pitch.amortizations.finance
Set once. Runs to zero.
Prepaid, deferred-cost, and intangible amortization schedules designed as typed calls: opening balance, period charge, remaining life, and the posting the schedule implies. Judgment once, at inception; arithmetic every period after. Inside the close suite. Nothing here is live yet, and this deck says so plainly.
↓ scroll · arrow keys
Every close has the tab. Prepaid insurance, the annual software invoice, deferred costs, capitalized commissions, the intangibles from the acquisition: each one set up honestly at payment time, each one amortized by a formula dragged down a column ever since. The arithmetic is the easiest in the building. The tab fails anyway, and always the same ways:
Amortization is judgment once, at inception, and arithmetic every period after. Deterministic work that drifts is not a math problem. It is a record problem: the schedule was never a record. It was a tab.
Now the agent arrives. An agent runs this arithmetic trivially, which is exactly the danger: clean computation against a drifted schedule is the old failure at machine speed. What the runoff needs from software is not faster division. It is a standing record: a schedule whose support is attached at inception, whose opening balance is the balance the ledger actually carries, and whose posting arrives as a prepared entry rather than an unattributed number.
Three schedule classes, matching the register at the apex: prepaid (insurance, subscriptions, anything paid ahead of its period), deferred cost (costs capitalized to be recognized over their term), and intangible (acquired and finite-lived, amortized to zero). One designed shape:
// the contract as designed · nothing at this domain is callable today
POST /v1/close/amortize
{
"class": "prepaid", // prepaid | deferred-cost | intangible
"account": "1420 · Prepaid insurance",
"inception": { "paid": "dated", "term": "stated in periods", "support": "attached" },
"period": "2026-06"
}
// designed response: the computed schedule · a read, not a posting
{
"schedule": {
"opening": "the balance the ledger actually carries",
"charge": "this period's computed charge",
"remaining": "periods left until the balance reaches zero",
"posting": { "implies": "a journal entry, prepared", "posts": "only via a recorded Authorization" }
}
}
The narrowness is the honesty, and it is inherited from the catalog of record, not invented here. This name answers for exactly one registry row, a data read that returns a computed schedule: opening balance, period charge, remaining life, and the posting the schedule implies. It does not post and does not roll forward. The posting is a separate act this row does not grant: an entity that wants it composes this read with the journal-entry row, under the posting boundary, on a recorded Authorization from a named party. A worker that posts on its own authority is not a faster close; it is an unattributed number. The entity closes its own books, always.
Everything on this slide is design intent. No schedule SKU exists, no call is servable, and the register at the apex says "Nothing at this domain is live." The contract ships as a servable surface only behind the family's publish gates, with its price posted on the family card in the same commit.
"Amortization" names two different objects in this family, and the family's catalog law keeps them apart on purpose. The live hub card posts a loan-amortization row: schedules on a Loan, Curtailment recalculation, per-diem interest, Payoff projection. That is a different computation on a different object, and this door may never render that row's price, quote it, or link it as though it were this capability. The canon calls a close-suite page rendering that live row's price the sharpest liveness violation this campaign could commit, precisely because the row is genuinely live, for a different thing. So this deck describes the neighbor without its figures, on purpose, and claims nothing from its liveness.
The apis.finance rate card serves today and posts a live row named Amortization math whose object is the Loan: schedules, Curtailment recalculation, per-diem interest, Payoff projection (curl-verified 2026-07-30). That row is not this capability, its price is not rendered here, and nothing about this door's liveness follows from it.
Agents can't sign an opinion. They can run the schedule. That is the vertical's ratified declarative shape applied to this door's own boundary. The amortization schedule is not a reserved act: nobody forbids an agent, or a controller, from dividing a prepaid balance over its term. The opinion is reserved: expressing assurance on financial statements belongs to a licensed firm under the state accountancy acts, signing a tax return to a credentialed preparer, representation before a taxing authority to a Circular 230 practitioner, investment advice to the advisers regime. A call that reaches any of the four is designed to return a typed BLOCKED with a cure that names the act, the reason, the party who may lawfully perform it, and the route. The platform performs none of the four, and neither does any pool.
