Building Responsible Agentic AI: Genpact's Blueprint For Enterprise Leaders
Tech Talks Daily
Designing observability and explainability early
Han explains Genpact's command center for real-time observability, health signals, storytelling, and embedded responsible AI.
In this episode of Tech Talks Daily, I sat down with Jinsook Han, Chief Agentic AI Officer at Genpact, to unpack one of the most misunderstood shifts in enterprise AI right now.
Many organizations feel confident about the value AI can deliver, yet only a small fraction are able to move beyond pilots and into autonomous operations that actually scale. Genpact's Autonomy By Design research puts hard data behind that gap, and Jinsook explains why optimism often races ahead of readiness.
We explore why agentic AI changes the rules entirely. When AI systems begin to act, decide, and adapt on behalf of the business, familiar operating models start to strain.
Jinsook makes a compelling case that agentic AI cannot be treated like another software rollout. It demands a rethink of data, governance, roles, and even how teams define work itself. The shift from tools to teammates alters expectations for people across the organization, from frontline operators to the C-suite, and exposes just how unprepared many companies still are.
Governance is a major theme throughout the conversation, but not in the way most leaders expect. Rather than slowing progress, Jinsook argues that governance must become part of how work happens every day.
She shares how Genpact approaches agent certification, maturity, and oversight, using vivid analogies to explain why quality and alignment matter more than simply deploying large numbers of agents. We also dig into why many governance models fail, especially when they rely on committees instead of lived understanding.
Upskilling sits at the heart of this transformation. Jinsook walks through how Genpact is training more than 130,000 employees for an agentic future, starting with executives themselves. The focus is not on abstract learning, but on proving that today's work looks different from yesterday's. Observability, explainability, and responsible AI are woven into this approach, with command centers designed to monitor both agent performance and health, turning early signals into opportunities rather than panic.
This conversation goes well beyond hype. It is about readiness, responsibility, and the reality of building autonomous systems that still depend on human judgment. As organizations rush toward agentic AI, are they truly prepared to change how decisions are made, how people work, and how accountability is defined, or are they still treating AI as a faster hammer rather than a new kind of teammate?
Useful LinksThe AI-powered Podcast Player
