India's AI story has moved past adoption. Now comes the hard part

In recent conversations among business leaders across boardrooms and operating teams, one theme stands out clearly: confidence in AI has become almost universal, but organisational readiness has not. AI is no longer discussed as a frontier technology in these rooms; it is assumed. What remains unresolved is whether enterprises are structurally prepared to absorb AI into how decisions are made and how execution is designed.

That reflection deepened recently at a leaders' roundtable with Sam Altman. His view was direct: we are roughly two years from early superintelligence. When you hear him walk through the progression—from solving basic math to producing original results in theoretical physics — it stops feeling like a prediction and starts feeling like a deadline. The question for every business leader in that room was the same: is your organisation structurally ready, or is it still running pilots?

Across large enterprises, the gap between AI ambition and execution has become increasingly visible. India reflects this pattern clearly. While most large Indian enterprises have invested in AI pilots, far fewer have embedded AI into core operations. A 2025 EY-CII study found that only 47% of Indian enterprises currently have multiple AI use cases running in production. This gap between confidence and execution is precisely where India's next phase of enterprise transformation will be decided.

What makes this moment different from even six months ago is that AI has moved from possibility to responsibility. Most organisations now accept that AI works. The harder decision is how much they are willing to let it change the way work actually gets done. Intent, at this stage, is not about adopting new tools, but about deciding which decisions no longer need to be made repeatedly by people, and redesigning systems around that choice.

One of the clearest shifts is the rise of agentic AI inside enterprises. Until now, AI has largely functioned as decision support—analysing data, generating insights, and recommending actions that humans still need to execute. What is changing is the willingness to trust AI with ownership of clearly defined outcomes. The inflection point comes when organisations stop asking where AI can assist humans, and instead ask which decisions humans should no longer be making repeatedly at scale. Deloitte's 2025 assessment of enterprise AI in India shows that over 80% of organisations are actively exploring autonomous or agent-style AI systems capable of executing multi-step workflows with limited human intervention. Human accountability does not disappear; it moves upstream into system design.

This shift was visible at Davos in January and has accelerated considerably since. In conversations with India's top business leaders and global technology principals over the past weeks, the discussion has moved decisively past the question of whether AI works. The question in every room is how fast enterprises can redesign around it. In that framing, India emerged as a critical reference point — not because it leads in frontier research, but because it represents the conditions under which AI must prove itself at scale. For Indian enterprises, this means AI cannot remain confined to pilots or prototypes. It has to work inside real workflows, across large teams, with clear accountability and measurable outcomes.

When AI is embedded into workflows rather than layered on top, the impact is measurable. Across operations-heavy sectors such as financial services, early adopters are reporting reductions of 30 to 50% in turnaround times for select processes after redesigning workflows around AI. These gains change cost structures and service benchmarks, and allow leadership teams to focus their attention on decisions that truly require human judgment. Equally, AI-driven forecasting is moving out of specialist analytics teams and into core financial planning and scenario analysis, becoming part of the executive decision-making workflow. Human judgment remains central—not in generating forecasts, but in asking the right questions and applying insights responsibly.

The transition to AI-led enterprises also forces a more honest conversation about reskilling. The hardest part of this shift is not frontline adoption, but the uncertainty it creates at the mid-management level. Roles defined primarily by supervision and coordination must be rethought when AI handles routine execution. What works is not training people on new tools, but redefining responsibility. When roles are redesigned around judgment, context, and problem-solving, and AI takes on repeatable work, people begin to see technology as an enabler rather than a threat.

India's operating context adds a further layer of complexity. Enterprises serving customers across languages, regions, and formats cannot rely solely on large, generic models. Smaller, focused language models trained on specific use cases and Indian languages are better suited to frontline realities. As multimodal AI matures across text, voice, and visuals, AI will move closer to the point of action.

Taken together, these shifts are reshaping India's services sector. The industry is moving away from people-centric delivery models toward outcome-driven ones, where consistency, speed, and quality matter more than sheer scale. That level of confidence is visible globally, including at forums like Davos—and now, closer home. What is notable in conversations across India's business leadership today is not the ambition; that has existed for years. What is notable is the specificity. Leaders are no longer talking about AI strategy in the abstract. They are talking about which decisions to automate, which workflows to redesign, and what accountability structures to put in place. That is a materially different conversation from the one India Inc. was having twelve months ago. But confidence alone will not deliver outcomes unless enterprises translate belief into operating change. Aspiration is not enough. Organisational intent and action are what will determine who leads and who falls behind.

The author is the vice chairman, RPSG Group. Views expressed by the expert are his own.