ROLF never sleeps. He is on duty day and night, keeps an eye on all the accounts, checks off incoming payments and takes note of open claims. When his colleagues come into the office in the morning, he supports them in their communication with defaulting payers. ROLF is an Artificial Intelligence (AI) system that works at the Norwegian branch of Svea Finans, a partner company of EOS.
“Thanks to ROLF, communication now runs far more smoothly,” says Stian Ellingsen, Team Leader Billing & International Debt Collection. The AI software prepares legal assessments for cases, updates address data and analyzes the information from authorities and credit agencies. Based on his analyses, ROLF recommends to his human colleagues which next step has the greatest chance of success for each individual case. Does it require a letter, an SMS or a phone call? Does the defaulting payer need a little more time? Should Svea Finans take legal action against the defaulting payer who has a large number of long-term outstanding claims? ROLF also learns. The AI system looks back to see which of its suggestions for making contact were successful and which were not. Has a person of this age and gender in the same situation responded to a letter and settled the claim? Or only a part? What is the best way to persuade him or her to pay the remaining outstanding amount? And what could work better in a similar situation next time? “We gave the system the ability to change itself continuously so it can increase its success rate in customer communication,” says Magnus Bødtger, Developer and AI Expert at Svea Finans.
Experts at the company met back in 2013 to identify which tasks and processes it would make most sense to automate. As a first step, they developed their own automation software Sidra. It now forms the basis for ROLF. “We wanted to be independent and work with a tailor-made software in which we could directly make changes ourselves,” says Bødtger. The name ROLF does not have any deeper meaning – the developers simply thought it was fitting.
The introduction of the AI system was both market and customer-driven. Svea Finans, which was founded in Sweden in 1981, now operates branches in eight European countries and is part of the EOS Global Collection network. To be able to keep the costs down in a highly competitive market, Svea Finans must automate tasks and processes. “Our customers too increasingly ask for AI-based processes,” says team leader Ellingsen. The reason: greater accuracy.
Previously, employees had to gather up all the data from the records for every single transaction and keep the entire customer communication in mind. Thanks to ROLF, the error rate has declined noticeably, while the success rate for settling outstanding claims has risen: “We are seeing a signifi cant improvement here,” says Ellingsen. At the same time, data protection is still essential. So one employee exclusively monitors compliance with the GDPR and other guidelines.
Not only are the customers happy, but so too are the defaulting payers. The reason: by means of the AI system, the people at Svea Finans have quick access to all the relevant data in every case. They can therefore answer inquiries more quickly and provide better information. Defaulting payers receive appropriate answers without delay, can view their financial status at a glance and easily request documents such as copies of invoices. “This, in turn, gives our administrative staff the time to take care of complex cases and maintain customer relations,” says Ellingsen.
Humans can take over at any time.
An important fact to note is that employees observe every step of the AI system and can intervene in particularly tricky cases to take over the entire process. Human intervention will be possible even when ROLF independently carries out not only the assessments but also actions, such as handing over entire processes to judicial officers.
The next step in automating the customer dialog will be a chatbot, which will answer questions from defaulting payers or send information. According to developer Bødtger, Svea Finans in Norway is currently investigating whether the company will purchase this kind of chatbot as a turnkey solution or – like Sidra and ROLF – if they will program it themselves. In any case, although the bot will be connected to all systems – a human will be able to switch it off at any time. “Machines are not taking over,” stresses Bødtger. “We will always need people to monitor these dialog processes.”