Debt recovery solution matches human satisfaction scores

Global digital business services provider TP (formerly named Teleperformance) has reported that its AI-powered collections solution, TP.ai FAB Collect, matched human-level customer satisfaction scores while delivering a 40% debt recovery rate in live client deployments. 

According to the International Monetary Fund, non-performing loan risks in parts of Asia remain elevated, pressurising organisations to modernise collections without damaging customer relationships. 

TP in the Asia-Pacific region (APAC), believes that AI-driven, human-supported debt collection models are emerging as a potential way forward. Such models can help lenders tackle rising credit risk and changing borrower behaviour. TP.ai FAB Collect is powered by an AI model trained on decades of human collections expertise and is deployed where volume demands scale, supporting human advisors who focus on preserving customer relationships.

TP.ai FAB Collect empowers lenders to operationalise more predictive, customer-centric engagement at scale. Built on TP’s proprietary TP.ai FAB or Foundational AI Backbone framework, the solution integrates advanced analytics, AI-driven decisioning, and omnichannel engagement to improve recovery outcomes while maintaining compliance and customer trust. 

“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,” shared Assaf Tarnopolsky, TP’s Chief Business Development & Customer Officer, APAC. 

When deployed by a leading financial institution, TP.ai FAB Collect’s AI agents achieved a customer satisfaction (CSAT) score that was slightly higher than human agents while also achieving a 40% debt recovery rate. At a leading telecommunications company, AI agents adapted outreach to customers based on local payment behaviour, achieving a 7%-point improvement in the pay-to-contact ratio compared to the human-only model. 

The solution also improved recovery performance over time while reducing collections costs by 40% compared to a human-only model. 

Tarnopolsky said that beyond BFSI institutions and telcos, TP.ai FAB Collect applies wherever customers owe regular payments and relationships matter.

“Any industry where customers and providers exchange regular payments; whether it's a utility managing arrears, a healthcare provider following up on patient balances, or a retailer handling buy now pay later defaults, they face the same core tension: recovering revenue without damaging the customer relationship. That's exactly the problem Collect was built to solve," he said.

"Because TP.ai FAB is grounded in decades of human-agent expertise across these industries, the solution travels well beyond banking. TP.ai FAB Collect has proven efficacy wherever payments and customer trust intersect."

According to Tarnopolsky, utilities may have overdue bills and payment plans to manage, whereas healthcare patients may have outstanding self-pay balances and co-pays. In addition to buy-now-pay-later defaults, store credit might be another pain point for retailers and e-commerce brands. Other companies where the product is relevant include insurance firms due to lapsed premiums, and loan and lease arrears for automotive finance.

"Built on our proprietary AI orchestration platform TP.ai FAB, TP.ai FAB Collect carries the same compliance, empathy, and outcome focus into these sectors," Tarnopolsky said.