The evaluation is collapsing into somebody else's, and the moment a growing business used to switch is being deliberately removed.
01
02
Two mechanisms are pulling in the same direction, and they need testing separately because they fail separately.
The first is that the procurement frame is collapsing. Three AI-native platforms shipped in three weeks across three geographies with payroll as a feature inside treasury, expense and reporting rather than as a product in its own right. The accounting-practice platforms absorb it the same way, framing themselves as where the practice operates with payroll one automated flow among many. The consequence for a payroll vendor is not competitive but structural: the evaluation collapses into the finance-ops evaluation, and the procurement cycle, the customer acquisition cost and the renewal motion all collapse into a decision made somewhere else.
The second is that the graduation moment is being deliberately destroyed. The richest customer-acquisition event in this market has always been a scaling business outgrowing its entry accounting product and re-platforming onto a mid-market suite. Xero Ultra exists so that moment never arrives, priced and shipping and UK-bound. Intuit can subsidise HR features to retain accounting customers, which makes total-cost-of-ownership competition impossible for anyone selling HR alone. That is an attack on the feeder channel rather than on the product.
The defensible response is visible and someone is already running it. Deel has certified its payroll into Workday Global Payroll Cloud, SAP, UKG and NetSuite, positioning as the execution layer underneath the systems of record rather than a competitor to them. Payroll that exposes itself as the trusted execution layer becomes infrastructure others build on. Payroll that does not risks being the feature an HCM bypasses.
The agent channel amplifies it. Four accounting-adjacent ledgers are now live on Claude through MCP. If accounting ledgers are agent-accessible and payroll ledgers are not, the agent relationship flows toward accounting by default, and the payroll vendor is downstream of a conversation it cannot see.
What would change our mind. Ardian put roughly $1B behind HR Path, a bet that deployments and managed payroll stay complex enough to pay for a global services consolidator right through the agentic transition. That cheque does not get written if payroll is becoming a checkbox. If services attach holds as a share of first-year contract value, the collapse is not happening.
Transformed 18 September. The first mechanism said AI-native platforms would build payroll into treasury and expense as their own feature. The collapse is happening, but along a different path: the platform that already holds the customer, a bank account, an accountant's platform or an HR suite, sells payroll that someone else runs. American Express with Gusto, Mercury naming Gusto rather than its own acquired payroll, HiBob with Zalaris and Workday's sales team reselling OneSource Virtual all arrived within five days. The prediction and the test that would kill it are unchanged.
03
Two mechanisms, tested separately. (a) The procurement frame collapses into the platform that already holds the customer - the bank account, the accountant's practice platform or the HR suite - and payroll arrives inside it as a named, rented engine more often than as the platform's own build, so payroll evaluation collapses INTO that purchase rather than the other way around. (b) The graduation moment is deliberately removed: the richest acquisition event in the mid-market is a scaling SMB outgrowing its entry accounting product, and Xero Ultra exists to remove it while Intuit can subsidise HR to hold accounting customers, making TCO competition impossible for a standalone.
04
A payroll vendor that is no longer the lead decision faces rising CAC for the same logo, because it must win inside someone else's evaluation rather than run its own. The defensible response is to become the certified execution layer underneath - which Deel has done, certifying into Workday Global Payroll Cloud, SAP, UKG and NetSuite: payroll that exposes itself as the trusted execution layer becomes infrastructure others build on; payroll that does not risks being the feature an HCM bypasses. The agent channel amplifies it - if accounting-adjacent ledgers are agent-accessible and payroll ledgers are not, the agent relationship flows toward accounting.
05
Given equal weight to the claim. Hiding the counter is how a prediction becomes an article of faith.
06
07
Specific public numbers we check on a schedule, so this claim can be tested without taking our word for it. Not checked means nobody looked. That counts as nothing, never as agreement.
| What we check | Status | Latest reading | Last checked |
|---|---|---|---|
| Payroll engines certified into a third party's system of record vendor partner directories · scan | changing | Globalli certified into Workday Global Payroll Connect (five GPC components; page byline 15 Sept, dateline 17 Sept). Deel's set unchanged as recorded. Zalaris certified integration into HiBob (15 Sept). | 2026-09-17 |
| Priced stay-put tiers aimed at the graduation moment vendor pricing pages · event | changing | Xero Ultra still Australia-only on the UK media-releases channel (07 Jul wording), promised UK autumn beta not yet announced; Intuit UK names Accountant Suite and a mid-market sales team, no payroll SKU, no price | 2026-09-10 |
| Accounting-adjacent vs payroll ledgers exposed to agents Anthropic Connectors Directory, MCP registries · scan | changing | Justworks adds a PEO's payroll history, paystubs and deductions as agent-readable, read-only. Payroll-first vendors with WRITE MCPs unchanged (isolved, Deel). | 2026-09-21 |
| Professional-services revenue as a share of total at public HCM suites (proxy for services attach falling) Workday quarterly results (Dayforce and UKG are private) · quarterly | changing | Workday Q2 FY27: professional services $178m of $2.649bn total revenue = 6.7% | 2026-08-27 |
08
Cumulative linked signals, by direction. A line that only climbs in green is being read generously.
