Hi everyone,

Welcome to the 172nd edition of The Pattern, a weekly newsletter where we dive into the latest from the world of economy, technology and finance. Let’s get right into it! 

Gone are the days when a collection agency paid a visit to a loan defaulter's home to collect the rest of the money.  

Today, the knock on the door has been replaced by a ring on the phone. And instead of a burly man with a clipboard, one might find staring at a woman in a lawyer's robe or a man in a sharply cut suit, brisk tone and legalese intact.  

The catch? Neither of them exists.

This isn't a sci-fi thriller; it's the new reality of loan recovery in India. They are the new-age debt collectors: AI-generated avatars. 

Debt collection turned ugly 

For decades, loan recovery in India has held the reputation of being humiliating and harassing. From excessive calls to relentless texts to public shaming, and behaviour that crosses the thin line between firm and abusive has only added salt to the wound of a distressed borrower. 

On the other hand, economics never really worked either. A human collection agent costs up to ₹30,000 a month while managing around 250 borrowers, and the data is often unreliable, outdated, and scattered across fragmented public records.  The available information is either poorly structured or excessively dense. 

In other words: too much wasted effort, too little recovered money. 

Defaults are climbing 

Loan defaults in India have risen to a six-quarter high, with Tier 3 towns and beyond facing the sharpest surge. According to a report, loans overdue by more than 90 days stood at 3.6% in March 2025, which is up from 3.3% a year earlier.  

With repayment stress mounting, banks had to rethink their playbook.   

Cue the AI collector 

These advanced platforms, powered by AI and real-time data, bring order to this chaos. These systems update profiles on the fly, do the heavy lifting of verification and organisation, and boost recovery success.  

Companies like Credgenics, Oriserve, and Sarvam AI, already working with giants in the BFSI space. Even state-run banks are exploring this option. Beyond making calls, these AI bots are learning, adapting, and refining their tactics—staying polite, thoughtful and respectful where humans sometimes fall short. 

According to a report by the Economic Times a bank executive said, "At present, most banks and financial institutions use a mix of these products, which includes AI-generated video calling, text messages and then, in more troublesome cases, follow-ups by real-life agents. Most banks have outsourced their collection business, and the ones which do it internally hire such firms for technology support." 

The pitch is simple: an AI agent can make 20 times more calls, costs 40–60% less and delivers resolution rates of 80–85% in the zero-to-30-day delinquency window, especially for personal and commercial vehicle loans. 

Between the fine print 

As AI steadily takes over the loan recovery ecosystem, the Reserve Bank of India is keeping a close watch—ensuring that technology plays by the same rules designed for human recovery agents. 

• No calls before 8 am or after 7 pm

• No intimidation

• No harassment

• No invasion of privacy

As one executive at a state-run bank put it: "While service providers are claiming that guardrails are in place, it needs to be seen whether the latest versions do not breach any regulatory guidelines." 

Man vs machine  

Rules can keep AI in check, sure. However, beyond regulations, there's a deeper tension: the fundamental shortcomings that arise from replacing humans with machines. 

 AI in loan recovery comes with its own set of dilemmas: avatars can’t truly show empathy and risk sounding tone-deaf; data privacy and bias remain real concerns; adoption is uneven, with private banks leading while PSBs hesitate and smaller lenders balk at costs; and looming over it all is the threat of scams getting smarter with deepfakes and fake IDs. 

In other words: efficiency comes with strings attached. 

AI or nay? 

Strip away the tech, and the heart of debt collection remains human: borrowers trying to survive, lenders trying to balance their books. AI changes the speed and scale of the chase, but it doesn't change the fact that behind every missed payment is a person, not a datapoint. 

Yes, banks are saving money. Yes, bots can recover more loans in less time. But what happens when a borrower wants to explain their situation or when a delicate case requires listening instead of scripted answers? That's where the human touch still matters. 

The future of loan recovery won’t be a battle between humans and AI.  It will be a partnership. With machines driving scale and humans adding empathy, collections can finally move from being combative to constructive—faster, fairer, and designed to protect dignity. 

That's a wrap for this week. As always, leaving you with a few reads to explore. Have a great weekend! 

 
Reading List: 

1. GST 2.0 ‘tap’ dance: ePayments leap 10x to Rs 11.3 lakh cr on Day 1

2. Changing face of banking mobilise India's financial markets: Report 

3. Real time payments growth to moderate at 4% after growing over 8% since 2019: BCG 

4. Payment, lending apps more likely to get installed than banking apps: RBI Bulletin 

5. Banks plan common portal to track consortium loans 

6. Changing face of banking mobilise India's financial markets: Report 

7. Vehicle loans and loans against gold appear robust for NBFC: RBI bulletin   

Thank you for reading. If you liked this edition, forward it to your friends, peers, and colleagues. You can also connect with me on X here and follow FinBox on LinkedIn to get the latest updates.   

Cheers,  
Mayank