Most enterprises are gaining value from AI. Few trust it to act on its own. That is the central finding of a new global study from Tata Consultancy Services (TCS).
Some 84% of executives rank productivity as the leading source of value from their AI investments. But only 10% of enterprises let AI agents carry out decisions without human approval across several functions or business areas. Just 9% say they have AI agents that can securely orchestrate workflows across enterprise applications.
The study is called TCS Enterprise AI Study: Intelligence that is Real. It points to a gap between the value firms get from AI-assisted work and their readiness to give AI agents more authority.
Security Leads the Agenda
Security, risk and governance rank as the top factor shaping how and where firms build and deploy AI. They come ahead of cost, speed and ease of deployment. This matters more as enterprises give AI greater access to applications, workflows and decision-making.
TCS sees a shift toward a Human+ AI operating model. AI moves from helping employees to taking on more responsibility and, where suitable, working with more autonomy. Getting more value along the way depends on having the right security, governance and controls. They decide where human oversight stays essential and where AI can act on its own.
Amit Kapur, Chief AI and Services Transformation Officer at TCS, said AI is moving from assisting people to taking more autonomous action, with deep access into the enterprise. Productivity is where firms see the most value today, he said. However, 10% also report a rise in revenue. He said concerns about security, governance and resilience are holding organisations back, and rightly so. The next stage will focus on deploying AI responsibly, with the right guardrails and enterprise controls.
Independence and Sovereignty
Executives also care about resilience and flexibility. Some 90% say model independence is an important factor when they decide on AI investments. And 34% rate AI sovereignty requirements as a high or very high organisational risk. These cover control over data location, model choice and compliance.
Concern is highest in Europe, where 43% rate sovereignty as a high or very high risk. Yet only 7% of global respondents have organisation-wide plans for switching providers. This suggests worry about model dependence and sovereignty is moving faster than operational preparation.
Other Findings
The study found that 39% of enterprises are still checking whether AI projects deliver enough value compared with token and compute costs. Another 44% say important business knowledge sits with employees or in scattered documents. That may limit AI’s access to enterprise context. Some 46% report targeted use of open-source AI for selected tasks.
AI’s effect on jobs is becoming harder to read. In 2024, 4% of leaders said it was too early to tell. In 2026, that figure is 19%.
After productivity, firms see value in better customer experience and in new products or services.
Regional Differences
North America stands out on several enterprise AI foundations. About 15% of respondents there treat AI as a strategic enterprise capability. Europe is focused on AI sovereignty, with 43% citing it as a high or very high risk. The UK and Ireland stand out on operational resilience. They report comparatively strong AI continuity planning and governance integration.
How the Study Was Done
The study surveyed 1,333 senior business and technology leaders in 25 countries and 12 industries. Respondents included C-suite executives, divisional leaders, and senior vice presidents and vice presidents. Their organisations had annual revenues from US$500 million to more than US$20 billion.
