LTIMindtree Embraces AI Adoption Strategy, Eschewing In-House Development

'Differentiator of LTIMindtree is AI Adoption, Rather than Another Shiny Toy'

LTIMindtree has decided to be transparent and said that there is no point in spending time and money on building AI products when so many have already been built. Ask an Indian IT company about how it uses AI, and you will learn nothing. This highlights why such companies remain service-driven rather than product-focused. Many of them also shy away from revealing the revenues they earn from their investments in generative AI. To get some clarity about why this is the case, AIM spoke with Nachiket Deshpande, COO at LTIMindtree, and he did not hold back. Commenting on the recent deal with Voicing AI, in which LTIMindtree committed $6 million to build human-like AI voice agents, Deshpande said, "There are so many startups that are coming all around the world. We would want to leverage those startups for those technologies that are coming up." Elaborating on the philosophy that LTIMindtree follows when it comes to AI, Deshpande said the company stands on three pillars: AI in everything, everything for AI, and AI for everyone. He added that AI solutions must integrate seamlessly into users' existing workflows. For example, if someone uses SAP, ServiceNow, or Outlook daily, the AI solution should meet them there; not as a separate platform. "This is what we refer to as the co-pilot approach, ensuring AI systems are agentic, API-driven, and omnipresent." "Over the last two years, there's been overwhelming noise around generative AI. My challenge was to build a strategy that is simple to adopt, enabling all 86,000 employees at LTIMindtree to align with it," Deshpande revealed. "More importantly, I wanted a pragmatic approach: something relevant to us and not just general AI industry discourse." Unlike other IT companies like Infosys or TCS, which are claiming to be building in-house generative AI solutions, LTIMindtree has decided to be transparent and said that there is no point in spending so much time and money on building AI products when so many have already been built. "Any technology we develop today risks becoming obsolete in a matter of months, and keeping up would require significant capital investment. The P&L structure of services companies like ours is fundamentally different from product companies," Deshpande explained. "They operate with 80% cross-market, and hence they have the ability to continue to do R&D, and we operate at 30-35% cross-market." He believes that if a company invests and builds a particular technology like AI, it might soon become irrelevant. "I need to get 86,000 people to reimagine their work with AI, but I only need 2,000 people to build AI solutions," he said. According to him, even that number is high and requires a lot of investment. "The differentiation of LTIMindtree will lie in terms of how we adopt AI, rather than saying I have another shiny toy which is better than somebody else. Because that differentiation is short-lived," he added. Deshpande explained that if a company wants to take an AI solution from proof of concept (POC) and roll it out to 10,000 customer service agents, it needs all of this to be built within the organisation in advance. Following the three pillars of its philosophy, LTIMindtree wants to reimagine every service it offers with AI - to move AI solutions from POC to large-scale deployment. For instance, rolling out solutions to 10,000 customer service agents. This creates a new revenue opportunity for us as customers invest in these capabilities. Apart from Voicing AI, the IT firm also closed a $240 million deal in the manufacturing sector, where it replaced three competitors by consolidating the client's application maintenance and running a portfolio with an AI-first approach. Since then, LTIMindtree has announced five to six similar AI-centric wins, with a large portion of the pipeline also reflecting this strategy. These are just a few of the examples. One of the most successful generative AI applications by LTIMindtree is Canvas, which is a platform specifically designed to take AI-driven ideas into production. Deshpande said that over 13 customers live on the platform, which integrates seamlessly with their other AI investments. "I predict every single dollar of revenue will have a generative AI component embedded in it," he revealed. Many IT firms like Mphasis and Persistent Systems are increasingly focusing on agentic AI. Larger players like Infosys and TCS are also building multi-agent systems. Deshpande believes this shift is happening because the biggest productivity gains lie in persona-centric AI solutions. For example, in a team, small productivity improvements for every individual don't create a monetisable impact. "But if a developer delivers better story point velocity, a support engineer resolves more tickets, or testing cycles shrink from weeks to days, the gains become significant. Agentic systems automate large portions of a persona's workflow, delivering tangible productivity outcomes," he added. When asked about why all Indian IT companies do not reveal the revenue from generative AI, Deshpande said that for LTIMindtree, almost every service has AI embedded in it. "It is not a revenue stream for IT services companies. For NVIDIA, it will be a revenue stream. For Microsoft, it will be a revenue stream." For LTIMindtree, however, it will not be a bigger revenue stream. Instead, it will be the key revenue stream. Considering that the impact of AI is much bigger than any revenue component any of the service companies will give, Deshpande concluded, "Counting AI revenue is not a measure that we would like to go after because I think it misrepresents the impact of AI.

