TP AI debt recovery matches human satisfaction in Asia

TP said its TP.ai FAB Collect debt recovery system matched human customer satisfaction scores in live deployments while delivering a 40% debt recovery rate.

The figures come as lenders across Asia face pressure to manage elevated non-performing loan risks without damaging customer relationships. Debt collection is becoming more important to both financial performance and customer retention as borrower behaviour shifts.

Debt recovery has long been one of the most sensitive parts of financial services, as banks and lenders try to collect overdue payments without unnecessarily harming customer trust. That tension has increased interest in systems that automate early contact with borrowers while leaving more delicate cases to trained staff.

TP said its model uses artificial intelligence for earlier-stage outreach and routes more complex or sensitive conversations to human advisers. The system was developed using data drawn from decades of collections work, according to the company.

Live deployments

In one deployment at a financial institution, AI agents produced a customer satisfaction score slightly higher than that of human agents while achieving a 40% debt recovery rate, TP said. The company added that the system cut collections costs by 40% compared with a human-only model and improved recovery performance over time.

At a telecommunications company, the system adapted customer outreach based on local payment behaviour. That deployment produced a 7 percentage-point improvement in the pay-to-contact ratio compared with a human-only approach, according to TP.

The results reflect growing use of AI in operational areas where institutions want greater efficiency but cannot afford missteps in customer treatment. Collections teams face particular scrutiny because aggressive or poorly timed contact can damage reputations and increase churn, especially in consumer lending and payments.

TP, formerly known as Teleperformance, has been expanding its profile in Singapore and the wider Asia-Pacific market. It describes the debt recovery tool as part of a broader push to combine automation with human oversight in banking, payments, lending and back-office operations.

Human oversight

Assaf Tarnopolsky, Chief Business Development & Customer Officer, APAC, said the system is designed to change how work is divided between software and staff rather than remove people from the process altogether.

"We trained our AI on 40 years of human collections expertise. Now it handles the first wave, so our human advisors can focus on the conversations that truly matter. TP.ai FAB Collect doesn't replace the human touch; it multiplies it," Tarnopolsky said.

The product sits on TP's TP.ai FAB framework, which combines analytics, decisioning tools and multi-channel customer contact. This allows lenders to move from traditional recovery practices towards what TP described as more predictive engagement.

Financial institutions in Asia have been examining how to modernise collections as loan quality weakens in some markets. According to the International Monetary Fund, non-performing loan risks remain elevated in parts of the region, leaving banks under pressure to improve recovery without undermining compliance or customer treatment standards.

That backdrop has created a market for technology vendors promising lower handling costs and better borrower engagement. But banks and lenders still face questions around governance, transparency and where human review must remain in place, especially when customers are vulnerable or disputes arise.

TP said its approach is based on AI-supported, human-led collections, with staff taking over where judgment and sensitivity are most needed. The company added that TP.ai FAB Collect received a 2026 Artificial Intelligence Excellence Award from the Business Intelligence Group in the Automation category.