Indian IT companies see business in policing AI agents

Summary

Tata Consultancy Services, Cognizant, HCLTech, and Tech Mahindra have each stated that managing AI agents and governing their use are emerging as new business opportunities

Companies that raced to unleash artificial intelligence (AI) agents to book meetings, fix code and pitch clients autonomously are now scrambling to keep them in check, amid fears that these digital workers could go rogue or access data well beyond their remit.

And India's largest information technology (IT) services firms are lining up to sell them policing software. Tata Consultancy Services Ltd, Cognizant Technology Solutions Corp., HCL Technologies Ltd, and Tech Mahindra Ltd have each stated that managing AI agents and governing their use are emerging as new business opportunities.

While India's largest IT services firm, TCS, and fifth-largest Tech Mahindra are doubling down on software to coordinate AI agents and govern their limits, third-largest HCLTech is acquiring firms to strengthen its capabilities in managing these agents. Cognizant, meanwhile, has launched its own offering to govern AI usage.

Disciplining AI agents

On 8 October, TCS's management stated that "strong AI governance with secure AI platforms is becoming a top need for enterprises".

"I think as the enterprises are focused on safe, secure, and responsible deployment of agentic AI, they need a strong foundation to scale. So, building the AI platform, and more so this year, the agentic control plane is becoming a key layer not just in terms of governance, but in terms of cost management, really looking at agent performance and the full end-to-end agent lifecycle," said Aarthi Subramanian, chief operating officer, during TCS's post-earnings analyst call.

"Agents inherit the access rights of the application or the human who deployed them, they operate in silos tied to a single vendor's platform, and the controls that stop two agents from duplicating work or reaching into data they should never see are mostly manual," said Phil Fersht, chief executive of HFS Research.

Nasdaq-listed Cognizant launched its own software layer to manage AI on 23 October. It launched Cognizant AI Nexus, which measures AI spend, governs AI output quality and documents results across teams.

It also gave a sneak peek at the ways it governs AI use. Nexus captures the best IT processes and turns them into templates that teams and AI agents can reuse. Because the same system tracks both spending and quality, companies can see whether their AI is getting cheaper, better and faster. As part of this offering, each AI agent gets its own ID, permissions and memory, and the company decides what it can access. Every task is logged, and humans still sign off on important decisions and compliance checks.

While Cognizant built its own, HCLTech acquired a firm to strengthen its AI management capabilities. On 28 September, it announced its intent to acquire Croatia-based software firm Robotiq.ai to strengthen orchestration across AI agents and enterprise applications.

The rise of automation further complicates the kinds of data these agents can access. Tech Mahindra's management addressed "heightened concerns about security and reliability of AI agents".

"Trust is fundamental to scaling AI. Responsible AI is embedded throughout the solution lifecycle, with red teaming, trusted data, traceability, security, and IP provenance built into how we design, deploy, and operate AI solutions," said Atul Soneja, COO, during its 15 October post-earnings analyst call.

TCS, HCLTech, and Tech Mahindra ended the September quarter with $7.64 billion, $3.72 billion, and $1.69 billion in revenue. Cognizant is expected to announce its September quarter results on 29 October.

Growth area

The rise of AI governance tools could also be attributed to the need to complete tasks assigned to an agent.

"Orchestration is the next significant opportunity because the concept of agentic AI has moved beyond isolated AI deployments. Just as teams need to work together to drive meaningful output, orchestrating agents to work together to complete a task is important. This will help in end-to-end scale implementations," said Namratha Dharshan, chief business leader for India research at Information Services Group.

She added that homegrown tech services firms are well-poised to meet demand because of their process, domain, and industry knowledge.

Both Fersht and Dharshan expect orchestration to be a growth opportunity in the near future.

"This is a constantly evolving field and not a one-time activity. As more agents enter the system, continuous monitoring and improvement will be critical, and these platforms will be highly integrated into enterprise environments and tools to provide real-time monitoring and visibility for the organization. Especially in a highly regulated environment, this will become even more critical," said Dharshan.

Fersht said that "for Indian IT firms, the test is whether they can move from reselling a hyperscaler's orchestration stack to owning a point of view on how an enterprise should govern intelligence across its whole estate, which means investing in forward-deployed talent, in proprietary harness IP and in commercial models tied to measurable outcomes rather than headcount".