First to the new tags. What: read ESAP, ESRS-tagged and FDTA-standardised data the day it appears and publish the first open scores on it. Defensible because a daily reader with a working pipeline is months ahead of a vendor that must price and package first. Cost: taxonomy work in 2027. First proof: an LEI on every company page now.
9. Signed facts. What: adopt a vLEI and sign the ledger and the scores when the XBRL signature specification lands. Defensible because ISO 17442-3 and the FDTA point to verifiable identity, and a signed public record is the strongest form of "no back-dating". Cost: a QVI fee and some engineering. First proof: a signed hash of the forecast ledger.
10. The ad-free, no-pay-to-rank newsroom as the trust layer. What: keep the five desks ad-free and keep the disclaimer. Defensible because Perplexity now pays for citations and Cloudflare for use, the economics will push publishers toward sponsored inclusion, and the one that refuses is the one agents are told to prefer. Cost: forgone advertising. First proof: the partner code below, published.
6. The partner code
The primary codes converge. The CFA Institute Asset Manager Code asks managers to "Act with independence and objectivity", to communicate "in a timely and accurate manner" and to "Act for the benefit of clients" (professional body). The ILPA Principles rest on alignment of interest, governance and transparency (industry body). The PRI commits signatories to "Seek appropriate disclosure on ESG issues by the entities in which a signatory invests" and to "Report on our activities and progress" (industry body). The SEC Marketing Rule (Release IA-5653, 22 December 2020) prohibits untrue or unsubstantiated statements, misleading implications, unbalanced treatment of benefits and risks, cherry-picked performance and gross performance without net, and permits testimonials and endorsements only with clear disclosure of compensation and conflicts (regulator 1, regulator 2). The FTC's revised Endorsement Guides (June 2023) treat fake reviews as endorsements, require disclosure of material connections, and name procuring, suppressing, boosting or editing reviews "so as to distort what consumers think of a product" as deceptive (regulator). The house's line, "reporting, not advice; no company pays to be included or ranked", is consistent with all five.
Sanpo yoshi is the oldest of these. ITOCHU states it as "good for all three sides": good for the seller (urite-yoshi), the buyer (kaite-yoshi) and society (seken-yoshi), traced to the Omi merchants and to founder Chubei Itoh's motto that trade is a compassionate business (company source). The business-history record is more careful: Suenaga Kunitoshi's 1999 paper examines Nakamura Jihei's testament as the claimed original (academic), and Usami Hideki's view, summarised in a Kwansei Gakuin paper, is that the three-part slogan was coined by Ogura Eiichiro in the 1980s from Omi practice rather than found verbatim in any house code (academic). Quote it as a modern statement of an old practice, not as a seventeenth-century text.
The partner code, in ten lines
1. We never take payment to include, exclude, rank or score a company. Scores come from filed results and a published, versioned method.
2. We never hold or process personal data to produce a score or a program, and we never keep or claim ownership of a client's personal or business data; our inputs are public filings and public records.
3. We never back-date. Forecasts are hashed before the outcome and graded in public after it; corrections are logged, dated and visible.
4. We disclose every commercial relationship with a company whose score we publish, on the page where the score appears.
5. We promise no returns, make no recommendation and give no advice; partners may not present our scores as ratings, recommendations or investment advice.
6. Partners attribute our data as CC BY 4.0 requires, with the source URL and the date, and do not strip the method or the disclaimer.
7. No partner has exclusivity over public data. Programs, events and pieces can be exclusive; the yardstick cannot.
8. We keep clients' names and program details confidential unless the client publishes them or asks us to; receipts are published anonymised unless the client chooses otherwise.
9. We hold partners to the disclosure standard regulators hold marketers to: material connections disclosed, no fabricated or suppressed reviews, net results shown with gross.
10. Disputes go first to a written correction request, answered within five business days with the evidence; if we were wrong we say so on the page and in the ledger; if we disagree, both positions are published; unresolved disputes go to a named independent reviewer whose finding we publish.
7. On one page: ahead of the curve
Where the world is going. Company data is becoming a dataset with a legal identity attached: the FDTA joint standards took effect on 1 October 2026 with agency rules due by 2028, the LEI is on 3 million entities and is the identifier of record, and ISO 17442-3 means a filed fact will soon carry a signature chain from the person who approved it. AI assistants cite sources by default and are starting to pay for citations and agent actions, and their own guidance says the way in is to be indexed, plain and quotable, not to buy a deal. The agent-commerce protocols from OpenAI, Google, Visa and Mastercard have settled on one shape: a signed mandate against a structured product feed, and a merchant who can quote. The corporate-intelligence incumbents say the same thing in their 10-Ks: AI over proprietary paid data, with the stated risk that third parties will use that data to make their own insights. Brand measurement is retreating from personal data to aggregated models and from platform-reported figures to independent, auditable proof.
Where the house stands. It reads every filing daily, publishes free under CC BY with machine twins and a keyless MCP server, and keeps a hash-sealed forecast record scored in public, which no competitor has and which cannot be bought back in time. Its yardstick is the one number the CFO already accepts, the filed result, measured against the field, and it has no licence revenue to protect, so the incumbents' main risk is its plan. It builds the program, the event and the pieces through 1,400 makers without a plant, so it can close the loop from measure to work to receipt, which no data vendor, agency or platform is structured to do. The next steps cost little: an LEI on every page, one sentence per fact, Dataset and ClaimReview markup, a product feed an agent can quote, a methods document with a named reviewer, a signed ledger when the signature specification lands. Published with the partner code, these make the house the source agents are told to call and the one buyers can check, and that position compounds for twenty years.
