Finance, accounts payable and back office

The ledger still needs a person. The keying does not.

Finance back office work is high volume, rule dense and unforgiving of error, which is the combination that makes it both a good candidate for an agent and a bad candidate for an unsupervised one. Everything on this page is built approval-gated by default, because a finance team that cannot show its controls to an auditor has not saved anything.

Book a scoping call Try the 2 minute fit check

Four workflows, in your vocabulary

The work, not the buzzword.

Each one says what arrives, what the agent does with it, where a person still stands in the way, and what gets counted afterwards. If a workflow cannot be written that way, it is not ready to be built.

Supplier invoice to purchase order matching

What arrives: Supplier invoices by email and portal, in every layout your supplier base produces, some with line items that match the purchase order and some that do not.
What the agent does: Reads each invoice, extracts supplier, dates, line items, GST and totals, codes it against your chart of accounts and matches it line by line to the purchase order and the receipt.
Where the human gate sits: Approval and payment release stay with a person. Anything outside your tolerance is queued as an exception with the discrepancy shown side by side rather than pushed through.
What gets measured: Cost and handling time per invoice, the share matched without intervention, and the exception rate, all against a baseline taken from your finance system.

Reconciliation exceptions

What arrives: A bank or subledger reconciliation where most lines match cleanly and a tail of them do not, currently worked through by one person against several screens.
What the agent does: Clears the routine matches, then assembles each remaining exception with the candidate matches, the supporting documents and the reason the automatic match failed.
Where the human gate sits: The agent proposes and never posts. Every journal or adjustment is created as a draft for the person who owns the reconciliation to approve.
What gets measured: Hours per reconciliation cycle, the number of unreconciled items carried forward, and the age of the oldest open item.

Debtor follow-up

What arrives: An aged receivables ledger where the follow-up happens when someone has time for it, which is rarely the same week twice.
What the agent does: Chases overdue accounts on the schedule you set, attaches the invoice and statement, records the response against the account and stops the moment payment lands.
Where the human gate sits: Escalation thresholds, tone and any account you mark sensitive are yours. Stopping supply, applying interest and anything with a contractual consequence stays with a person.
What gets measured: Days sales outstanding, the share of the ledger over your ageing threshold, and promise-to-pay conversion, all already tracked in most finance systems.

Month-end pack preparation

What arrives: The same close checklist every month: accruals to check, schedules to roll forward, variances to explain, inputs pulled from the same handful of systems.
What the agent does: Runs the checklist, gathers and reconciles the inputs, drafts the variance commentary against prior period with each figure linked to its source, and lists what is missing rather than carrying last month's number forward.
Where the human gate sits: The financial controller reviews and signs the pack. The agent prepares, it does not certify, and it does not close a period.
What gets measured: Days to close, the number of post-close adjustments, and hours spent preparing rather than reviewing.

The results we can actually evidence

No invented finance study. Here is how yours would be measured.

faster intake to instruction

45% Client's own tracked operating KPI, before and after the same definition

per land sale contract reviewed

1-2 hrs to ~10 min Client's existing time tracking, same review step before and after

Both measured results come from legal and conveyancing clients, against metrics those firms were already tracking. We will not publish an accounts payable figure until an accounts payable client produces one. The method is the part you can hold us to now: cost per invoice taken from your finance system over a three month baseline and a three month post-deployment period, on the same cost definition both sides, with the exception rate counted in the after figure rather than excluded from it. Your finance system already holds the baseline.

Read how both were measured

Audit, segregation of duties and the controls you already have

What the sector will not let an agent do.

The reason finance automation projects stall is rarely accuracy. It is that nobody can explain the control environment afterwards. This is how each control survives the deployment.

Segregation of duties: The agent prepares and a person approves, which is the separation your auditors already expect between the person who enters and the person who authorises. The agent holds no approval authority, and the log records both sides of every transaction.
An audit trail that stands up: Every action is logged: what ran unattended, what waited for approval, who released it, what escalated and which version of the rules applied. The trail is built as the work happens rather than reconstructed when an auditor asks.
Payment release and delegation of authority: Nothing is paid by an agent. Payment release stays with a person under your existing delegation limits by default, and we would advise leaving it there permanently however accurate the matching becomes.
Supplier fraud and changed bank details: A change to a supplier's bank details is a stop condition, not a field to be updated. It is flagged, held and routed for out-of-band verification by a person, because that specific change is where the loss actually happens.
Tax and statutory accuracy: GST treatment, coding rules and statutory positions are captured during solution design and applied consistently, but the return is lodged by a person. The agent escalates what falls outside the rules rather than guessing at a treatment.

This is a description of how the deployment is built, not legal, financial or compliance advice. Your own obligations should be confirmed with your own advisers. The security page sets out hosting, model choice, retention and access in full.

How it starts

A pilot, not a transformation programme.

One workflow, a baseline agreed before anything is built, an agent running supervised inside the tools you already use, and a keep-or-kill number at the end. Pilots start from $5,000.

Book a scoping call Try the 2 minute fit check

Common questions

The ones people actually ask.

Would it pay anything?

No. Payment release stays with a person by default and we would advise keeping it there permanently, however accurate the matching becomes. The agent prepares the payment run, it does not authorise it.

How does it handle segregation of duties?

The agent prepares and a person approves, which preserves the separation your auditors expect between entry and authorisation. The agent holds no approval authority and the log records both sides.

Our coding rules are unusual.

Most are. Rules are captured during solution design and applied consistently, and the agent escalates what falls outside them rather than guessing at a treatment.

Where would we start?

One invoice type or one exception category, with the current cost per item and handling time captured from your finance system before anything is built.

Where else this shows up

Document-heavy operations Professional services Melbourne and Victoria

All industries Use cases by workflow Security and trust Case studies How we work