The surge of attention around recent artificial intelligence releases has amplified fears across markets even though the underlying technology has existed for years, Mphasis Global Delivery Head Ravi Vasantraj said, arguing that visibility – not capability – has driven the latest reactions.
In an interaction with FE on the sidelines of the Nasscom Technology and Leadership Forum, Vasantraj said tools that automate or convert legacy COBOL code into Java have long been available in the market. What has changed, he said, is the prominence given to such tools following high-profile releases from companies like Anthropic, which has led to outsized interpretations of their immediate impact. Markets, he added, tend to overestimate short-term effects while underestimating how technology adoption actually plays out over time.
Visibility Trap
According to Vasantraj, the heightened focus on AI is nevertheless forcing a shift in enterprise behaviour. Increased scrutiny pushes employees, service providers and customers to accelerate decision-making and experimentation, helping reset expectations around how AI can be deployed within organisations. These moments, he said, should be viewed as an adjustment phase rather than a disruption to existing delivery models.
He was critical of recent commentary suggesting that falling costs of “AI intelligence” could sharply erode employment in labour-intensive markets such as India. Referring to a Citrini Research report that predicted AI agents could become as cheap as electricity, Vasantraj said such arguments oversimplify a far more complex transition. Global enterprises, he said, continue to work with firms like Mphasis not because they lack access to low-cost labour, but because of the value created through execution and scale.
Beyond Code
Vasantraj said much of the current narrative ignores the realities of enterprise technology delivery. While artificial general intelligence could emerge sooner than expected, fully autonomous and unmonitored AI systems remain some distance away. Writing code, he said, is only a small part of the process. Making applications production-ready involves navigating security protocols, data governance frameworks and intellectual property standards, which cannot be automated away easily.
From discussions with clients, Vasantraj said enterprises remain cautious about direct AI integration. A large proportion of proof-of-concept projects fail, even as announcements around new models generate headlines and lift valuations of AI developers. In his view, markets and observers often focus on promised outcomes while overlooking the execution journey that determines whether technology can be deployed at scale.
On the business environment, he said demand has remained resilient. Mphasis continues to hire, including at the entry level, with contract closures rising over the last three quarters. Enterprises are also beginning to address accumulated technology debt, which had earlier slowed transformation spending.
Vasantraj said AI-led automation in legacy modernisation could ultimately work in favour of service providers by shortening delivery cycles. Tasks that once took weeks can now be completed in hours, lowering costs and accelerating onboarding, even as enterprises remain cautious in how they deploy the technology.