India, April 12 -- Infosys cofounder Nandan Nilekani believes that building enterprise AI at scale is challenging and takes time to mature, but the most difficult task is implementing AI in the public sector.

Speaking at the Carnegie Global Technology Summit 2025 yesterday, Nilekani said that adopting AI is relatively easier in the consumer space, but at an enterprise level, companies are putting their brand behind an AI-powered offering and even minor errors can affect their image.

He further noted that one of the reasons enterprise AI is progressing slowly is the absence of guardrails to prevent machine errors in operations.

"If the enterprise provides AI at scale and that AI even has 2% error in the way it answers, then that affects...