By Jonathan Nwosu, Product Manager
The month-end close is not a law of nature. It's a batching artifact: the shape finance work takes when reconciliation, matching, and eliminations are expensive human tasks that have to be saved up and done in one push. For most multi-entity companies that push consumes the first week of every month, and everything the business learns from its numbers arrives that much later.
Before arguing about how to shorten it, it helps to say plainly what the process actually contains.
The 7 steps
Most closes are the same sequence, whatever the company calls it internally.
1. Cut off the period. Stop new transactions landing in the month being closed, and agree what counts as in and out.
2. Reconcile the bank. Every statement line matched to a ledger entry, with the differences explained.
3. Match payments to bills and invoices. Which payment settled which obligation, including the partial ones, the batched ones and the ones that arrived net of a fee.
4. Chase what is missing. Receipts nobody submitted, documentation nobody attached, coding nobody confirmed.
5. Eliminate intercompany. Find both legs of every internal transaction, confirm they agree, and post the entry that cancels them at group level.
6. Post accruals, prepayments and deferrals. The entries that put costs and revenue in the period they belong to rather than the period they were paid or billed in.
7. Review, adjust and sign off. Judgment, estimates, and somebody attesting the period is final.
Steps 1 and 7 are genuinely periodic. Steps 2 through 6 are not periodic at all. They are continuous work that got saved up, and that is where the week goes.
The cost is the delay, not the work
A payment matched the day it lands takes seconds, because the remittance is recent, the reference is intact and the amount is recognisable. The same payment matched on day 26 requires somebody to work out what a round number from a vendor with 4 open invoices was settling.
Every step in the middle of that list has the same property. A missing receipt chased today is a Slack message. Chased 3 weeks later it is an archaeology project with a colleague who has genuinely forgotten. An intercompany recharge eliminated as it posts is arithmetic. Eliminated at month-end across 2 ledgers it is a reconciliation with a difference column.
Nothing about steps 2 through 6 is hard. They are only hard late, and the batch is what makes them late.
Work the moment it appears
Agentic accounting removes the reason the batch exists. Agents don't save work up. A payment lands and is matched against its bill within minutes. A receipt is missing and the chase starts the same day in Slack, not 3 weeks later from a spreadsheet of exceptions. An intercompany transaction posts and its elimination is drafted immediately. Reconciliations run continuously against live bank feeds instead of once, heroically, at month end.
Each task is small. That's the point. The close was only ever a mountain because 30 days of small tasks were left to accumulate.
What's left of "the close"
Not nothing: review remains. Judgment calls, estimates, accruals that need a human decision: those still happen on a cadence. But they happen against books that are already current, so the close stops being reconstruction and becomes confirmation. Teams running this model, like Oper Credits, watched a half-week consolidation grind become something that posts as transactions happen.
The numbers move in the same direction everywhere it has been tried. Tillo went from 12 days to 5 across 4 entities and 23 currencies. All Gravy closes 4 countries in 5 days with a finance team of 3. Ocean.io cut close time 60%. Alva Labs is targeting a same-day close.
Why this needs one ledger
The continuous close has an architectural precondition: the agent has to see the whole transaction. If AP lives in one tool, expenses in another, and the ledger in a third, no agent can match, chase, and post across the gaps, and the batching returns at every system boundary. This is why the continuous close is a property of agentic accounting platforms rather than a feature that can be added to a Frankenstack.
The close was always a workaround; agents remove the need for it and give finance back its first week.
See how Consolidation posts continuously across entities in Light. For the buying decision, read what to evaluate in close software and why the batch itself is the thing to remove.