How an AI Agent Works a Reconciliation Break

Reconciliation is a useful process for seeing how an agent works, because most of the job is already automated and the part that isn't is easy to describe. Software matches the bulk of settlement lines to the bank and the ledger on amount and reference. The time goes on what's left over, called breaks. Each break has to be traced to a cause, valued and fixed, and that means opening several systems.

If you're new to the terms model, copilot and agent, the first piece in this series explains them.

The break

A merchant's matching software flags that its acquirer's settlement for 12 October is £1,240 short of what the ledger expects. Worked by hand, an analyst downloads the settlement file, searches the ledger, pulls up the contract and may end up emailing the acquirer. Here's what an agent does with the same case.

1. Read the break

The matching software puts the case in the agent's queue. The model reads it: which acquirer, which date, how much is missing.

2. Decide what to look up

The model picks a tool and the inputs to give it, for example "get the settlement lines for batch 4471". The tools are small pieces of code the firm wrote: fetch the settlement lines, search the ledger, read the fee schedule in the contract, get the bank line. The model chooses which to call and in what order.

3. Read what comes back and pick the next step

The settlement file shows three chargebacks totalling £1,050 and a fee line. The model searches the ledger and finds no entry for the chargebacks. It reads the contract, which sets the fee at 0.25%, and sees the statement charged 0.30%. Code works out the difference: 0.05% on £380,000 of sales is £190. Together that's the full £1,240.

4. Explain the cause and the amount

The model writes up the break in plain language. Each number points back to the record it came from, so a reviewer can check the explanation in a minute or two.

5. Draft the fix and queue it

The agent drafts a journal for the three chargebacks and a £190 fee claim to the acquirer. Both go into an approval queue. Nothing is posted or sent until a person approves it.

Steps 2 and 3 repeat until the model has a cause and an amount. A simple break might need two or three lookups, and a messy one ten or more.

What the model did and what code did

The model did the reading and the deciding. It read a settlement file laid out the way this acquirer lays it out, matched the fee line to a clause in the contract, chose what to look up next and wrote the explanation. Rules struggle with this part, because every provider's file and every contract is a bit different.

Code did the sums, the data access and the limits. The £190 was calculated in code, because models still make arithmetic mistakes. The model only saw what the tools returned. And the tools were read-only, so the agent couldn't post the journal or send the claim itself.

Where a person comes in

The approval queue is where the firm keeps control. A reviewer sees the break, the explanation, the records behind each number and the drafted fix, and approves, edits or rejects it. Every lookup the agent made is logged, so an auditor can follow the case later.

The firm sets these limits, and the model can't change them. Hence an agent can be given a lot of room to investigate, across files, ledgers and contracts, while money still only moves when a person approves.

What drives the cost

An agent calls the model once for each lookup and again for the write-up, and every call reads the case so far. A break that takes eight lookups costs several times as much in model fees as one that takes one or two. So the cost per case is the number of calls times the price per call, and a cheaper model that needs more attempts, or gets the cause wrong and has to start again, can end up costing more than a dearer one.

The model also decides how good the explanation is, and a wrong explanation that a reviewer approves costs far more than the model fees. Hence two questions matter when you buy or build: which model does the product use, and what does it cost per case at your volumes?

Step 3 is the reasoning core of the loop: reading the records and finding and valuing the break. That's the step I test models on. In my first test, 34 models read UK card acquiring statements, and I scored each one on accuracy in pounds and cost per statement.

Where reconciliation agents are today

Several reconciliation vendors started selling agents in 2026. BlackLine made its Verity Prepare agent generally available in July 2026, and Trintech launched an exception management agent on 29 September 2026. Others offer copilots that suggest causes for a person to check. The Reconciliation page in ScalePoint Research covers the work step by step, and the vendor table for the cluster is coming soon.