Some South Africans prefer to talk to AI about their debt.
Image: Supplied
Millions of South Africans are struggling with debt, and the way they engage with creditors may be quietly shifting. Some consumers appear more comfortable discussing overdue accounts with artificial intelligence voice agents than with human debt collectors, according to industry practitioners.
The trend, says Bruce von Maltitz, CEO of 1Stream, reflects the emotional weight that often comes with debt collection rather than a purely technological shift.
“Debt is a difficult topic to address under the best of circumstances,” von Maltitz said. “When debt becomes overdue or the topic moves to debt collection, things become even more difficult – practically and emotionally.”
He said some consumers appear to respond more readily to AI voice agents because the interaction feels less personal and less judgmental than speaking to a human agent.
“There are people who have serious financial problems leading to defaulting, but there are also many cases where a debit order simply didn’t go off because of a mistake, or a payment was missed and the customer fully intends to pay,” he said. “In those instances, you often need a fast, consistent way to notify the person and help them resolve the issue.”
The scale of debt pressure remains significant. The latest available National Credit Regulator Credit Bureau Monitor shows that 10.54 million of South Africa’s 29.24 million credit-active consumers had impaired credit records as at the end of June 2025. That represents just over 36% of credit-active consumers.
The NCR defines impaired records as consumers who are three months or more behind on repayments, have adverse listings, or are subject to judgments or administration orders.
Separate industry research points to continued financial strain across households. According to Eighty20’s Credit Stress Report, about 40% of credit-active consumers are more than three months in arrears on at least one credit agreement, while TransUnion industry insights indicate that roughly 35% of consumers expect to miss at least one bill payment in a given cycle.
Retail credit remains a key pressure point. Experian data has previously shown relatively high default levels in the retail credit segment, which includes millions of small-value accounts across the country.
Von Maltitz said AI voice agents are increasingly being explored as a way to handle early-stage contact with customers at scale.
Historically, he said, it was often not economically viable for organisations to use human call centre agents to repeatedly contact low-value retail accounts. AI systems, he argued, can reduce that cost while maintaining consistent communication.
“AI voice bots have no egos, no moods, and no capacity for personal judgment, so consumers feel far less embarrassed,” he said. “It normalises the interaction. The bot is simply there to confirm details and arrange a payment path.”
He added that AI tools are not intended to replace human agents, but to manage the initial stage of contact.
“The first conversation is where we see the demand,” he said. “If you can get a promise to pay using a bot, that is quick and inexpensive. If the situation is more complex, it moves to a human agent.”
Von Maltitz said early and scalable contact with customers could also improve outcomes for businesses, particularly when reaching consumers before financial pressure escalates further in the month.
However, he cautioned that any AI-driven collections system must be carefully designed to avoid overwhelming consumers with automated contact, given already high levels of unsolicited communication.
Used carefully, he said, AI voice agents could reduce friction in the early stages of debt collection while allowing human agents to focus on more complex, distressed or disputed cases.