Why resilience matters as much as innovation in NBFCs
The NBFC space is set to enter a new phase where AI, cloud computing, analytics, and digitisation will no longer be different ways of innovating but an essential part of a new business model. Furthermore, financial institutions will be working in a more complex business environment driven by dynamic consumer needs, regulations, and resilience.
As HDB Financial Services, the topmost priority right now goes beyond embracing advanced technological innovations to creating an enabling digital ecosystem.
In an interview with Express Computer, Mathew Panat, CTO of HDB Financial Services, shed light on the efforts made to build capabilities and use AI in lending, collections, customer services, underwriting, and operations in a way that ensures both consumer experience and security.
Building the digital infrastructure before rolling out AI
Panat believes the current wave of AI-led transformation is possible largely because financial institutions spent the last few years strengthening their technology foundations.
He explains that technologies such as cloud computing, mobility, cybersecurity, data lakes, and modern digital platforms became critical building blocks as the industry prepared itself for long-term scale.
“There were new technologies that were adopted, and these were primarily in the nature of social media. There was adoption around cloud computing, data, data lakes, and those kinds of concepts. There was adoption around mobility and how new mobile solutions and platforms come into play. There was a lot of adoption around info security, a lot of new trends, zero-day threats and that kind of stuff.”
Alongside these technology shifts, HDB Financial Services also focused on ensuring that the organisation had the right foundational architecture to support scale, resilience, and operational efficiency across a highly diversified lending portfolio. “We needed to ensure that we could scale at a low cost or a low price point because those total unit economics also matter for us. We also cater to requirements or loan products which are very small tickets-think of microfinance in that space, to very large tickets, which is like construction equipment or a commercial vehicle and everything in between.”
According to him, technology decisions in the NBFC sector cannot follow a “one-size-fits-all” approach because each product category comes with different economics, customer behaviour, and profitability models. “We should be able to cater to that, but we should also be able to bring in the economics of scale aspect into the whole thing,” he says.
At the same time, while fintech firms brought speed and agility into the ecosystem, Panat believes regulated financial institutions must maintain a stronger balance between innovation and governance.
“Fintechs have a lot that we could learn from. But at the same time, there is a lot that we need to unlearn from them, because they are very good from a technology perspective, but when it comes to ensuring that it consistently works, which is the resilience, governance, and security part, they have taken their prime focus has been faster time to market,” he adds.
For HDB Financial Services, the objective has been to combine speed with long-term sustainability.
AI moves from pilots to enterprise-wide deployment
Over the last year, the company has accelerated investments around AI-led innovation, with multiple workstreams now running simultaneously across business functions.
Panat says the organisation has identified nearly eight to nine major AI themes spanning sales, collections, operations, HR, customer service, risk management, IT, and information security. “AI for sales, AI for collection, AI for credit facilitation, AI for operations, AI for IT and infosec, AI for risk management and regulatory aspects of it, and AI for HR. We’ve got some eight to nine themes that we are running in parallel.”
However, he clarifies that the organisation’s approach remains practical and outcome-driven rather than driven purely by hype.
Some of these AI initiatives are already operational. In collections, for instance, AI-based voice bots are now being used for early delinquency reminders and payment nudges. In customer service, AI is helping categorise incoming requests and automate responses for simpler queries. One of the more interesting implementations has been around multilingual voice-based customer onboarding journeys.
“We have created a voice-based AI acquisition engine. Right now we have got eight Indian languages that we support. So let’s say you want to speak in Hindi; the bot will take you through the entire loan journey where a decision can be taken, ” he explains.
Internally, the organisation is also experimenting with AI-driven employee assistance tools for HR-related queries and workflows.
AI-assisted underwriting, not AI-only underwriting
While AI is increasingly becoming part of underwriting workflows, Panat says the organisation is deliberately taking a cautious approach when it comes to fully autonomous credit decisioning. “We are using it to facilitate underwriting. We are not using it to underwrite.”
According to him, explainability and accountability remain critical concerns in financial services, particularly in regulated lending environments. “We are very cognisant of the fact that this is still a technology that has to go through some level of maturity curve. If tomorrow I have to explain as to why a model decided to do what it decided to do, I should be comfortable as to being able to do that.”
Instead of replacing human judgement, the current objective is to reduce manual effort and accelerate decision-making.
Using data intelligence for proactive risk management
Risk management remains central to the NBFC business model, and HDB Financial Services has built a large part of its technology strategy around centralised data intelligence.
As part of its broader transformation journey, the company created a cloud-based data lake on AWS that consolidates data from loan origination systems, loan management platforms, and peripheral systems.
“We said we’ll bring in data from predominantly all the loan origination systems and loan management systems and some more peripheral systems into this data lake.”
The organisation now uses this infrastructure to build early-warning systems and predictive risk insights.
Panat explains how borrower behaviour patterns often serve as strong indicators of future stress. “For example, if I look from a collections perspective, the same data can give me a flavour as to whether there’s a potential that this customer could default on their loan payments.”
The same datasets are also helping the company identify broader trends across geographies and product lines. According to him, the value lies in deriving multiple business insights from the same unified data foundation.
Keeping customer experience simple and frictionless
Panat believes technology should simplify experiences rather than complicate them.
“The way we look at this problem is keeping the customer at the centre of all the things.”
He explains that customers across age groups, geographies, and income segments engage differently with digital financial services. Some customers prefer entirely self-service models, while others still rely heavily on assisted journeys and relationship managers.
“There are customers who are very happy with an unassisted model where there’s absolutely no human being supporting them. There are certain customers who are more comfortable with a human touch point.”
The organisation therefore focuses on delivering consistency across both assisted and digital channels.
Why ideathons matter in financial services innovation
Interestingly, HDB Financial Services recently conducted its first ideathon with participation from nearly 38 companies and startups.
For Panat, these initiatives are important because they challenge institutional inertia and introduce unconventional thinking into large enterprises. “Disruption is the best way to bring in innovation. If things are maintained in a status quo mode over a period of time, that sets in some level of inertia. These startups come up with brilliant ideas. These ideas force you to think, can I use these ideas to grow my business?”
He also believes startups often operate without the constraints that traditional institutions place upon themselves.
Interestingly, Panat also believes AI-assisted development tools could fundamentally reshape how enterprises innovate in the coming years.
The road ahead: AI at scale, edge computing and privacy-first finance
Looking ahead, Panat expects AI adoption to deepen significantly across the organisation over the next few years.
“AI has brought in a lot of things, so obviously I will adopt AI in a larger and bigger way,” he points out.
At the same time, the organisation continues to prioritise resilience, reliability, scalability, and operational predictability. “We will continue to focus on resilience, reliability and scalability because as you keep on building out your foundation, you still need to ensure that predictability and guaranteed availability are baked into things.”
They are also exploring edge computing as a strategic area for the future. According to him, edge-led architectures could help improve agility, reduce latency, and accelerate experimentation without requiring heavy capital investments.
At the same time, he believes the future of embedded finance will increasingly be shaped by privacy regulations and consent frameworks under DPDP compliance.
While fintech partnerships will continue to grow, Panat says regulated entities like HDB Financial Services will continue to play a stabilising role in balancing innovation with governance. “These fintechs have solved a problem which traditional banks and NBFCs have struggled to solve, which is around customer experience. But what they have not solved is ensuring that the risk-reward and compliance aspects of it are aligned with what the regulator wants,” he concludes.