Where a schedule surfaces judgment that needs a credentialed reviewer (an impairment question is not an arithmetic question), the route is typed, and today it is also honestly refused, because the accounting supply face has not opened:
gigs.accountants serves its RESERVED register leaf (curl-verified 2026-07-30): "Credentialed accounting professionals staffing routed books and close work. Nothing at this domain is live." The credentialed route this deck names is typed and refused, not staffed.
| instrument | returns | meter | price |
|---|---|---|---|
| Amortization schedule (prepaid · deferred cost · intangible) | the computed schedule: opening balance, period charge, remaining life, the posting it implies | no record, no charge | Price posts with the SKU |
| The posting the schedule implies | a journal entry, prepared: the flagship catalog’s row, never this row’s grant | posts only via a recorded Authorization | Price posts with the SKU |
Flat per record, fixed at post, posted on the family's one rate card, and released on the record produced. A call that fails its type check or cannot tie its opening balance to the ledger produces no artifact, an honest typed refusal, and meters nothing. No seats, no minimums, no percentage of anything, and no fee that varies with how fast a close landed. Until a SKU posts, every price cell above reads exactly what the vertical's publish gates require it to read. And per the collision law one slide up, no figure from the hub's live loan-amortization row appears at this door, in a price cell or anywhere else.
The fourth ProofPredicate was written for data products. Whether a close work product, here a computed schedule, is a served record is an open owner question in the vertical's canon, recommended for ratification rather than minting a fifth predicate. Until it is ruled, the meter promise here is design intent, and no SKU ships against it.
substrate — apis.finance
the controller who owns the close
the whole close, calendar grain
the agent proposing period-end estimates
accruals: proposal, posting, reversal
the agent tying out accounts at close
reconciliations as typed calls
the agent carrying the balance-sheet runoff
amortization schedules as typed calls
the builder embedding the family
one key, posted prices, the money kernel
A brand here is one ICP and one motion, chosen explicitly, and the instrument doors divide by rule rather than by vibe. The accrual is a period-end estimate: judgment at every close, proposed and then reversed. The reconciliation is a period call: two sides tied out at close, each month anew. The amortization schedule is neither: a standing record set once at inception and run deterministically to zero across periods. Same hero as the flagship, different grain, different caller on the wire, and no door claims another's instrument. The filing is worn openly rather than smoothed over: the vertical's owner ruling files this name among the work-product aliases that 301 into the flagship once the flagship ships. The flagship has not shipped, the apex already posts a door-grain identity, and this record proposes the runoff-grain door as a graduation, on the estate's own precedent. Ratify or reverse; the register stays honest either way.
monthend.finance, the flagship, serves its RESERVED register leaf (curl-verified 2026-07-30): "Nothing at this domain is live." The ruling that would 301 this name into the flagship has not triggered.
accruals.finance, the sibling instrument door, serves its RESERVED register leaf (curl-verified 2026-07-30): "Reserved for accruals, posted as typed calls. Accrual proposal, posting, and reversal as part of the month-end close suite. Nothing at this domain is live."
reconciliations.finance, the other sibling instrument door, serves its RESERVED register leaf (curl-verified 2026-07-30): "Account reconciliation with evidence attached: bank, subledger, intercompany, and suspense, inside the close suite."
apis.finance serves today: "Agents can't lend. They can call it." The family hub with the Mandate specimen, the key funnel, llms.txt, and the rate-card frame any close-suite SKU would post on. One key is designed to open every live door in the family.
The apex serves the family register's own leaf (curl-verified 2026-07-30): "amortizations.finance · RESERVED · apis.finance family register. Reserved for amortization schedules as typed calls. Nothing at this domain is live." The identity this deck argues is already posted at the apex, in the estate's one honest word.
Until the close-suite ruling lands, no close-suite copy may claim a capability, and this deck claims none: every capability slide above is design intent, and every suite domain remains a holding leaf.
One seam, worn openly. The apex leaf's ratified reserved-for sentence reads "Prepaid and intangible amortization schedules computed, posted, and rolled forward inside the close suite": three verbs. The catalog of record grants this name exactly one row, a data read that returns the computed schedule and neither posts nor rolls forward; the campaign's own copy corrections already recut the flagship's register line for this exact overreach. This deck adopts the row's narrow grant and states the wider sentence as the seam it is.
The graduation this record proposes, a work-product door at the runoff grain rather than a 301 alias, is a filing decision the owner has not made. It is recorded in this file's frontmatter reconciliation note and queued; the deck's green Claims stay curl-true under either outcome.
The register at amortizations.finance already says what this door is reserved for, and keys@apis.finance, the register's own door line, is where you hear when it opens.
If this was forwarded to you: amortizations.finance is the reserved runoff door of an accounting close suite. Prepaid, deferred-cost, and intangible amortization schedules designed as typed calls: the schedule computed with its opening balance read against the ledger, its period charge, its remaining life, and the posting it implies prepared for a named person to authorize, never posted on the software's own authority. Nothing at the domain is live; the deck above states every capability as design intent and carries the evidence for every liveness fact it posts. The live front door of the family is apis.finance. To hear when this one opens, write to keys@apis.finance.