When AI becomes an agent: Economic implications for India

The advent of agentic AI, one of the latest innovations in artificial intelligence, gives humanity its first glimpse into an era where machines can work alongside humans, sometimes outperforming them and even potentially replacing them. While full replacement is still a distant possibility, the rise of these new AI systems brings unique policy challenges for our nation, particularly when we consider their potential impact on productivity, competitiveness, exports and jobs.

Agentic AI is still fundamentally built on large language models (LLMs). An LLM, which predicts the most likely continuation of a user’s prompt and simulates human-like reasoning, forms the core of an agentic AI. What sets an agent apart from a simple LLM is its ability to access its environment, gathering information from APIs, sensors, online data and user interactions. This lets the agent go beyond the limitations of a standard LLM, which can only make predictions based on the data it has been trained on.

With this environmental access, an agentic AI can tackle a single task far more comprehensively than a simple LLM. It can not only set objectives but also reason, plan, execute and monitor actions independently. Agentic AI therefore marks a significant step toward creating a true AI co-worker.

These agents are autonomous and capable of handling multiple tasks in a workflow. For instance, a single agent could manage inventory, forecasting and report preparation: tasks that would normally require several people. As a result, job displacement may not be one-to-one and could be substantially higher. Considering these realities, agentic AI embodies many of the common concerns about AI taking over human jobs.

The policy implications of agentic AI can be broadly divided into two scenarios. In the first scenario, it’s important to recognise that exports continue to play a key role in India’s growth story, even as domestic demand grows.

Total exports reached a record USD 824.9 billion in 2024–25. Some sectors, such as IT, are especially service export intensive. In 2024-25, India exported USD 387.5 billion worth of services, mainly in IT, consulting, finance and digital technologies. Export-intensive sectors like IT directly employ around five million workers, while textile industry; another export intensive sector, employs over forty million.However, the arrival of agentic AI could significantly change how India exports. Maintaining a strong export balance requires higher productivity and efficiency to maximise output at minimal cost. Countries that adopt AI more quickly are likely to reduce production costs, making their goods more competitive internationally, partly due to the non-linear job displacement we discussed earlier and partly due to the expertise of the AI itself.If rival countries in IT, apparel, fabrics, jewellery or pharmaceuticals implement agentic AI before India, it could directly affect India’s export prospects, giving importing nations more options and bargaining power.

However, AI adoption in these sectors doesn’t necessarily replace all direct workers. Instead, it can have a direct effect in sectors like IT and indirect effects in sectors such as textiles, where agentic AI is applied to backend operations like logistics.

Ultimately, this could give India’s export rivals a competitive edge, potentially leading to export erosion and disproportionate job losses, especially in high-paying sectors like IT.

Recent developments show how AI adoption is already leading to uneven outcomes for workers. On September 4, Salesforce announced over 250 job cuts in San Francisco, shortly after its CEO credited AI with reducing reliance on human labor. A few days later, on September 9, Oracle laid off more than 100 employees in India as part of a shift toward AI and cost cutting.

In contrast, Walmart is using AI to streamline operations while keeping its 1.6 million-strong U.S. workforce intact, betting that technology will transform roles into higher-paying, more technical positions rather than eliminate them.

Together, these cases highlight two sides of AI adoption: either enhancing productivity or replacing workers outright. In both situations, lower costs put pressure on exporting nations like India, which may struggle to hold on to their export share

Coming to the second scenario, let’s assume India adopts agentic AI to stay competitive. While this may look beneficial on paper, it brings a different set of challenges. Using agentic AI could disrupt the traditional Solow growth model, which assumes technology complements labour and boosts productivity.

In contrast, these AI agents may replace labor rather than enhance it. As a result, India might maintain its export share even as rising output comes with stagnant or slower GDP growth. However, implementing AI at scale in India would not be straightforward. Even if it succeeds, much of the wealth generated by these agents could flow westward, since India is still behind in preparing for the AI race.

Tools like ChatGPT already have a large user base in India, while homegrown LLMs have yet to gain significant traction. According to Mary Meeker’s AI Trends Report, India is now the largest user base for ChatGPT, surpassing the United States in monthly active users. India accounts for roughly 13.5% of global ChatGPT users, compared with 8.9% in the United States and 5.7% in Indonesia.

Ultimately, both scenarios come with their own challenges. Not adopting AI could weaken competitiveness; while adopting it risks jobless growth. The arrival of AI therefore demands a carefully designed policy response, as it is likely to expose India’s structural weaknesses in ways that are hard to predict.

On one hand, AI could boost the productivity of existing workers if used well, but on the other, it can also create the policy dilemmas we outlined earlier. This makes careful balancing and deep introspection essential to understand the spectrum of effects agentic AI could have on the economy.

For instance, policymakers must identify sectors where agentic AI poses a high risk of workforce replacement and distinguish them from sectors where the risks are lower or where AI could actually bring clear benefits. An effective response would therefore be a comprehensive plan that integrates multiple policy goals into a single, coherent framework.

Amit Kapoor is Chair and Mohammad Saad is Researcher at Institute for Competitiveness.