Only 14% of Indian Companies Have Full Visibility Into AI Agents: Okta Research

While Indian organisations are working toward AI-powered digital experiences, many have not yet built the governance and access control systems needed to manage AI agents and non-human identities. The findings are based on polls conducted by cloud-based identity and access management platform Okta at Okta AI Identity Summits held in Mumbai and Bengaluru.

According to the research, close to 73% of respondents in India said their organisations were either very prepared or somewhat prepared to deliver AI-powered experiences across applications, customer services and AI agents. 

Okta noted that this level of preparation contrasts with the governance measures currently in place, describing a gap between adoption plans and the controls needed to support secure deployment.

The polls found that only 14% of respondents had full visibility into all AI agents and non-human identities operating within their organisations. A further 17% said they could restrict or suspend an AI agent across all environments if required, while 51% said this was possible only in some cases.

On managing non-human identities such as service accounts, APIs and AI agents, 17% of respondents described themselves as very confident, and 53% said they were somewhat confident. When asked about access governance, only 14% said they were very confident that AI agents were limited to the systems and data required for their tasks.

"As organisations move beyond experimentation and begin embedding AI into business processes, governance becomes just as important as innovation," Matthew Graham, Chief Security Officer for APJ at Okta, said in a statement. 

He added that organisations gaining the most value from AI would be those building visibility, accountability and security into their AI work from the start.

Graham also pointed to a shift in how organisations are approaching AI deployment. "The conversation around AI is shifting. The focus is no longer solely on adoption," he said. 

He noted that organisations are now focused on governing AI consistently, managing access, and maintaining trust as AI agents become more embedded in daily operations, and said identity management has a role in addressing these questions.

Okta said the findings point to a need for stronger visibility, lifecycle management, and access governance as AI agents, service accounts, APIs and other non-human identities become more connected to enterprise systems and data in India.