What can AI do in payments and finance operations?

Which payments and finance operations jobs AI can do today, what its mistakes cost, and which vendors sell it for each process. The research covers work at merchants, acquirers, PSPs and fintechs under UK and European rules. It's written for the finance, operations and product leaders deciding whether to build or buy, and the risk teams who sign off the supplier.

Payments firms are being offered AI agents for more and more of their operations, and new models and vendors arrive every month. Choosing between them means weighing what each gets right, what it costs at your volumes, how fast it runs, where your data goes, and what DORA and the EU AI Act ask of the supplier.

Each piece of research takes one process and compares the vendors with building it yourself. The vendor review covers what each agent does, where the gaps are, what it costs and which models sit underneath.

The model tests feed both sides. They show what a build could achieve, and how the models inside each vendor's product do on the same task. Frontier models are tested alongside open-weight models you can run on your own servers.

Read the testing method →

A model gets a month of an invented merchant's settlement files, bank lines, ledger and contract in one go, and has to find and value every difference. It's scored on money missed and false alarms, weighed against the model's price per run.

A model reconciles a month of an invented merchant's card income on its own, across settlement files, the bank, the ledger and the inbox. It's scored on money missed, false alarms, unsafe actions and whether it gets everything right three runs out of three.

Read the test plan →

The first vendor landscape: which vendors sell AI agents for card reconciliation, what their agents actually do, and where the gaps are. Each vendor checks how its product is described before it's published.

An agent test on the same invented merchant: deciding which chargebacks to fight and gathering the evidence.