Over the last few years, we’ve moved from just experimenting with AI to having our expectations set by it. Today, AI sits at the center of boardroom conversations across industries. Leaders are investing aggressively, hoping to unlock productivity, improve decision-making, and build long-term competitive advantage.
However, on deeper and closer observation, a more complex picture begins to emerge.
According to NASSCOM, despite growing adoption, only about 20% of organizations report actual revenue impact from AI so far. And IBM’s research shows that while 42% of enterprises have deployed AI, another 40% are still stuck in experimentation.
Where most companies are struggling is with this gap between adoption and results.
In many places, AI is treated as something you can buy and deploy quickly. But enterprises don’t work like that. Each organization has its own way of functioning, making decisions and scaling up. When AI is introduced without understanding this context, it operates in isolation. You end up with something that looks powerful but doesn’t quite fit.
Even in India, where adoption is accelerating faster than the global average, the same challenge exists.
Compared to the global average of 28%, nearly 40% of organisations report using AI significantly. While 97% of respondents anticipate increased productivity from AI, most of them are still more concerned with small-scale enhancements than with real change.
The effectiveness of the AI systems depends heavily on the context. Models that were trained on generalised data now struggle to understand nuances unique to specific firms. When not integrated into critical systems, these models are not included in the decision-making processes. Additionally, without frequent and punctual updates to align with business changes, they swiftly become obsolete. Due to this, many businesses often find themselves investing a lot without getting the desired returns.
Measuring the productive results of the rollout is one of the largest issues businesses encounter while implementing AI technologies.
In spite of increasing AI deployment, about 70% of organisations continue to lack a structured and defined framework in order to quantify ROI, according to a recent industry research. Without such clarity, it becomes difficult to separate real value from perceived progress. The shift that needs to happen now is from adoption to alignment.
At CloudThat, we have seen this play out in our own journey. Even for something as widely adopted as Microsoft Copilot, we did not approach it as a simple license rollout. We implemented a structured, multi-level training and assessment program internally to evaluate readiness and real usage. Access was granted based not only on availability but also on how well teams could use it in their daily job.
Instead of being just another underutilised tool, this made sure that adoption was in line with results from the beginning.
Early indications of this change are already apparent. Businesses are investing in stronger data foundations, shifting toward domain-specific AI, and emphasising integration over discrete use cases. They are also realising how important talent is. Technology is insufficient on its own.
You need people who understand both the business and the technology to make it work.
This becomes even more important when you look at India’s ambition to become a Viksit Bharat. AI is going to play a central role in that journey. It has the potential to reshape sectors like healthcare, manufacturing, financial services, and public governance in a meaningful way.
But that won’t happen if we keep relying on one-size-fits-all AI solutions built for very different markets. India’s challenges, data environments, and sheer scale are unique, and the systems we build need to reflect that reality.
Which brings us to talent. We don’t have a shortage of technical capability; India already has a strong base in engineering and technology. The real opportunity now is to layer that with domain expertise, so AI isn’t just deployed, but applied in ways that actually solve problems and drive economic value.
At a national level, it’s not about how much AI we adopt. It’s about how well we use it.
