Technovation with Peter High (CIO, CTO, CDO, CXO Interviews) Why Citi Re-Engineers Workflows Before Applying AI
Mar 26, 2026
Shobhit Varshney, Global Head of Artificial Intelligence at Citi and former IBM data and AI leader, discusses re-engineering workflows before adding AI. He covers prioritizing high-value use cases, building a central AI platform with shared components, scaling from pilots with ROI accountability, fostering an AI-first culture, and embedding responsible AI and governance across the firm.
AI Snips
Chapters
Transcript
Episode notes
Value First Plus Platform Equals Scalable AI
- Pair a value-first strategy with platform and governance to ensure AI is not an afterthought.
- Shobhit frames this as 'doing the right AI' (value) and 'doing AI right' (responsible controls and platform).
Reengineer Workflows Before You Automate
- Re-engineer processes end-to-end before applying AI to maximize value.
- Simplify, eliminate, then automate so AI doesn't just codify inefficient workflows, a theme Shobhit calls "doing the right AI."
Central Platform Multiplies AI Value
- Centralize common AI capabilities on a platform to reduce duplication and speed time to value.
- Build shared components (e.g., unstructured content consolidation) once so one improvement lifts multiple use cases like AML and fraud.