Indian IT companies are developing tools and services to govern artificial intelligence (AI) agents as businesses deploy them across enterprise operations. Their work includes controlling access, monitoring performance and costs, and enforcing security policies across enterprise systems.
Tata Consultancy Services (TCS) said enterprises need “safe, secure and responsible deployment of agentic AI”. During its Q2 FY27 earnings call, TCS management said an agentic control plane was becoming a key layer for governance, cost management, agent performance and managing agents through their lifecycle.
TCS said it had won an engagement with a global insurer to build an enterprise AI platform and a control tower for in-house and third-party agents. TCS said: “As agents scale, enterprises will need more robust orchestration, access controls, security, observability, cost controls, and policy enforcement.”
Kalyan Kumar, president of HCLSoftware, said AI governance was becoming important as companies allowed agents to carry out tasks on behalf of employees. “AI governance is becoming a very important thing because you've got to scale this. You are creating like an agent which is executing works on behalf of a human. How do you govern it?” he said in an interview with TNIE.
Kumar said companies also needed to strengthen data governance and observability as agents accessed enterprise information and executed workflows. “That is where observability becomes extremely critical,” he said, adding that organisations needed to combine AI-based monitoring with deterministic and probabilistic systems.
Further, Tech Mahindra said it had established a responsible AI practice across consulting, delivery and talent enablement. It also launched Zero Gravity Telco Architecture to address constraints on safe AI agent deployment in telecom networks.
The security concerns are reflected in a Cognizant report that found 83% had experienced an AI-related security incident in the previous 24 months, while 82% had paused an AI deployment over security concerns in the previous year. Another 86% said AI had accelerated the identification of new vulnerabilities.
The survey found that 97% of respondents had increased cybersecurity budgets over the past year, with 43% raising them by more than 15%. However, organisations reporting best-in-class performance in individual areas such as visibility, detection, remediation, guardrails or security harnesses still reported incident rates of 67% to 78%, against 83% across the sample. Only 38% said they had complete visibility into AI in their environments.
Vishal Salvi, senior vice-president and global head of cybersecurity at Cognizant, said in the survey announcement: “The next step is connecting visibility, detection, remediation and governance so security operates as one system and can keep pace with how quickly AI is changing the risk environment.”
Kesavardhanan Jayaraman, founder and chief executive of K7 and chief executive of the Association of anti-Virus Asia Researchers, said: “Organisations must move beyond periodic audits towards continuous verification, testing and clear accountability, including third parties.”
He said public key infrastructure security depended on key generation, entropy, secure storage and lifecycle management, while customer data required layered access controls, segmentation and continuous monitoring. “Trust can’t be assumed from certification or years of secure operation; it must be verified continuously,” he said.