All sources for the evidence
SEC, FDTA Joint Data Standards final rule page (Release 33-11420): sec.gov
Federal Register, FDTA Joint Data Standards, 25 June 2026: federalregister.gov
OCC Bulletin 2026-25: occ.gov
Federal Reserve staff memo on the FDTA final rule (PDF, not machine-read today): federalreserve.gov
SEC, EDGAR application programming interfaces: sec.gov
SEC, EDGAR Next filer information: sec.gov
SEC, Investment Adviser Marketing (Release IA-5653): sec.gov and press release sec.gov
FTC, Guides Concerning the Use of Endorsements and Testimonials in Advertising (16 CFR Part 255): ftc.gov
ESMA, ESMA launches data collection under the first phase of ESAP (10 July 2026): esma.europa.eu
Google Privacy Sandbox, Next steps for Privacy Sandbox and tracking protections in Chrome (22 April 2025): privacysandbox.google.com
XBRL US, Final Joint Rule for FDTA (9 June 2026): xbrl.us
GLEIF, The LEI in numbers Q1 2026: gleif.org
GLEIF, ISO standardizes vLEI (14 October 2024): gleif.org
ISO 17442-3:2024: iso.org
GLEIF, vLEI Ecosystem Governance Framework: gleif.org
XBRL International, Verifiable LEIs: digital trust you can bank on: xbrl.org
IFRS Foundation, ISSB proposes update to IFRS Sustainability Disclosure Taxonomy (July 2026): ifrs.org
EFRAG, ESRS XBRL Taxonomy project page: efrag.org
schema.org Dataset: schema.org ; ClaimReview: schema.org
llms.txt proposal: llmstxt.org
MCP specification 2025-11-25 changelog: modelcontextprotocol.io
MCP Registry preview (8 September 2025): blog.modelcontextprotocol.io ; terms: modelcontextprotocol.io
Linux Foundation, formation of the Agentic AI Foundation (9 December 2025): linuxfoundation.org
AP2 protocol site: ap2-protocol.org
OpenAI commerce developer docs (ACP specs): developers.openai.com
CFA Institute Asset Manager Code: rpc.cfainstitute.org
ILPA Principles: ilpa.org
PRI, What are the Principles for Responsible Investment: unpri.org
S&P Global FY2025 10-K: sec.gov ; Kensho factbook page: investorfactbook.spglobal.com
FactSet FY2026 10-K: sec.gov
Morningstar FY2025 10-K: sec.gov
Gartner FY2025 10-K: sec.gov
Palantir FY2025 10-K: sec.gov
Zeta Global FY2025 10-K: sec.gov
Sprinklr FY2026 10-K: sec.gov ; FY2026 results release: d1io3yog0oux5.cloudfront.net
Omnicom FY2025 10-K: sec.gov ; completion of IPG acquisition: omnicomgroup.com
Snowflake FY2026 10-K: sec.gov
C3.ai FY2026 10-K: sec.gov ; preliminary results and CEO change: c3.ai
Google Search Central, AI features and your website: developers.google.com
Google Meridian: developers.google.com ; github.com
OpenAI, Introducing ChatGPT search: openai.com
Anthropic, Citations: platform.claude.com ; Web search tool: platform.claude.com
Perplexity, Introducing Comet Plus: perplexity.ai
Stripe, Instant Checkout and Agentic Commerce Protocol (29 September 2025): stripe.com
Visa, Trusted Agent Protocol (14 October 2025): usa.visa.com ; Visa and partners complete secure AI transactions (18 December 2025): corporate.visa.com
Mastercard, Agent Pay (29 April 2025): newsroom.mastercard.com
ITOCHU, Mission (Sampo-yoshi): itochu.co.jp
ANA, Programmatic Media Supply Chain Transparency Study: ana.net ; Q1 2025 benchmark coverage: lbbonline.com
ANA, 2025 Trends in Agency Compensation: ana.net ; May 2026 client-agency relations: ana.net
IAB/MRC Retail Media Measurement Guidelines (January 2024): iab.com
ASI, North American promo industry grows 4.2% in 2025: asicentral.com ; Q2 2026 sales: members.asicentral.com ; 2026 State of the Industry: members.asicentral.com
PPAI, 2025 US distributor sales: ppai.org ; 2024 Sales Volume Report: ppai.org
IAPP, US State Privacy Legislation Tracker: iapp.org
Suenaga Kunitoshi, on Nakamura Jihei's testament and the origins of sanpo yoshi, Doshisha Shogaku 50 (1999), NDL record: ndlsearch.ndl.go.jp
Kwansei Gakuin University paper summarising Usami Hideki on the coinage of the slogan: kwansei.ac.jp
KPMG, Joint data standards interagency final rule: kpmg.com
Proskauer on the semiannual reporting proposal: proskauer.com ; CLS Blue Sky Blog, Atkins statement: clsbluesky.law.columbia.edu
DLA Piper on Omnibus I and CSRD: knowledge.dlapiper.com
Ahrefs, OpenAI publisher partnerships and citations (November 2026): ahrefs.com
Press Gazette on Comet Plus: pressgazette.co.uk
IT Brief on Cloudflare crawler controls (2026): itbrief.co.uk
Fortune on ChatGPT search launch: fortune.com
The New Stack on the OpenAI Apps SDK: thenewstack.io
PYMNTS on Google's Universal Commerce Protocol: pymnts.com
Benzinga, FactSet Q4 FY2026 call transcript: benzinga.com
Sheppard Mullin on the 2023 FTC Endorsement Guides: sheppard.com
Dechert, Overview of the PRI: dechert.com
O'Melveny, 2026 data security and privacy compliance checklist: omm.com