Finance workspace where close, payables and receivables data live in separate systems
Oracle Fusion · Payables

Oracle Fusion Payables
Automation in 2026

Fusion has a generally available Payables agent. The invoices sitting on hold in your queue are still the ones nobody has researched.

Short answer

Oracle made a Payables agent generally available in Fusion Release 26B, bundled into existing subscriptions. It assists with payables work inside Fusion. The invoices that consume real time are the ones on hold for price variances, missing receipts or suspected duplicates, and resolving those requires research across the purchase order, the contract and the receipt history.

Key takeaways
  • The Payables agent reached general availability in Release 26B and is bundled rather than separately licensed.
  • Capture and coding are largely solved. Holds and exceptions are where payables time goes.
  • Resolving a hold means research across the order, the contract, the receipt and prior vouchers.
  • Enable the native agent first, then measure the hold queue that remains.

Payables on Fusion has never lacked structure. Purchase orders, receipts, tolerances, holds and approval hierarchies are all there, and the addition of a generally available Payables agent in Release 26B added assistance on top of that structure at no separate licence cost.

Open the hold queue on any Fusion instance and the picture is familiar. Invoices sit against price variances, quantity mismatches, missing receipts and suspected duplicates. Each one is a research task, and research is precisely what a structured hold mechanism was never designed to perform.

Oracle FusionAccounts payableHoldsMatchingDuplicates

The five holds that consume a Fusion payables team

Each section covers one hold type, what causes it, and what resolving it automatically actually involves.

01

Price variance against the purchase order

An invoice arrives above the ordered price, the tolerance rejects it, and someone has to decide whether the difference is a legitimate contractual escalation, a freight charge that was always expected, or a supplier testing whether anyone checks.

How it runs

The agent reads the purchase order line, the contract terms and the invoice, assembles the comparison and checks the supplier's history of similar variances. The finding arrives with all three sources attached, so the decision takes seconds rather than a walk across three screens.

The variance arrives with the order, the contract and the history already assembled.
02

Quantity mismatch and partial delivery

Goods arrive in two shipments, the invoice covers the full order, and the receipt against which it should match does not exist yet. The invoice sits on hold while somebody works out whether the balance is still coming.

How it runs

Receipt history is read against the order, the partial delivery pattern is identified, and the open balance is shown with the expected remainder. Where the pattern suggests a receipt was never posted rather than never delivered, that is flagged separately because the fix is different.

Distinguish the two cases. Undelivered goods and unposted receipts need different actions.
Show the open balance so the approver is not guessing at what remains.
Payables tasks from receipt to posted voucher, and who does each one.
Payables tasks from receipt to posted voucher, and who does each one.
03

Suspected duplicates that hide behind a changed reference

Duplicate detection based on exact invoice number matching catches the obvious cases and misses the expensive ones, where a supplier resubmits with a modified reference or a slightly different date.

How it runs

Comparison runs across amount, supplier, order reference and line detail rather than on the invoice number alone, and prior vouchers are shown side by side with the differences highlighted. The decision stays with a person because occasionally the second invoice is genuine.

Duplicates are found by comparing the substance, not the reference field.
04

Invoices with no purchase order at all

Professional fees, utilities, insurance, regulatory charges and anything urgent frequently arrive without an order, and the entire matching structure has nothing to match against.

How it runs

Coding is proposed from your own history and policy, the charge is checked against any contract on file and against its own pattern, and anything that broke that pattern is surfaced even though the coding is technically valid. Approval routing follows the policy rather than the tolerance.

Pattern checks catch what coding cannot. A valid code on a forty percent increase is still a problem.
Contract checks catch the rest, including billing that continued past a term.
Native assistance compared with agent led exception research.
Native assistance compared with agent led exception research.
05

Supplier bank changes, the control that actually matters

The expensive payables failure is not a miscoded charge. It is a payment sent to an account that changed because of a plausible email referencing a genuine overdue invoice.

How it runs

Every bank detail change is treated as a control point rather than a data update. The change is surfaced against the existing vendor record, verification through a channel the requester does not control is required, and the approval is logged. Changes arriving close to a payment run are flagged specifically.

A bank change is a control event, not an edit, and it should be evidenced like one.
One layer across payables, receivables and close on a Fusion estate.
One layer across payables, receivables and close on a Fusion estate.
"

Structured holds tell you an invoice failed. They were never designed to tell you why, and why is the entire job.

Ashok Manthena, CEO and co-founder of ChatFin

Clearing a Fusion hold queue in four weeks

Ordered so the largest hold category is addressed first.

1

Categorise the queue

Sort current holds by type and count the minutes each category takes to resolve, which gives you the real business case.

2

Enable the native agent

Turn on what your subscription includes and run a full cycle, recording what it cleared and what it handed back.

3

Automate the largest category

Start with whichever hold type consumes the most time, usually price variance or missing receipts.

4

Add the control checks

Bring vendor bank changes and pattern breaks under evidenced review, which is where the expensive failures live.

Fusion payables automation: common questions

Does Oracle Fusion have an AI payables agent?
Yes. A Payables agent reached general availability in Release 26B, bundled into existing Fusion subscriptions rather than licensed separately. It assists with payables work inside Fusion, so exception research and the resulting corrections still involve your team.
How do you resolve invoice holds faster?
Automate the research rather than the routing. Each hold type needs a specific comparison assembled: the order and contract for a price variance, receipt history for a quantity mismatch, and prior vouchers for a suspected duplicate. When that arrives assembled, the decision takes seconds.
How do you catch duplicate invoices reliably?
Compare substance rather than the invoice number. Matching on amount, supplier, order reference and line detail catches resubmissions with modified references, which exact number matching misses and which are the expensive cases.
What is the biggest payables control risk?
Supplier bank detail changes. Treat every change as a control event requiring verification through a channel the requester does not control, log the approval, and flag changes that arrive close to a payment run.
How ChatFin fits

Research the hold, not just route it.

ChatFin connects to Oracle Fusion through supported interfaces, matches invoices against orders and receipts, and researches every hold until it can present a finding with the order, contract, receipt history or prior voucher attached. Vendor bank changes are treated as evidenced control events.

The corrected voucher or documented hold goes to a named approver, and the same layer covers reconciliation and cash application. Bring your current hold queue and we will work it live.

Book a demo

Sources, images and disclosure

ChatFin publishes this article and sells finance automation software, so treat product claims as a vendor point of view and verify vendor capability and release status directly before you rely on it. Release and capability details reflect published vendor documentation as of August 2026 and change frequently, and features marked as previews are subject to change by the vendor.

Photography from Pexels, used under the Pexels License, which permits free commercial use without attribution. Credit is given here as a courtesy: source photo on Pexels. Diagrams are original ChatFin artwork and remain the copyright of ChatFin. Product names and logos referenced on this page are trademarks of their respective owners and are used for identification and comparison only.

Get started

Bring your hold queue.

Price variances, missing receipts and suspected duplicates from a real week, researched live during the session.

✓Watch agents run a real finance workflow on your own stack, not a slide
✓Bring one of your own processes and we will map it end to end
✓Leave with a 30 day plan to deploy your first agents
Request a demo
Confidential & private