How to use AI for credit management

This wealth of information is a godsend for credit managers – if they have the tools to take full advantage of it. 

That is why companies like Atradius are implementing sophisticated systems based on Artificial Intelligence (AI) and Machine Learning technologies to help clients take accurate decisions based on scientifically collected, crunched and analysed information. 

The use of AI and Machine Learning in credit management was the topic of the tenth episode of Atradius’ podcast series, which starred Stan Chang, the director of Group Buyer Underwriting in Amsterdam. 

Chang explained how Atradius is maximizing data analytics via AI and Machine Learning and listed the benefits that the deployment of state-of-the-art technologies can bring to clients and business partners. 

“Atradius uses machine learning to do some simple but very powerful tasks, such as retrieving and processing information, in order to optimize risk decisions,” Chang told podcast hosts Mary Ibrahim and David Finn. “We use web scrapers, APIs and various associated technologies to pull in information in real time. It is information that is published in hundreds of thousands of websites, in multiple languages, on a 24/7 basis.” 

Web scrapers are software that roam the internet to collect specified bits of information that are particularly relevant for a company. APIs, or application programming interfaces, are tools that enable different kinds of software to talk to each other. Although not exactly new, technologies like those are transforming the way that business functions work, and credit management is no exception to the rule. 

With those technologies, Atradius can raise information that is extremely valuable when assessing credit risk. It includes data about M&A transactions, changes in management, payment defaults, the launching of new products, job advertisements, sanctions, litigations, labour strikes and many other developments that provide hints to the abilities of companies to fulfil their credit commitments.  

“It is qualitative or unstructured information that, historically, has been very hard to use in models and algorithms,” Chang said.  When we need to make sense of all that data and turn it into accurate decisions, AI and Machine Learning come into play.  

“Big Data is the oil that runs all AI and Machine Learning engines. It helps us understand performance and relationships, and to make the best risk decisions that our customers demand today,” Chang pointed out. “In the B2B world, AI is a bit less visible than in consumer markets. But the change is definitely happening, and it is transformational.” 

Once the data is extracted, Atradius crunches it through natural language processing models, which the company has developed in house from AI technology, and with machine learning systems. The information is classified, and sentiment analysis, which takes into account subjective information, is performed. The data is also matched and compared to massive amounts of related information stored in databases.  

Alongside the ability to process large amounts of news and qualitative information, Atradius also uses AI or Neural Networks to detect relationships in drivers that cause non-payment. “This way, we are issuing more decisions to support trade and getting them out faster,” Chang said.  

With so much data being generated in the world today, it is not hard to understand why it is vital for companies to try to make the best of it. But Chang warned that investments in technologies such as AI and Machine Learning must not be made lightly. They require significant commitments in money and human resources and there is always a risk of squandering it all if the company does not have a very clear idea of why it is taking such a step. 

“AI and Machine Learning are about bringing efficiency, productivity and better results. They can bring new services and new propositions, which means future-proofing a business,” he said. “But one should not focus on AI and Machine Learning because they are sexy topics, and everybody is talking about them. They are not goals in themselves. They are tools and technologies designed to help us.” 

Once the investment decision is made, Chang’s recommendation is to scale it step-by-step, and not try to decipher the whole world of Big Data in the first try. 

“Keep it simple,” he said. “Adoption is far more important than impact when you start.” 

Listen to the full episode How to use AI for credit management