AI Will Change the Nature of Jobs, Not the Volume, Say TCS Top Brass

Everybody is starting with a pilot—identifying an area or a use case that is relatively easy to implement, testing the waters, and ensuring the necessary guardrails are in place to meet expectations and comply with policies.

There are already a few examples where initiatives have gone into production beyond the POC stage. Some of these are operating at scale, but of course, this remains an evolving space with much more to come.

We are seeing an increasing traction in the government vertical. This includes AI-based chatbots and applications like image analytics for object recognition, among others.

BFSI is clearly active, with numerous POCs completed and several implementations already underway. These are mostly focused on areas like risk management and fraud analytics. BFSI is definitely leading the way, but that doesn’t mean other sectors are far behind.

In the telecom space, a lot of work is being done with AI and generative AI, especially in pilot projects. In manufacturing, many of the emerging use cases are on the predictive side—predictive maintenance and analytics. We're also beginning to see a few use cases in the tourism sector, again primarily focused on predictive applications.

So, across sectors, there is visible momentum, but BFSI continues to lead—at least to the best of our knowledge.

Not specifically. We are not directly measuring the impact of AI on margins and related metrics. We've always focused on improving productivity internally—not just through AI.

If you look at TCS, we have tools like MasterCraft and several others, where the goal has always been to generate more code with less effort or conduct more testing with fewer resources. So the emphasis on productivity has always been part of our approach.

With AI, tools like Copilot and others are now being used in many projects to assess potential productivity gains. But for a direct correlation to margins to emerge, this would need to happen at scale across the enterprise.

Right now, it's limited to project-specific implementations, so it’s too early to talk about its impact on margins at the enterprise level. That said, wherever we have implemented AI, we are seeing clear productivity improvements.

Every customer is asking two types of questions in response to Request for Proposal (RFPs). One type involves open-ended questions like: “Given the rest of the RFP, please suggest areas where we could benefit from the use of AI or generative AI.”

In other cases, customers have already done some groundwork and identified specific use cases, saying: “These are the areas we’re considering for AI implementation. As part of your response, can you provide an AI-based solution showing how a particular use case can be realized?” Some even ask for a proof of concept (POC) as part of the evaluation process.

During our technical bid defense, they expect a demonstration of the POC to gain confidence that we’ve understood the requirements and can implement the solution. There really is no RFP these days without AI-related questions.

In government projects specifically, RFPs are usually structured as fixed-price contracts. So, while they may not break down the costs attributed to AI separately, AI requirements are still clearly present.

Even when AI isn’t explicitly mentioned, there’s still an expectation to include what AI capabilities we can offer. During technical evaluations, it’s assumed that some AI use case will be part of the submitted bid.

Another point is that we also offer various products and platforms. We’re being asked about, and are actively planning, the integration of AI into the next generation of these offerings. Customers want to know what AI features we will incorporate into our product roadmaps and existing platforms. That’s the current landscape we're seeing.

No, we don't see any significant change in the deal pattern. As both of us mentioned earlier, there is an expectation that AI will be integrated or at least touched upon as part of the solution we offer.

Some specific deals do ask explicitly, “Can you do XYZ using AI?”—but those tend to be smaller deals. In larger deals, the expectation is more implicit: that some form of AI will be included or seamlessly integrated into the overall solution.

Obviously, we are all watchful. As of now, we are not anticipating any direct impact.

Maybe here and there, but most government deals are typically long-term. Some were signed a few years ago and are still ongoing. Others may start in six months or so, and by then, things will likely have stabilized. So, at this point, we’re not seeing any major impact.

Even in the private sector, we’ll have to wait and see how things unfold. For now, the momentum from previous periods continues.

No, we haven’t seen any drop in the number of deals. There haven’t been any cancellations or withdrawals of projects we’ve bid for. But again, we’ll need to watch how things evolve.

So, on this matter, we’ll go with what the CHRO mentioned during the earnings call. At TCS, we assess overall requirements and publish a single hiring number, which includes India as a part of the total. Arriving at this number involves extensive evaluation based on our global needs.

AI skills will certainly be required, but the percentage of total hiring that involves such skills is determined at the corporate level, based on global demand.

As you know, entry-level hiring at TCS is typically conducted through the TCS NQT and interviews. The level of difficulty varies across cadre-based roles like Ninja, Digital, and Pine. For freshers, especially those from colleges, there isn’t a strict requirement for AI knowledge. If a candidate has it, it's certainly welcome, but it’s not a mandatory filter.

For need-based hiring—meaning experienced professionals—it entirely depends on the project. These hires are made for specific roles, and if AI skills are required, they are definitely considered.

Additionally, within TCS, we run several programs to build AI competencies and skills across all levels, including leadership. Based on project or role requirements, nearly everyone at TCS has undergone some form of training on various aspects of AI.

No, actual AI technical training is being provided even at our level—to understand what a model is, how a model is trained, and all the relevant aspects of AI that we need to know. This enables us to effectively interact with CTOs or representatives from the customer organization.

At least we are not envisaging. We've heard this kind of comment through every technology change cycle. Of course, there'll be some part of what gets done today manually that will get automated. In that sense, the nature of the job will change. But the actual quantum, I don't think so.

One thing that companies like TCS have very strong is the domain and contextual knowledge. So, anything that we want to build using AI will require a solid understanding of the domain and the associated contextual knowledge. These companies are very well positioned to build products and solutions using AI in the near future.

I think it’s very important. Sovereignty as a principle is becoming very strong among all governments across the globe—not just in India. Everyone has realized that geopolitical situations can cause disruptions. Whether it’s running your operations or managing your data, you need to have security and controls in place to ensure that the government retains strategic control over these assets. So, it is absolutely essential that we have our own data models and LLMs.

Yes. For many situations, models trained on outside data may not be effective. So India-specific LLMs and SLMs are required, depending on the project needs. I think we’ll see more and more of these models becoming mainstream