---
title: "Credit management for energy and utilities with AI"
description: "A model ranks overdue bills by risk, the agent proposes the repayment plan and the credit manager approves or changes it."
url: https://www.joinx.it/en/industries/use-cases/credit-management
---

# Overdue bills in order of risk and repayment plans ready to approve

The credit department works through overdue bills by date or by amount, with the same reminders for someone who has forgotten a payment and for someone who has not paid for months.

## The same reminders for different situations

Every billing cycle leaves a list of unpaid bills, which the credit staff work through in the order the system presents it. They send the reminders, phone the customers with the largest amounts and agree repayment plans case by case, looking at invoices, payments and contacts in the CRM side by side. So a customer with an open complaint on that very bill may receive a reminder, while the accounts most at risk grow old in the queue behind others that would have settled themselves.

## The model ranks the accounts, the credit manager approves the plan

In the example, the Data Intelligence models read the payment history, the overdue amounts and the reminders left unanswered, and put the accounts most at risk at the top with the reasons alongside. The agent checks whether there is an open complaint on the bill, in which case it holds back the reminder and flags it, then proposes a repayment plan within the company's rules. The credit manager approves or changes it, and the errors they report are used to correct the model.

1.  The model ranks the accounts by risk, with the reasons
2.  The agent checks for open complaints and proposes the repayment plan
3.  The credit manager approves or changes the plan
4.  Reported errors correct the model

Illustrative example

## Data, modules and decisions for a utility's credit

| Item | In this case |
| --- | --- |
| Data | Invoices and receipts matched to bills, payment history, reminders sent, open complaints in the CRM, company rules on repayment plans |
| Modules | Data Foundation, Data Intelligence, Agents |
| Who decides | The credit manager on the repayment plan and on any action that affects supply; the head of credit on the agent's rules and the model's thresholds |
| What we measure | Collections on the accounts worked, the ageing of overdue amounts, repayment plans kept to and reminders sent on bills with an open complaint, against the starting point |
| AI Act risk class | Ranking bills that are already overdue by risk is not the assessment of creditworthiness for granting a loan, the use that Annex III classifies as high-risk for individuals. Because the model also ranks household accounts, we set the class with risk and compliance before development. |
| Where to start | With Deployment, if receipts and bills are already matched and the head of credit has written down the rules on repayment plans; otherwise we start from the data with Data Foundation |

## If receipts cannot be matched to bills

When a payment arrives without a clear reference and stays unmatched for weeks, the model sees bills already paid as overdue, and the wrong reminder reaches a customer who is up to date. Receipt matching has to be fixed first. Repayment plans granted verbally, with no written rules on duration and amounts, likewise leave the agent without criteria, and writing them down with whoever is accountable for credit comes before building.

A utility's other processes are on the [Energy & utilities](https://www.joinx.it/en/industries/energy-utilities) page, and all the sheets are in the [use cases](https://www.joinx.it/en/industries/use-cases).

## Let's look together at a list of overdue bills

With the credit department we take the latest list of overdue accounts and see which information was missing to put them in order of risk.

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