Insight
Individual processes are moving faster than anyone expected. The institutions running them still can't show a return most boards would call meaningful.
Insights / Loan Turnaround Dropped From Weeks to Minutes. Median ROI Is Still 10%.
India's regulated financial sector is under pressure from several directions at once: customers who now expect instant credit decisions, a rapidly expanding digital public infrastructure through the RBI's Unified Lending Interface, and a supervisor that has started asking, in detail, what "AI" actually means inside a regulated entity. Individual processes are already responding to this pressure. Rural and MSME credit underwriting that used to take two to four weeks under the RBI's Unified Lending Interface now clears in under 30 minutes. Commercial lenders using AI-driven origination report 50 to 75% faster time-to-decision, and banks running agentic KYC are cutting onboarding cost by up to 50%. These are not projections; they are what specific processes are doing right now, in production.
Yet a 2025 BCG survey of 280+ finance executives found the median realised ROI from AI in finance is just 10%, well below the 20% most firms are targeting, and fewer than half of executives can even quantify their return. We believe the gap between "this process got dramatically faster" and "this shows up as a number the CFO cites" comes down to a small number of decisions institutions have or haven't made, not to which vendor they bought.
Audited results from institutions that went past pilots, next to where most of the sector still sits. Figures shown as percentage change or adoption share.
| The upside available | ||
| Rural/MSME credit turnaround time reduction | up to 99% | |
| Commercial loan decision-time reduction | up to 75% | |
| Fraud false-positive reduction | 60% | |
| KYC/onboarding cost reduction | up to 50% | |
| Retail credit underwriting time reduction | up to 30% | |
| Where most institutions are today | ||
| Finance executives who can quantify their AI ROI | 45% | |
| NBFCs using AI in any form | 27% | |
| Median realised ROI from AI | 10% | |
We believe the difference is rarely the model itself. Across the institutions in the "upside" column above, the same handful of decisions recur.
Organisations that fundamentally redesign a process around AI, rather than inserting a model into the existing one, are the ones that show up in the upside column. Adding a tool to an unchanged process caps the gain at whatever that process already allowed.
A typical finance function runs six use cases in proof-of-concept and five in production, but few take more than ten live at once. Value comes from depth in a handful of processes, not breadth across dozens of pilots.
Underwriting, KYC, and fraud models only compress decision time when they can reach the account, bureau, and transaction data those decisions depend on. Where that access stops at a departmental boundary, so does the gain.
A model's behaviour on day one and its behaviour a year in are not the same question. Institutions that name an owner for ongoing performance catch drift before it becomes a regulatory finding; institutions that treat launch as the finish line don't.
The RBI's own data shows most regulated entities using AI at all are running simple, rule-based systems rather than the kind that produce the gains above. For an NBFC or bank board, closing that gap is now a supervisory question, not only a technology one.
Asset Reconstruction Companies reported no AI usage at all to the RBI's committee, and most regulated entities that do report using it are running simple, rule-based systems rather than anything that produces the gains above. As regulators start asking what "AI governance" actually means in practice, those institutions will need to answer that question with more than a policy document.
The number worth sitting with isn't the size of the opportunity. It's how few institutions have actually captured it, and how much of that gap comes down to the five decisions above rather than to which model was bought.
Sources: Reserve Bank Innovation Hub, Unified Lending Interface turnaround-time data; McKinsey & Company, "Banking on Gen AI in the Credit Business" (2025) and 2023 underwriting analysis; Boston Consulting Group, agentic KYC benchmarking and 2025 survey of finance executives; HSBC/Google Cloud transaction-monitoring case data; Reserve Bank of India, Framework for Responsible and Ethical Enablement of AI (FREE-AI) Committee Report, August 2025.
Talk to us about where the gap actually is for you.
Get in touch →