Can this AI agent redefine debt collection?

You wouldn’t expect empathy from machines, and particularly not from a bot that collects debt. But the makers of ProAgent say it’s doable.

The AI agent, made by US-based startup Prodigal, incorporates expressions of understanding in its interactions.

If a caller mentions that they are driving, the bot pauses. “Are you on hands-free? Please don’t hold your phone while driving,” it reminds.

If someone says their child has exams, it responds, “I hope the exams go well, let’s see how we can help you catch up on your payments.”

Prodigal co-founder Shantanu Gangal tells Tech in Asia that ProAgent has “infinite patience” and is available even at midnight when human agents are not.

“It can talk patiently even if it’s 10:30 pm. That kind of service and convenience is only possible with AI.”

But behind the seemingly friendly voice is a technology that’s reshaping a controversial corner of finance: debt collection. And this begs the question: Is an AI agent for debt collection even a good thing for borrowers?

Can AI empathize?

Debt collection is a strange test case for AI. The tech not only has to recognize speech and automate calls, but it also needs to understand the tone, legality, and empathy required to carry out its task efficiently.

To address this, Prodigal designed ProAgent as a three-layer stack.

The foundation layer runs on general language models such as Claude Sonnet, the GPT-4 series, and Llama 4. The middle layer adds financial and legal context, while the top layer is tuned for each client’s rules, disclosures, and tone.

“Think of it as a layered cake,” Gangal says. “The last layer is highly customized, just like how human agents are trained for each lender.”

From SaaS to AI debt collection

The human cost