Agentic AI is Now On Mid-Sized Indian IT's Mind

The biggest reason for not building in-house models for Indian IT firms is that generative AI is too expensive to experiment with. The era of agentic AI is upon us, and surprisingly, Indian IT is also ready for this revolution. What is even more surprising is the fact that mid-sized and small IT firms are also ready for adoption. As usual, they are taking the acquisition route to achieve these capabilities. The latest example is LTIMindtree. Last week, the IT firm announced that it has committed $6 million to Voicing AI, a US-based startup specialising in human-like AI voice agents. This investment aims to bring human-like voice capability across more than 20 languages with conversational, contextual, and emotional intelligence. Moreover, LTIMindtree has formed an alliance with GitHub for its Copilot coding tool to train its workforce on the GitHub ecosystem. Similarly, Mphasis has focused on conversational AI to boost user experiences. In July, the IT solutions provider launched NeoCrux, a tool for improving developer productivity by streamlining software developer cycles with AI agents and orchestrators. In October 2023, the company acquired Silverline, a New York-based Salesforce partner, for $132.5 million. During the announcement, Nitin Rakesh, CEO of Mphasis, said, "The acquisition aligns with our strategy to drive customer experience transformation, modernise contact centres, and enable conversational AI automation to meet evolving client needs." It seems evident that Mphasis has decided to embrace AI and agentic AI for customer interactions as well. Though it is not clear whether the acquisition of Silverline increased Mphasis's current capabilities, the impact is definitely visible. Persistent Systems, on the other hand, is addressing AI-related privacy challenges. In September, the company acquired Pune-based data privacy consultancy Arrka for ₹14.4 crore ($1.7 million). "Arrka's mature frameworks and privacy management platform ensure a scalable and governance-driven approach, critical for successful AI implementations," Persistent Systems CEO Sandeep Kalra said in a press release. Meanwhile, bigger IT giants like TCS, Infosys, and Wipro have decided to develop a different strategy for building AI agents for their customers. In the latest quarterly results, TCS posted $1.5 billion in bookings with AI and Infosys has started building small language models and multi-agent frameworks for its clients. A similar situation was observed at Wipro, HCLTech, and Cognizant. TCS, India's largest IT services exporter, launched a new AI-focused business unit, AI.Cloud, in May, by merging its cloud and AI divisions. Infosys, the second-largest IT firm, unveiled Topaz, an AI suite offering solutions and services. Notably, it has several clients in its portfolio. The mid-sized IT companies are, however, positioning themselves differently here. Speaking with AIM, Nachiket Deshpande, COO at LTIMindtree, said Indian IT firms often do not want to build these capabilities in-house. Commenting on the recent deal with Voicing AI, Deshpande added, "There are so many startups that are coming all around the world. We would want to leverage those startups for those technologies that are coming up." Deshpande's biggest reason for not building in-house models is that generative AI is so expensive that it doesn't make sense to experiment with it while investing so much. "I don't think the technology in generative AI would be the real differentiator because of the level and the speed of innovation that is happening, whatever technology you may develop will become obsolete within a few months, and you have to keep doing that," Deshpande said, adding that it requires a lot of capital. Mohandas Pai, the founder of Aarin Capital and former CFO of Infosys, also told AIM that IT services companies are not built to invest disproportionate capital R&D. "Indian IT services companies are not product companies." "Creating an LLM or a big AI model requires large capital, time, a huge computing facility, and a market. All of which India does not have," Pai pointed out, adding that even though Infosys, TCS, and others might have the funds, their focus is to provide vertical solutions and not horizontal ones like ChatGPT. Similarly, Deshpande also said that the services companies' P&L is structured differently than that of product companies. "They operate with 80% cross-market, and hence they have the ability to continue to do R&D, and we operate at 30-35% cross-market," he explained. He believes that if a company invests and builds a particular technology, tomorrow it might become irrelevant. "I need to get 86,000 people to reimagine their work with AI, but I only need 2,000 people to build AI solutions." "The differentiation of LTIMindtree will lie in terms of how we adopt AI, rather than saying I have another shiny toy which is better than somebody else. Because that differentiation is short-lived," he added. Apart from the technological capabilities, Deshpande said it's important that IT companies give startups the space they need to grow on their own. "Hence, acquisition was not the best way, but at a meaningful scale was a better way where we get to leverage the technology, we get to take them to the customer, but also provide those entrepreneurs space to continue their innovation," he further said. With generative AI systems going autonomous and several big-tech firms announcing models like Devin and others, Indian IT is also ready to adopt that change quickly. Deshpande said that one of the biggest reasons for this is the outlook on productivity. "The idea of productivity outlook is persona-centric, and you have to look at each persona becoming more and more productive," Deshpande said. Hence, he said that agentic systems, which do not examine a particular business process or task but the entire persona and try to automate large parts of those personas, are the way to adopt that productivity. That's why agentic systems are becoming increasingly focused. Earlier this year, small IT firms were taking the acquisition route for building AI capabilities. Happiest Minds, Hexaware, Quest Global, Coforge, Sonata, and GlobalLogic have all announced acquisitions in this space. Now, the conversation has shifted to agentic AI.