Clients expecting Gen AI capabilities to be part of large deals, says Infosys CTO Tarafdar

Rafee Tarafdar, Chief Technology Officer, Infosys.

Trade

• Watchlist
• Portfolio
• Message
• Set Alert

 

 

Choose Stock Exchange: BSE LIVE NSE LIVE

26 Aug, 2025 12:21

Volume

Todays L/H

More

×

Clients are expecting Generative Artificial Intelligence (Gen AI) capabilities as part of large deals at a time when the nascent technology is becoming a part of both discretionary and operational spending, said Infosys’ Chief Technology Officer Rafee Tarafdar.

“Today, there is also a lot of pressure that our clients are having from their boards and CXOs to show some Gen AI capabilities and start making some difference,” Tarafdar told Moneycontrol in an interview. “Our clients are expecting that we bring Gen AI capabilities as part of those large deals.”

He added that AI and Gen AI are leading to the re-imagining of processes, apart from generating productivity and efficiency benefits. Consequently, the company is witnessing enterprises eliminating tasks, making them simpler and faster, thereby generating more value.

On October 29, Infosys’ Knowledge arm released a report showing that companies expect enterprise AI to boost productivity by 10-40 percent. However, many companies lack the foundational building blocks for successful enterprise AI adoption, the report revealed, which was corroborated by over 1,500 respondents and 40 senior executives in the US and the United Kingdom.

On October 24, the Bengaluru-based company launched two small language models (SLMs) built using the NVIDIA AI stack for banking and IT-specific applications. Chief Executive Officer Salil Parekh said this was prompted by clients' strong focus on Gen AI while speaking at the company's Q2 earnings press conference.

An SLM is an AI model designed to process and generate text but is smaller in size and tailored for specific tasks or industries, often using less data and computational resources compared to larger models.

Gen AI is no longer just an experimental technology. According to Tarafdar, enterprises have begun allocating discretionary budgets toward AI projects, forming dedicated teams to scale their AI initiatives.

"We are seeing clients forming projects, creating teams, and scaling, and I expect this to accelerate as more solutions move into production," Tarafdar added.

The growth of Gen AI comes at a time when the cost of deploying AI has dropped dramatically. The cost per million tokens has come down significantly—anywhere between 100x to 200x over the last two years, according to Tarafdar.

This reduction in costs, coupled with advancements in the AI models themselves, is driving a higher return on investment (RoI).

The real benefits of Gen AI, however, go beyond just improving productivity as the quality of work has also improved significantly.

“In addition to productivity, the quality of work and the quality of outcome is also improving. And when you put that along with the productivity and efficiency benefits, then it is a bit higher than what we see here as the numbers,” he noted.

Additionally, India’s second-largest IT services company observed that the rollout of Gen AI solutions is occurring in phases.

The first wave focused on productivity and efficiency, involved AI assistants in areas like coding, customer service, sales, and marketing. "Now, a lot of these today have started moving into widespread rollout and adoption,” Tarafdar explained the second wave.

“I see that a select number of our clients today have thousands of users who are already using these in order to drive productivity and efficiency," said Tarafdar.

With the second wave underway, it is now about tackling more complex use cases, particularly in areas like IT operations, business process management, risk, and compliance.

"These use cases are currently being tested with select users in pilot phases," Tarafdar further said, hinting that the next wave will involve rethinking business processes entirely.

However, scaling Gen AI is about more than just technology—it requires change management. Infosys' clients have realised that introducing AI tools isn’t enough; they need to focus on effective adoption.

“What they (clients) are doing is they are creating a small team of mentors and coaches who can tell them how to best use these tools and techniques,” Tarafdar said.

Companies are also organising hackathons to encourage collaboration and prompt engineering to solve problems creatively, along with gamification and reward programs to further drive adoption.

"A combination of these three is what also needs to be done as part of change management in order to make sure that the adoption is higher than 80 percent across the user base," Tarafdar added.