Why AI Agents Break Down In Production Environments: VeloDB

6 Nov 2026, 1:52 pm · 4h ago · 1 min read · YourStory

Agentic AI deployments frequently fail when transitioning from pilots to live production due to inadequate data architecture, according to VeloDB APAC Regional Director Nitin Waghulde. Research firm Gartner projects over 40% of agentic AI initiatives will be abandoned by 2027 due to rising costs, unproven business value, and governance issues. Waghulde highlighted that AI agents consume data differently than human analysts, transforming single-query reporting tasks into hundreds of unstructured lookups across CRM, ERP, and support databases. Consequently, enterprises must redesign underlying data infrastructure to ensure agents can scale reliably in enterprise workflows.