As we transition to more digital lending processes, the real challenge for NBFCs is not just to roll out apps but also to redesign the whole loan journey to make it faster, fairer and resistant to fraud while ensuring inclusion for those who prefer branch visits or need support. Kunal Dikshit, CTO at Fedbank Financial Services, is doing just that.
In conversation with CIONOW, Dikshit talks about how Fedbank measures the impact of digital on its operations, how it governs AI before it can influence credit decisions, how it balances frictionless onboarding with tighter identity validation, and how it builds cyber resilience across its expanding ecosystem of cloud, fintech and vendor partners.
What has been the biggest impact of the digital, paperless lending experience for Fedbank Financial Services? Which parts of the loan life cycle have delivered the biggest improvement in turnaround time and in the customer experience?
At Fedbank Financial Services, our focus has been on progressively removing manual intervention and unnecessary physical touchpoints from the customer journey.
On the customer-facing side, we have enhanced our mobile application and website to enable services such as online gold-loan renewals, digital interest payments, AutoPay setup and document downloads. These initiatives have helped customers complete routine activities without having to depend on branch visits.
We are also working on digitising key parts of the lending process, particularly KYC and document-related activities. Our AI initiatives around document collection, document extraction, field extraction and automated form filling are aimed at reducing repetitive manual work and improving the speed and quality of processing.
We measure the impact through a combination of turnaround time, digital adoption, reduction in manual touchpoints, transaction migration to digital channels, operational productivity and customer-service metrics.
The larger objective is to move from simply making individual processes digital to creating an increasingly paperless, connected and frictionless lending journey.
As lending becomes increasingly digital, how are you using data, analytics and AI to improve credit assessment without making the process opaque or unfair for customers? What governance controls are essential before putting AI into production?
We see AI as an enabler for better and more consistent decision-making, but not as a black box that replaces responsible lending judgement.
Our current focus has been on practical AI applications across areas such as KYC document collection, document understanding, field extraction, fraud detection and automated form filling, particularly for mortgage-related processes.
The immediate value is in making information available faster and more accurately to the lending process, reducing manual data entry and enabling employees to spend more time on judgement-based activities.
We believe for any AI solution before it enters production, governance needs to include data quality, validation of models, interpretability, bias and fairness, information security, access management, auditability, human oversight and ongoing monitoring of performance. For a financial institution, the question should not only be ‘Can AI make the decision?’ but ‘Can we explain, govern, monitor and challenge the decision?’. And that becomes all the more critical when AI decisions affect customer eligibility, credit or risk.
Many financial institutions struggle to integrate legacy platforms with modern digital systems. What technology and architecture decisions helped create a scalable, modular and API-led lending ecosystem?
In a lending organisation, completely replacing established legacy platforms such as ‘FinOne’ is neither practical nor necessary. Our approach is to create a modern technology layer around the legacy ecosystem so that we can continue to leverage the stability of existing systems while building new digital capabilities independently.
A key part of this architecture is an API Gateway, where APIs from different systems and platforms are integrated and managed through a common layer. This provides a standardised way for digital channels, applications and external partners to communicate with the underlying systems.
We are also building a Data Lake as a central data layer, bringing together data from different systems. This enables middleware and business applications to be developed on top of the data layer rather than creating direct dependencies on individual legacy systems.
The objective is to have middleware applications sitting between the legacy platforms and modern digital applications. This decouples the new applications from systems such as FinOne, allowing us to introduce or modify digital capabilities without making extensive changes to the legacy platform.
This architecture gives us three important advantages: reduced dependency on legacy systems, faster integration of new digital capabilities and greater flexibility to scale.
So, rather than replacing the legacy ecosystem, our strategy is to progressively decouple it through APIs, middleware and a common data layer, creating a more scalable, modular and API-led lending architecture.
For an NBFC serving customers through digital, physical and assisted channels, how do you maintain a consistent customer experience while also catering to customers with limited digital access or low financial literacy?
For an NBFC, digital cannot mean digital-only. Our customers have different levels of digital familiarity and different preferences. Some customers are comfortable using the mobile application or website, while others may prefer support through an RM or a branch.
Our objective is therefore to make the underlying customer experience consistent, irrespective of the channel through which the customer interacts with us.
For digitally comfortable customers, capabilities such as online renewals, digital payments, AutoPay and document downloads provide convenience and reduce the need for branch visits.
For customers who may not be as digitally comfortable, our RMs and branch teams continue to play an important role. They can assist customers in completing digital processes and guide them through the journey, rather than making the customer completely dependent on physical paperwork and manual processes.
The idea is to make the assisted journey frictionless and smooth as well, where the RM or branch becomes an enabler of the digital journey. For example, an RM can assist a customer in initiating a digital process, completing the required information or using the available digital services, while the underlying process remains digital.
Technology should therefore not create a divide between digital and physical channels. Instead, we want our physical and assisted channels to enable customers to move towards digital adoption comfortably.
The goal in the longer term is to give customers seamless access to all three channels – digital, branch or assisted – with identical customer experience, information and service levels.
Advances in fraud, identity theft and synthetic identities are all becoming more sophisticated. How should CIOs manage the friction of onboarding – as well as identifying, preventing and protecting against fraud?
The balance is between making the right customers move faster while identifying higher-risk cases earlier. Our initiatives around AI-based KYC document collection, document extraction and fraud detection are aimed at strengthening this part of the customer journey.
As part of our onboarding controls, we are also looking at using a combination of AI and non-AI models, along with deduplication checks, to identify duplicate or potentially suspicious customer records at the onboarding stage.
Technology can help us identify inconsistencies, extract relevant information from documents and identify potential fraud signals without requiring every customer to go through the same level of manual verification.
The ideal model is therefore a risk-based journey. Genuine, low-risk customers should experience minimal friction, while applications with unusual or suspicious signals should trigger additional verification.
At the same time, customer data protection has to be treated as a fundamental requirement. KYC and financial information is highly sensitive, so access controls, secure data handling, auditability and appropriate governance need to be built into the solution from the beginning.
For us, fraud prevention and customer experience are not competing objectives. Good technology should enable us to improve both.
With increasing dependence on cloud platforms, APIs, fintech partners and third-party service providers, how are you approaching cyber resilience, vendor risk and business continuity? Which capabilities should every BFSI technology leader prioritise?
As financial institutions become increasingly dependent on technology ecosystems, resilience has to extend beyond our own infrastructure.
Our approach needs to consider the entire ecosystem—cloud platforms, APIs, fintech partners, technology vendors and other critical service providers.
The core capabilities that every BFSI technology leader should prioritise include identity and access management, data protection, API security, security monitoring, vulnerability management, disaster recovery, backup and recovery, incident response and third-party risk management.
Vendor Audit and Assessment is particularly important because the security and resilience of our organisation can also be impacted by the controls and practices followed by our external technology partners. Vendors need to be assessed appropriately before onboarding and periodically thereafter, based on the criticality and nature of the services they provide.
An important shift is to treat cybersecurity and resilience as an enterprise and business risk, rather than simply an IT responsibility.
For every critical technology dependency, we should understand what happens if the service becomes unavailable or compromised and how quickly critical business operations can be restored.
As we increase our use of AI, APIs and external technology ecosystems, security and resilience have to grow at the same pace as innovation.
