Genpact Architect outlines enterprise AI strategy

In a newly published interview, Genpact Chief Enterprise Architect Jayant Swamy outlined how large organizations must restructure their workflows to support autonomous artificial intelligence. Swamy, who joined the global professional services firm in 2024 after leadership roles at Xtrac8.Tech, Accenture, Fannie Mae, and Oracle, emphasized that successful AI deployment relies heavily on surrounding architecture rather than just the underlying models. Genpact itself, which originated as a General Electric Capital initiative in 1997 before going independent in 2005 and public in 2007, is currently focusing its efforts on agentic systems and automated enterprise operations.

Swamy cautioned that common customer-facing metrics like containment rates and speed can mislead businesses. While a high containment rate suggests efficiency, it often hides unresolved issues where customers simply give up. Instead, he argued that practitioners should measure downstream signals, such as whether a customer query remains resolved 30 days after the initial contact. To prevent autonomous agents from optimizing for speed at the expense of trust, Swamy advised building balanced objectives and strict guardrails directly into the system's orchestration layer.

For enterprise developers and architects, this shift requires designing clear escalation thresholds based on financial and regulatory risk. Routine tasks like password resets can run with full autonomy, but high-stakes transactions must automatically route to human workers. Swamy noted that the human role is evolving from executing tasks to supervising and auditing agent networks. Ultimately, successful integration requires a robust data layer that provides agents with governed, role-based access to systems of record without exposing sensitive information.