AI steps into debt collection as credit stress climbs in South Africa

South African firms are piloting AI voice agents to handle debt-collection calls as mounting credit stress makes human conversations harder for many consumers. Technology company 1Stream has developed what it calls Agentic AI to take on those difficult initial conversations, the company’s CEO said.

Companies cite emotional barriers to human-led collections

Bruce von Maltitz, CEO of customer engagement technology company 1Stream, said debt collection carries an emotional burden that can block engagement. He told reporters:

“Debt is a difficult topic to address under the best of circumstances. When debt becomes overdue, or the topic then moves to debt collection, things become even more difficult – practically and emotionally.”

Von Maltitz said the company has developed Agentic AI to handle calls, and that operational data from South African contact centres using voice automation or Agentic AI suggested some consumers are more comfortable engaging with an AI system during initial conversations.

Credit stress statistics underpin the push to automation

Recent data cited in reporting underscores rising payment strain among consumers. Researcher Eighty20’s Credit Stress Report found that around 40% of the country’s credit-active population is more than three months in arrears on at least one loan. Credit bureau TransUnion reported that about 35% of consumers expect to miss at least one bill payment during the current cycle.

Sector peer Experian reported a 15.77% default rate on retail credit products, and that there are more than 22 million retail credit accounts currently active in South Africa.

Why AI may lower psychological barriers

Von Maltitz cited international research to explain why machines can change the tone of collections calls. He referenced an early 2026 study titled AI in Debt Collection called Estimating the Psychological Impact on Consumers, saying the work found that

consumers experience heightened feelings of stigma and social evaluation when talking to a human, who they perceive is capable of morally judging them

and that

interacting with an AI system that lacks ego or judgment significantly lowers these psychological barriers

.

Not a wholesale replacement for humans, experts warn

Carmen Murray, described in the reporting as a cultural forecaster, cautioned that businesses should not treat AI as a substitute for people. She said the technology works best when it expands human capability rather than removes people from the equation.

Von Maltitz reinforced that view for collections work:

“AI should never be used to manage complex negotiations, handle formal account disputes, or engage with vulnerable consumers experiencing severe financial or personal hardship,”

he said.

“Human professionals remain irreplaceable in collections. The technology is simply there to handle the routine and simple tasks, freeing up human agents to focus on the cases that truly require a human touch.”

What this means for consumers and lenders

Within the limits of the reporting, companies are testing AI to reduce stigma and improve first-contact rates while keeping humans for complex or sensitive cases. As credit stress indicators rise, lenders and contact centres face pressure to balance automation with oversight and humane handling of vulnerable customers.