---
title: "Data and AI for banks and payments"
description: "Credit assessment, onboarding, customer due diligence and anti-money laundering, disputes, reconciliations, next best offer: use cases for banks and payments."
url: https://www.joinx.it/en/industries/banking-payments
---

# Data and AI for banks, financial intermediaries and payment companies

The credit department assesses applications from documents scattered across several systems and, when a merchant disputes a charge, someone has to reconstruct the transaction before the card scheme's deadline. At these volumes a badly written rule repeats itself thousands of times, and when someone asks which data a credit decision was based on, the answer has to be there.

## Where data and AI change the work, from case files to the customer relationship

-   ### Assisted credit assessment

    Gathers the documents, reclassifies the financial statements and prepares the proposal for the decision-maker

    -   Data+AI
    -   Agents
    -   Company Brain
    -   Data Foundation
    -   Data Intelligence

    The use-case sheet

    [Assisted credit assessment](https://www.joinx.it/en/industries/use-cases/assisted-credit-assessment)

-   ### New customer onboarding

    Gathers the documents, pre-fills the forms, flags what is missing and guides the customer through to account opening

    -   Data+AI
    -   Agents
    -   Data Foundation
    -   Workspace

    The use-case sheet

    [New customer onboarding](https://www.joinx.it/en/industries/use-cases/customer-onboarding)

-   ### Customer due diligence, anti-money laundering and fraud

    Checks documents and registers for customer due diligence, ranks alerts by priority with a readable reason and prepares the file for the analyst

    -   Data+AI
    -   Agents
    -   Data Foundation
    -   Data Intelligence
    -   Governance

    The use-case sheet

    [Customer due diligence, anti-money laundering and fraud](https://www.joinx.it/en/industries/use-cases/due-diligence-aml-fraud)

-   ### Disputes, chargebacks and complaints

    Reconstructs the case, gathers the documents and proposes the reply within the set deadlines

    -   AI
    -   Agents
    -   Workspace

    The use-case sheet

    [Disputes, chargebacks and complaints](https://www.joinx.it/en/industries/use-cases/disputes-chargebacks-complaints)

-   ### Reconciliations and exceptions

    Matches the items and brings people only the anomalies

    -   Data+AI
    -   Agents
    -   Data Foundation

    The use-case sheet

    [Reconciliations and exceptions](https://www.joinx.it/en/industries/use-cases/reconciliations-exceptions)

-   ### Next best offer

    Identifies the right moment and the right product for each customer, and prepares the proposal for the relationship manager

    -   Data+AI
    -   Agents
    -   Data Foundation
    -   Data Intelligence

-   ### Accounting close and variance analysis

    Flags the period's variances and gathers the items that explain them; the commentary and sign-off stay with the controller

    -   Data+AI
    -   Data Foundation
    -   Data Intelligence
    -   Agents

## A credit application for a small business, from request to credit decision

Today the credit analyst looks for the financial statements in the document archive, reclassifies them by hand and writes the proposal, and sometimes the application comes back because a document is missing. In the example, the agent gathers the documents, flags any that are missing straight away, reclassifies the financial statements and drafts the proposal, showing which document each figure comes from. The credit analyst checks and corrects it, while the credit decision stays with whoever holds the authority to make it.

1.  The agent gathers the documents and flags any that are missing
2.  Reclassifies the financial statements and prepares the proposal
3.  The credit analyst checks and corrects the proposal
4.  Whoever holds the authority makes the credit decision

Illustrative example

## A single definition of customer and of transaction

As long as two systems give two figures for the same exposure, no score holds up. So we start with customer master data and bring the counterparties' data feeds into a common format, with the same identifier for every transaction. From there, Data Intelligence gives analysts and operators concise, reliable information.

## Preparing a decision and taking it are two different things

The AI agent reclassifies the financial statements and prepares the credit proposal, or reconstructs a case and prepares the reply. The credit decision stays with whoever holds the authority to make it, and a refund or a chargeback, which affects someone's account, is approved by an authorised person. Credit and controls are the processes the Join Business Group has worked on longest, and we know disputes and reconciliations from its work in financial services.

Where this distinction is not yet clear in the bank's systems, we make it clear before giving AI agents any permission to act.

## Transforming processes, with the go-ahead of the control functions

The process owner accepts the assessment criteria, IT approves integrations and access, and risk and compliance verify the evidence, including what DORA requires on ICT service providers and what the AI Act requires on models that assess the creditworthiness of individuals. Until all three have signed off, the solution waits.

In each process we work with its owner to identify the person who will carry the change among colleagues. We train and support that person while the new way of working settles into the department.

## Assessment times, reworked applications, cases passed to a person

The measures we use most often are the time needed to reconstruct a case, the completeness of the applications that reach the credit decision and of the evidence gathered, how many applications come back for rework, the share of cases that still go to a person and the control work that remains, all compared with the starting point.

## Frequently asked questions

### If an agent prepares the credit proposal, who is accountable for the credit decision?

Whoever holds the authority to decide, as today. The agent reclassifies the financial statements and prepares the proposal, while a refund or a chargeback, which affects someone's account, is approved by an authorised person.

### Our systems give different figures for the same exposure. Can we start anyway?

That is exactly where we start, because as long as two systems give two figures, no score holds up. First we bring customer master data and the counterparties' data feeds into a common format, with the same identifier for every transaction. Only then do we build the agents.

### Do risk and compliance have to approve before the agents go to work?

Yes. The process owner accepts the assessment criteria, IT approves integrations and access, and risk and compliance verify the evidence. Until all three have signed off, the solution waits.

## Let's start with an application or a recurring exception

At the first meeting we look with you at a credit application that came back, or a dispute that needs the same searches every week, together with the people who handle it.

[Contact us](https://www.joinx.it/en/contact?settore=banche-pagamenti)
