A variance outside tolerance, a missing PO, a mismatch with no clean answer. This is the case that used to sit in a queue.
Finance operations
The exceptions your rules engine already throws back.
Across P2P, O2C, R2R, tax and treasury, most of the volume is already handled. Rules match it, post it, close it. What lands on a person's desk is the fraction rules were never written for. That fraction is where we build.
Where the work actually goes
01
Procure to pay
Highest volumeThree-way match handles most invoices. The agent takes the rest: price and quantity variances outside tolerance, duplicate submissions, invoices with no matching PO at all.
- Variance reasoned against contract terms, not just tolerance bands
- Vendor query drafted with the evidence attached, sent, and chased
- Duplicate and split-invoice patterns flagged before payment, not after
02
Order to cash
Highest volumeCash comes in against the wrong invoice, the wrong amount, or no remittance at all. Someone has to read the bank statement, the email, and the ledger, and decide. That decision is where the agent starts.
- Short and split payments matched across invoices with a stated reason
- Dispute reasons read from the customer's email and coded consistently
- Credit hold cases assembled with payment history, not left to sit
03
Record to report
Strong fitClose-cycle work is a sequence of small judgements under a deadline. The agent assembles the schedules, flags what looks wrong, and leaves the posting decision where it belongs.
- Intercompany mismatches surfaced with both sides of the entry shown
- Flux review notes drafted from the prior period, not from scratch
- Journal entries proposed and cited, never posted without sign-off
04
Tax
Evidence work onlyIndirect tax coding and jurisdiction checks generate a steady stream of small mismatches. The agent finds them and builds the case. A person still files.
- Tax code mismatches flagged against the invoice line, not the batch
- Jurisdiction and rate changes cross-checked before the return is due
05
Treasury
Strong fitCash positioning and bank reconciliation break in small, repeatable ways: a fee not booked, a timing difference, an FX rate applied late. The agent finds the break and shows its working.
- Bank reconciliation breaks traced to a specific transaction, not a total
- FX exposure flagged against policy limits before the position moves
01
The intelligence layer
Rules handle the standard path. This is what happens on the rest of it.
Variance tolerance · illustrative, not a measured result
1
The rule declines it
2
The agent gathers context
The PO, the contract, the email thread, the prior exception on this vendor. Pulled from every system, without a ticket.
3
It proposes, with sources
A recommended action, every source cited. Nothing is posted on the agent's own authority.
4
A person decides, it learns
Approval or correction becomes the pattern the next exception is checked against.
02
What the industry measures
These are not our numbers. They are the ones the research firms publish.
We have not published a client result yet. Until we do, the honest thing is to show you what the benchmark studies say about accounts payable and let you place your own team against it.
Invoice exception rate
the slice we build on
Ardent Partners,AP Metrics That Matter, 2025
Touchless processing rate
posted with no one touching it
Ardent Partners,AP Metrics That Matter, 2025
Invoice cycle time
receipt to approved
Ardent Partners,AP Metrics That Matter, 2025
Cost per invoice
the sources disagree. see below.
APQC via CFO.com;Ardent Partners, 2025
Read these as direction, not as a target. APQC and Ardent Partners report different cost-per-invoice figures because they survey different samples, and most benchmarks skew to mid-market and enterprise. If you process under 200 invoices a month your unit cost will look worse than any of these, for reasons that have nothing to do with your process. Your own baseline tells you more than the cross-industry median does. We would rather you measured it than took our word for it.
03
Process fit
Where an agent earns its place, and where it does not.
This is our assessment, not a measurement. Four things decide whether a finance process is worth building an agent for. Darker means a stronger case.
EXCEPTION
VOLUME
VOLUME
RULES ALREADY
WRITTEN
WRITTEN
SPREAD ACROSS
SYSTEMS
SYSTEMS
LOW LEGAL
EXPOSURE
EXPOSURE
OUR READ
Procure to pay
START HERE
Order to cash
STRONG
Record to report
GOOD, SCOPED
Treasury
GOOD, SCOPED
Tax
EVIDENCE ONLY
STRONG CASE
PARTIAL
WEAK
COUNTS AGAINST
Where we say no
Anything with tax or regulatory exposure gets a recommendation with evidence, never an unattended posting. Anything with no documented policy behind the judgement stays with the process owner until one exists. And anywhere a scheduled job already fixes it for less, we say so.
04
How we engage
Three steps. You can stop after the first one.
Discovery
We sit with whoever actually clears the exception queue and watch a normal week. You get a written assessment either way, including the case for not building anything. If a scheduled job fixes it cheaper, we tell you that and we are done.
Build
One process, scoped and priced before we start. Built against your systems and your rules, not a demo environment. You see it running on your own data, not on ours.
Handover
Code, documentation and runbook handed to your team. No licence, no lock-in, no retainer you have to keep paying to keep it working. Keep us on if you want more built.
Build pricing is set per engagement and quoted in writing before any work starts.
Which queue is this?
Tell us the process and the system it lives in. We'll tell you honestly whether an agent clears it.