14 counted: 5 diverges · 9 supports · 1 predate the claim and are not counted
| Date | Signal | Bearing | On which part of the mechanism |
|---|---|---|---|
| 2026-08-05 | signal-deel-oracle-fusion-payroll | Before the claim (not counted) | Payroll engines certified into a third party's system of record grow faster than standalone payroll products winning net-new mid-market logos on their own evaluation. |
| 2026-08-12 | signal-xero-payroll-powered-by-gusto | supports | (a) The procurement frame collapses: payroll is evaluated inside another purchase (finance operations, the accounting practice, the bank account or the HR suite) rather than on its own. |
| 2026-08-12 | signal-intuit-midmarket-no-payroll | diverges | (b) The graduation moment is removed: a scaling SMB no longer re-evaluates payroll when it outgrows its entry accounting product. |
| 2026-08-19 | signal-xero-payroll-fronted-denver | supports | Payroll engines certified into a third party's system of record grow faster than standalone payroll products winning net-new mid-market logos on their own evaluation. |
| 2026-08-20 | signal-xero-agent-rail-no-payroll | supports | (a) The procurement frame collapses: payroll is evaluated inside another purchase (finance operations, the accounting practice, the bank account or the HR suite) rather than on its own. |
| 2026-08-20 | signal-warp-stripe-atlas-payroll-bundle | supports | (a) The procurement frame collapses: payroll is evaluated inside another purchase (finance operations, the accounting practice, the bank account or the HR suite) rather than on its own. |
| 2026-08-24 | signal-asure-aws-marketplace-cosell | diverges | (a) The procurement frame collapses: payroll is evaluated inside another purchase (finance operations, the accounting practice, the bank account or the HR suite) rather than on its own. |
| 2026-08-25 | signal-intuit-fy26-payroll-undisclosed | supports | (a) The procurement frame collapses: payroll is evaluated inside another purchase (finance operations, the accounting practice, the bank account or the HR suite) rather than on its own. |
| 2026-09-10 | signal-cegid-silae-merger | supports | (a) The procurement frame collapses: payroll is evaluated inside another purchase (finance operations, the accounting practice, the bank account or the HR suite) rather than on its own. |
| 2026-09-10 | signal-intuit-foster-uk-midmarket | diverges | (b) The graduation moment is removed: a scaling SMB no longer re-evaluates payroll when it outgrows its entry accounting product. |
| 2026-09-10 | signal-intuit-free-to-midmarket-one-platform | supports | (b) The graduation moment is removed: a scaling SMB no longer re-evaluates payroll when it outgrows its entry accounting product. |
| 2026-09-16 | signal-infoniqa-becomes-suiteify | diverges | (a) The procurement frame collapses: payroll is evaluated inside another purchase (finance operations, the accounting practice, the bank account or the HR suite) rather than on its own. |
| 2026-09-16 | signal-amex-embeds-gusto-payroll | supports | (a) The procurement frame collapses: payroll is evaluated inside another purchase (finance operations, the accounting practice, the bank account or the HR suite) rather than on its own. |
| 2026-09-16 | signal-mercury-books-ledger-in-the-account | diverges | (a) The procurement frame collapses: payroll is evaluated inside another purchase (finance operations, the accounting practice, the bank account or the HR suite) rather than on its own. |
| 2026-09-16 | signal-integral-ai-run-licensed-firm-funding | supports | (a) The procurement frame collapses: payroll is evaluated inside another purchase (finance operations, the accounting practice, the bank account or the HR suite) rather than on its own. |
09
The market moved along a different path than the one drawn. The prediction keeps its identity and its history. A narrative that changed is itself evidence.
Was: The number of payroll engines certified into a third party's system of record grows faster than the number of standalone payroll products winning net-new mid-market logos on their own evaluation.
Became: The number of payroll engines certified into a third party's system of record grows faster than the number of standalone payroll products winning net-new mid-market logos on their own evaluation.
Why: Part (a) said AI-native platforms would ship payroll as their OWN feature inside treasury and expense. Three scored divergences say the collapse is happening but along a different path: Asure reached buyers through a hyperscaler marketplace (24 Aug), Suiteify put payroll inside a DACH HR/finance/ERP suite (16 Sept), and Mercury built the ledger into the bank account while naming Gusto, not its own acquired payroll (16 Sept). Amex (15 Sept), HiBob with Zalaris (15 Sept) and Workday GO reselling OneSource Virtual (17 Sept) show the same shape: the platform holding the customer sells payroll that someone else runs. Proposed 17 Sept, done 18 Sept. The prediction and kill are unchanged; only the mechanism's path is redrawn.
7 signals carried forward; 2 dropped as no longer bearing on the claim: A hyperscaler marketplace is a procurement door (bench candidate procurement-door), not a platform that holds the customer; it bears on how payroll is bought, not inside what.; The agent-channel asymmetry is an instrument reading (ledger-mcp-ratio), not a test of the procurement path.
10
(a) is new - crowded out of the first cluster set by agent-weighted salience. (b) re-derives an earlier mid-market thesis and the W14 editorial of April 2026. Declared, dated 2026-08-12, not backdated.
This prediction carries no resolution date. We cannot predict when evidence will arrive, so the review cadence attaches to the instruments above rather than to the claim. It runs until the market proves it, moves it, or twelve months pass with no material signal against it.
How a signal becomes a prediction →