Indian banks shift from AI experimentation to scale, focus on credit, productivity and risk
New Delhi, August 12
Indian banks are moving rapidly from experimenting with generative artificial intelligence to deploying it across core operations, but scaling these initiatives remains the key challenge, according to the report released by FICCI, Boston Consulting Group and the Indian Banks' Association at FIBAC 2026.
The report noted the banking sector's AI playbook is increasingly centred on four priorities -- democratising credit, unlocking productivity, building risk capabilities and expanding AI-led customer engagement.
It shows that the share of banks with GenAI use cases under implementation has risen sharply, while GenAI has become a top-three strategic priority for a significantly larger proportion of Indian financial institutions. However, data and infrastructure readiness, talent and skills shortages, regulatory and governance concerns, and uncertainty over returns remain major barriers to wider adoption.
One of the biggest opportunities lies in transforming the credit journey through agentic AI. Banks could potentially reduce turnaround times by 50-90% by deploying AI agents across application, document processing, identity verification, credit assessment, fraud detection, collateral verification, sanction and disbursement. The report also envisages more than 95% first-time-right processing, a 40-60% reduction in operating costs and a 20-30% reduction in credit mortality rates.
On productivity, the report argues that the benefit of AI should extend beyond simple cost cutting. Banks can automate low-value and repetitive tasks, allowing relationship managers to redirect their time towards complex advisory, cross-selling, customer retention and relationship building. AI-assisted advisory can support activities such as next-best-product recommendations, personalised campaigns, early-warning alerts and collections.
At the same time, banks will need to strengthen their risk-management capabilities as the risk landscape expands beyond conventional credit risk. Geopolitical and climate risks require more granular modelling, while AI is making cyberattacks faster and cheaper. Rising digital interdependence is also increasing operational vulnerabilities, making machine-speed defence, real-time fraud monitoring and stronger operational resilience increasingly important.
The report therefore suggests that the next phase of AI adoption will be less about isolated pilots and more about embedding intelligent, agent-led systems into the banking operating model while strengthening governance, infrastructure and risk controls.
— ANI
Reader Comments
As someone who works in the compliance side of banking, my main concern is regulation. RBI needs to update its guidelines for AI faster than the banks are adopting it. We're talking about systems that handle crores of rupees and sensitive customer data. The mention of "regulatory and governance concerns" is correctly highlighted as a barrier — wait until a GenAI model hallucinates on a fraud detection case. Speed is good, but trust is better.
I was just reading about the FIBAC conference and this is the buzz everywhere. The 40-60% reduction in operating costs is attractive, but here's my question – what happens to the banking staff who are replaced? India's banking sector is a major employer. Are we looking at large-scale displacement? I think the report's suggestion to use AI for "unlocking productivity" should go hand in hand with serious reskilling and upskilling of the workforce. Otherwise, we're creating a social problem while solving an operational one. 🤔
From a customer perspective, I've already seen the difference in my bank's mobile app. The "next-best-product recommendations" are actually getting good — they suggested a better fixed deposit plan that matched my savings goal. The risk is in personalisation becoming creepy. I stopped using one fintech app because the AI kept sending me offers at inappropriate times. Balance is everything.
The section on geopolitical and climate risks is important and often ignored. Indian banks haven't even properly modelled heat stress risks for agriculture loans, and now we want to throw AI at everything? I hope they walk before they run. The foundation — data quality, infrastructure — is still weak in tier-2 and tier-3 cities. You can't scale what you can't trust. That said, I'm cautiously optimistic.
We welcome thoughtful discussions from our readers. Please keep comments respectful and on-topic.