A calendar, not a spreadsheet
Every task with an owner, a due day, a dependency, and a state. The list is the same each month, which is what makes it improvable rather than heroic.
Platform · financial core
Close duration is the metric that degrades quietly — nobody notices the month it went from eight days to eleven, because it happened over four quarters. A structured close with agents doing the mechanical work makes the duration a number you manage rather than a thing that happens to you.
Tell us your size, entities, and stack. We will tell you where the days are actually going.
What it does
Every task with an owner, a due day, a dependency, and a state. The list is the same each month, which is what makes it improvable rather than heroic.
AR and AP must agree with their control accounts before the period can advance. A break blocks progression rather than producing a note nobody reads.
Reconciliation, accrual schedules, depreciation, rev rec runs, and intercompany elimination are prepared automatically and held for review.
Period-over-period movement explained by the transactions that caused it, with a draft narrative you edit rather than write.
Closing a period is an assertion by a person and is never delegated to an agent at any authority level. The signature means something because it cannot be automated.
Every task carries its supporting records, so the close file an auditor asks for already exists rather than being assembled in March.
Finance teams usually assume the close is slow because of the volume of work. In the closes we have instrumented, it is nearly always waiting — for a bank feed to populate, for a manager to approve an accrual, for someone to answer a question about a variance, for the last three bills to arrive.
That distinction matters because it changes what fixes it. More people does not shorten a close dominated by waiting. Removing the dependency does — bank feeds reconciled continuously rather than on day one, AP cut-off enforced by policy rather than by chasing, accrual schedules prepared before the period ends rather than after.
The second largest category is rework: something did not tie, and the investigation consumes a day. Subledger tie-outs running continuously mean the break surfaces on the day it occurs, when the transaction is recent and someone remembers it, rather than on day four of the close when nobody does.
Bank reconciliation, accrual and prepaid schedules, depreciation runs, revenue recognition, intercompany elimination, and the chasing. All of it is prepared and held for review — Level 1 by default — so your controller spends the close reviewing judgements rather than performing arithmetic.
The chasing is worth calling out separately because it is the least glamorous and most effective piece. The close agent knows which task is blocking which, who owns it, and how long it has been sitting, and it follows up without anyone having to feel awkward about following up.
Closing a period asserts that the numbers are right. That assertion has a person behind it — for an audit, for a bank covenant, for a board. Delegating it to software would make the assertion meaningless, and every gain from automating the work leading up to it depends on that final step remaining a judgement.
So sign-off is absent from the agent permission model entirely. Not a default that an administrator can lower — there is no configuration that produces it, which is the same treatment we give releasing funds.
Six days instead of thirteen is a week of finance capacity returned every month, which is the visible benefit. The larger one is decision timing: numbers landing on the sixth rather than the twentieth means three extra weeks each month where operational decisions are made on actuals rather than on instinct.
Questions
Send your checklist and last three close dates and we will show you where the days are going.