AI-native lending will replace static credit decisions with continuous risk intelligence: Vartis CTO

Artificial intelligence is fundamentally changing how lenders assess credit risk, moving the industry away from one-time underwriting decisions towards continuous, real-time risk intelligence, according to Dipesh Karki, Co-founder and CTO of Vartis Platforms.

The Mumbai-based fintech, which last year unified its lending businesses—LenDenClub, InstaMoney and VartisOne—under a single technology platform, is betting heavily on AI-native lending as it works towards its target of ₹1,000 crore in revenue by FY28. The company has already rebuilt its lending infrastructure around RBI's digital lending guidelines while embedding AI across underwriting, servicing and portfolio management.

"Traditional credit models offered only a snapshot of a borrower's financial health," Karki said. "AI allows us to evaluate a much broader set of signals—from banking behaviour and repayment patterns to transaction data and alternative datasets—to build a more dynamic view of creditworthiness."

The bigger shift, he argues, is that underwriting no longer ends once a loan is approved. Instead, AI enables lenders to continuously monitor portfolios, identify emerging risks early and intervene before borrowers slip into default.

Predictive analytics is also reshaping collections by helping lenders move away from uniform recovery strategies. Rather than treating every delinquent borrower the same way, institutions can prioritise accounts based on repayment propensity, recovery potential and behavioural patterns, making collections more targeted and efficient.

For lenders under pressure to reduce loan turnaround times, automation has become equally important. Karki believes speed and governance are no longer opposing objectives. Automated workflows can perform identity verification, KYC, bank statement analysis and policy execution while ensuring every application follows the same decision framework. The result is faster approvals, improved auditability and stronger regulatory compliance.

However, he cautions that AI adoption is rarely limited by algorithms alone.

"The biggest challenge isn't building the model. It's fragmented data, legacy technology and disconnected workflows," he said. "Moving AI from experimentation into live lending operations requires organisational transformation across technology, risk, operations and compliance."

The company is investing in OmniCredit, its AI-powered credit decisioning platform, allowing risk teams to simulate lending policies, compare strategies and assess portfolio impact before deploying changes. Alongside it, InstaMoney uses automated decisioning to expand credit access for underserved borrowers, while NeuraVoice is exploring AI-driven voice interactions for servicing and collections.

As AI assumes a larger role in credit decisions, explainability and governance become equally critical, Karki said. Risk teams must understand why models reach particular decisions, while human oversight remains essential for policy reviews and high-impact lending decisions.

Looking ahead, he expects India's digital public infrastructure—including Aadhaar, UPI and the Account Aggregator framework—to serve as the foundation for AI-native lending.

"AI will become the intelligence layer on top of India's digital rails," he said. "The next phase of financial inclusion won't be driven by more data alone, but by better interpretation of that data."