Your cart is empty
Browse transcripts and add items to get started.
An enterprise AI leader with two decades inside Indian banking and financial services unpacks why generative and agentic AI pilots move fast but stall on the way to production. He walks through what a real production-grade architecture looks like — integration layers, RAG, business-domain agents, DAG-based orchestration and day-one observability — and why regulated decisions like loan and claims underwriting stay human-in-the-loop while productivity, customer service and internal automation scale fastest. Legacy systems, regulator traceability, information-security risk and hard ROI thresholds emerge as the real constraints separating a working POC from a deployed system.
This expert call examines the execution gap between AI experimentation and production inside Indian financial institutions. Building a proof-of-concept in a GenAI world is now easy; taking it into production at scale is where "hell breaks loose" — integration with legacy monoliths, compliance and information-security review, regulator demands for traceability, and the hunt for defensible ROI. The expert details how production architectures are converging on agent-plus-API designs — an integration layer, vector-database RAG, business-domain agents, DAG-based orchestration, and observability built in from day one — and explains why financial decision-making stays human-led even as autonomous capabilities are deployed in lower-scrutiny, non-customer-facing workflows.
