GenAI could boost India’s outsourcing business margins

Tech services major ServiceNow has realised $10 million in annualised tangible gains due to cost savings and increased productivity across the organisation as a result of Gen AI usage, a top executive said.

The company’s artificial intelligence-enhanced product Now Assist — which offers virtual agent customer support, customer service management, HR service delivery, ITOps management, field service management, strategic portfolio management, AI search, etc — has become the company’s fastest growing product with 11 deals exceeding $1 million in annual contract value in Q2 (April-June), Pat Casey, CTO and EVP of DevOps, ServiceNow told ET.


Casey also cited Wipro and LTI Mindtree as case studies that explain the impact AI has had. “Through the integration of Virtual Agent with Microsoft Teams, Wipro employees can now quickly find answers to requests. Common tasks such as password resets or new laptop orders can be resolved in less time than they did with a human agent. This increase in efficiency has resulted in a 40% reduction in service requests and a 25% cut in support call volumes,” he said.

The virtual agent now handles 16,000 requests a month, growing by 40% each month. At Mindtree, meanwhile, IT operations costs have come down by 50%, he said.

Benefits for India

India, a leader in the IT services outsourcing market, stands to benefit the most with Gen AI applications, since it offers an opportunity to expand margins. “(The outsourcing business) is a huge part of the Indian economy. And that business runs on razor-thin margins and has been squeezed down for years on the margin front,” Casey said, adding that Gen AI could quadruple these margins which are as thin as 5%.

When asked about the adoption of its text-to-code StarCoder AI model, Casey said that developers are using it for code completion, summarisation, snippet retrieval, etc. Internally, the model has boosted developer productivity and speed of innovation by 52%.

Will AI replace humans?

He cautioned that people are overselling the notion that AI-generated code is replacing developers. “You look on the internet, and there’s always somebody saying that developers can just sit on the beach and sip drinks, and the code will write itself. But anybody who works in the industry knows that’s just not true. It’s very much human in the loop. You need to know your code well enough to know what you want to prompt it for,” he said.

The StarCoder LLM is an open access large language model (LLM) for code generation developed by ServiceNow Research and Hugging Face. Trained with a trillion tokens, the model is 15 billion parameters in size and is trained using permissively licensed source code in over 80 programming languages.

In terms of workforce, Indian employees represent more than 20% of ServiceNow’s global workforce and Hyderabad has emerged to be its largest site it terms of employee headcount, the executive said. “The India team has evolved into a microcosm of ServiceNow globally. In India, 85% of its employees are engineers and developers, and more than 40% of our global product engineering is done from India,” he said.

When asked about building AI processing infrastructure in India, Casey said, “Right now, I don’t have India-specific GPUs (graphics processing units) on my roadmap, but that is partly dependent on how successful we are in the local market. If there’s market demand, it’s something that we would definitely look at.”