For years, collections remained one of the least digitised functions in financial services. While lenders invested heavily in underwriting, fraud detection and payments innovation, debt recovery largely relied on manual processes, fragmented agency networks and standardised recovery playbooks.
That is beginning to change. According to Ranjith B.R., Founder and CTO of DPDzero, artificial intelligence is helping transform collections from a reactive recovery function into a predictive, intelligence-led operating model that improves both recovery outcomes and borrower experience.
In an exclusive interaction with Express Computer, Ranjith explains how AI is reshaping collections, why the function is uniquely complex from a technology perspective, and how lenders can balance automation with compliance and empathy.
Moving beyond one-size-fits-all collections
Traditional collections approaches have largely treated borrowers the same way, regardless of their financial situation, intent or behaviour. Ranjith believes that this is where AI is creating the biggest shift.
“The fact that each borrower is unique and so are their reasons to not pay is never taken into account in the traditional collection methods. Each borrower is treated likewise. AI changes that.”
By analysing borrower interactions across voice, WhatsApp, SMS, payment activity and call outcomes, AI helps build a unified understanding of customer behaviour. This allows lenders to personalise recovery strategies rather than relying on blanket outreach campaigns.
One of the strongest insights emerging from large-scale borrower data is that context often matters more than outreach volume.
“Borrower behaviour is highly contextual – timing, tone, and channel selection often outperform aggressive outreach volume.”
Instead of sending the same reminder to every borrower, AI can identify who is likely to respond to a digital nudge, who may require a restructuring conversation, and where human intervention is necessary. As unsecured lending continues to grow rapidly in India, such intelligence-led collections models become increasingly important.
Why collections is a difficult AI problem
Unlike underwriting, which is largely a one-time decision, collections is a continuously evolving process where borrower behaviour changes over time.
Ranjith argues that collections sits at the intersection of behavioural psychology, financial risk, operational execution and regulatory compliance.
“Collections is uniquely complex because it sits at the intersection of behavioural psychology, real-time operations, financial risk, and regulatory sensitivity,” he points out.
Defaults are often driven not only by financial stress but also by behavioural and situational factors such as temporary liquidity gaps, communication fatigue or lack of repayment clarity. This means AI systems must continuously adapt rather than operate on static rules.
The challenge becomes even greater in India, where lenders serve borrowers across different geographies, languages and repayment patterns. Modern collections infrastructure must coordinate AI voice systems, telecalling operations, messaging platforms, payment systems and field teams while maintaining a unified borrower view.
Compliance must be built into the platform
As AI adoption accelerates, compliance is becoming a core technology requirement.
Ranjith emphasises that collections cannot operate as a black-box AI system. Explainability, auditability and borrower dignity must be embedded into the platform architecture from the start.
Technology can enforce regulatory requirements such as outreach limits, permitted contact hours and communication monitoring. AI can also analyse call recordings and flag potential violations before they become larger issues.
Importantly, he believes AI should enhance accountability rather than replace it. “AI should be seen as an enabler in building more context across conversations, sharpening decision-making and at the same time not removing accountability.”
Human oversight therefore remains critical, particularly in sensitive cases where judgement and empathy are required.
AI as a copilot, not a replacement
Despite growing automation, Ranjith does not see AI replacing collection professionals.
Instead, he views AI as an augmentation layer that automates repetitive tasks such as reminders, payment nudges and routine follow-ups while enabling human agents to focus on more complex conversations.
“We see AI as an augmentation layer for collections rather than a replacement for human judgment.”
Financial distress often involves emotional and personal circumstances that technology alone cannot fully address.
He reminds, “Collections still remain a human-centric function. Financial distress often involves emotional sensitivity, negotiation, and contextual understanding that still require empathy and judgement.”
In this model, AI serves as a copilot, helping agents prioritise accounts, recommend next-best actions, surface borrower insights and monitor compliance in real time.
The future of collections
As lending volumes grow and customer expectations evolve, collections is becoming more than a recovery function. It is emerging as a strategic technology layer that influences risk management, operational efficiency and customer trust.
For Ranjith, the future lies in combining intelligence, automation and human expertise within a single operating framework. “Technology should own outcomes. That means building platforms where intelligence, automation, and human workflows work together seamlessly.”
The result is a shift from reactive debt recovery towards predictive, data-driven collections infrastructure, one that enables lenders to scale efficiently while maintaining compliance and borrower dignity.