AI Playbook for financial services 2026: Insights from technology leaders in Indian financial services

Indian banks, insurers, NBFCs, and fintechs are moving AI out of pilots and into fraud detection, risk, and customer decisions, with AI agents next in line. The ambition is clear, but can the data, visibility, and evidence around those systems keep pace when a regulator, auditor, or customer asks how a decision was made?

In this report, an ET Intelligence survey commissioned by Elastic, 100 technology, risk, and security leaders across Indian financial services share where AI breaks down, how well they can defend its decisions, and what stands between them and AI agents in production.

Findings from this report conclude that:

• The problem with AI in financial services starts before the model runs: 78% say inaccurate AI output comes from the data around the model rather than the model itself, with wrong or outdated retrieval the largest single cause (42%). In real-time decisions, 64% point to the data layer, compared with just 4% who cite model accuracy.
• Few institutions can prove how their AI reached a decision: Only 16% say their AI-driven fraud decisions are fully traceable and backed by evidence. Most (64%) say decisions are mostly explainable, but only with effort, leaving 84% short of audit-ready as expectations from regulators and auditors rise.
• Governance, not technology, is holding back AI agents: 77% cite governance as a barrier to deploying AI agents, compared with 14% who cite infrastructure. Visibility is part of the gap: 70% cannot yet fully see the actions their AI agents take.

Additional resource

• How financial services companies are building contextual intelligence at scale