India’s IT workforce is undergoing a structural shift. Software engineers who once spent their time writing and debugging code are increasingly moving into supervisory roles as agentic AI takes over core development tasks.
Recently, even OpenAI CEO Sam Altman wrote on X: "I have so much gratitude to people who wrote extremely complex software character by character. It already feels difficult to remember how much effort it really took. Thank you for getting us to this point."
Unlike earlier forms of generative AI, which could answer questions or draft emails, agentic AI goes much further. It can plan, execute, and complete complex software projects on its own from beginning to end. It writes code, tests it, and fixes bugs without waiting for human instructions.
As a result, today’s engineers are becoming gatekeepers. They define business rules, set security standards, and supervise the AI’s work, making sure it does not make errors or introduce risks.
“I haven’t written a raw block of functional code in three months,” said Rahul Kumar, a 28-year-old software engineer at a leading IT company in India. “My job has completely changed. I now give the AI a design plan, and it produces the entire module in minutes. My role is to review its output, checking for security issues, efficiency, and whether it meets our needs. I’m less of a builder now, and more of an inspector.”
This view is widely shared across the large campuses of India’s major IT firms, many of which have introduced hundreds of AI agent licenses since late last year. The shift is moving workers up the value chain and reducing the need for repetitive manual tasks.
Pooja Suresh, a technical lead at a global technology company, says this transition requires a completely different skill set. “Earlier, you were valued for knowing the details of a programming language. Today, the AI understands the language better than any person,” she said. “What it lacks is context and judgement. My team focuses on ‘spec-driven development’. We define clear rules and parameters, the AI does the heavy lifting, and we review the final result.”
Indian technology firms such as Wipro, Infosys, TCS, Tech Mahindra, and HCLTech are already building advanced agentic systems for global clients.
Speaking at the World Economic Forum in Davos earlier this year, C Vijayakumar, CEO of HCLTech, highlighted the vast potential of autonomous systems. Referring to the rapid pace of development, he said that “physical AI, which can sense and understand actions, is a major opportunity,” and estimates suggest it could become a trillion-dollar industry by 2030.
However, this trillion-dollar future still depends heavily on human oversight. Companies understand that allowing an autonomous system to operate without human supervision could create serious risks.
“The biggest fear is an AI agent quietly introducing a critical error into a live system because it misunderstood a prompt,” said Karthik Narayan (name changed), a cloud architect at Wipro. “You can’t simply trust a black box. We act as the safety net. We review the AI’s reasoning, test its dependencies, and give the final approval. The machine is fast, but we are responsible.”
For the Indian IT sector, the era of the 'code monkey' is officially over. As agentic AI takes the wheel, the traditional developer is stepping back to become the navigator.