At a start-up office in Bengaluru, India, developers are fine-tuning artificial intelligence (AI) chatbots that talk and message like humans.
The company, LimeChat, has an audacious goal: to make customer-service jobs almost obsolete. It says its generative AI agents enable clients to slash by 80 percent the number of workers needed to handle 10,000 monthly queries.
“Once you hire a LimeChat agent, you never have to hire again,” said Nikhil Gupta, its 28-year-old co-founder.
Illustration: Mountain People
Cheap labor and English proficiency helped make India the world’s back office — sometimes at the expense of workers elsewhere. Now, AI-powered systems are subsuming jobs done by headset-wearing graduates in technical support, customer care and data management, sparking a scramble to adapt, a Reuters examination found.
That is driving business for AI start-ups that help companies slash staffing costs and scale operations — even though many consumers still prefer to deal with a person.
This account of the disruptive changes transforming India’s US$283 billion information technology (IT) sector is based on interviews with 30 people, including industry executives, recruiters, workers, and current and former government officials.
Reuters also visited two AI start-ups and tested voice and text chatbots that handle increasingly sophisticated customer interactions in human-like ways.
Rather than pump the brakes as the technology threatens jobs built on routine tasks, the country is accelerating, wagering that a let-it-rip approach would create enough new opportunities to absorb those displaced, Reuters found.
The outcome of India’s gamble carries weight far beyond its borders — a test case for whether embracing AI-driven disruption can elevate a developing economy or render it a cautionary tale.
The global conversational AI market is growing 24 percent a year and should reach US$41 billion by 2030, consultancy Grand View Research estimated.
India — which relies on IT for 7.5 percent of its GDP — is leaning in. In a February speech, Indian Prime Minister Narendra Modi said: “Work does not disappear due to technology. Its nature changes and new types of jobs are created.”
Not everyone shares Modi’s confidence in India’s preparedness. Santosh Mehrotra, a former Indian official and visiting professor at the University of Bath’s Centre for Development Studies, criticized the Indian government for a lack of urgency in assessing AI’s effects on India’s young workforce.
“There’s no game plan,” Mehrotra said.
Business process management employs 1.65 million workers in call centers, payroll and data handling in India. Hiring has plummeted due to increased automation and digitalization, despite rising demand for AI coordinators and process analysts, said Neeti Sharma, chief executive officer of staffing firm TeamLease Digital.
Net headcount in the segment, which represents one-fifth of IT output, grew by fewer than 17,000 workers in each of the past two years, down from 130,000 in 2022 and 2023, and 177,000 in 2021 and 2022, TeamLease Digital figures showed.
Reuters spoke to three current and five former customer-service workers, who described increasing job insecurity and integration of AI, including tools that suggest responses and bots that handle nearly all routine queries autonomously.
Megha S., 32, was earning US$10,000 a year at a Bengaluru-based software solutions provider. She said she was laid off last month, just before India’s festive season, as the company moved to implement AI tools to review the quality of sales calls.
“I was told I am the first one who has been replaced by AI,” said Megha, who spoke on the condition that her full name and former employer not be identified.
“I’ve not told my parents,” she said.
Former Indian Ministry of Labour and Employment official Sumita Dawra who oversaw an Indian government taskforce on AI’s impact on the workforce before retiring in March, said while the technology offered productivity gains that would lead to new jobs, India could consider stronger social security measures, such as unemployment benefits, to help those displaced during the transition.
However, a senior Indian official said that the Indian government believed that AI would ultimately have little impact on overall employment.
India’s IT and labor ministries, and Modi’s office, did not respond to requests for comment.
Besides AI, factors clouding the outlook for India’s IT sector include US tariffs; a proposal by a US lawmaker for a 25 percent tax on firms using foreign outsourcing services, and US President Donald Trump’s US$100,000 fee on new H-1B visas, which are widely used by tech firms to sponsor Indian workers.
Investment bank Jefferies last month predicted that India’s call centers would face a revenue hit of 50 percent — and about 35 percent for other back-office functions — from AI adoption over the next five years.
That would spell near-term job losses in India, which accounts for 52 percent of the global outsourcing market.
“The biggest impact is going to be on young students coming out of college,” said Pramod Bhasin, who in the 1990s established India’s first call center with 18 employees for GE Capital, where workstations were partitioned by saris strung from the ceiling.
In the longer run, India could transition from “back office” to the world’s “AI factory” by capitalizing on demand for AI engineers and automation deployment, said Bhasin, who went on to found IT services firm Genpact.
One beneficiary of that demand is LimeChat, which Reuters visited in August. Gupta said his developers and engineers have helped automate 5,000 jobs across India. The company’s bots handle 70 percent of customer complaints for its clients, and it plans to achieve 90 to 95 percent within a year, he said.
“If you’re giving us 100,000 rupees [US$1,136] per month, you are automating the job of at least 15 agents,” said Gupta.
At that price, the service costs roughly the same as three customer-care staff, he said.
LimeChat’s sales soared to US$1.5 million last year from US$79,000 two years earlier, regulatory disclosures show.
Last year, the firm began integrating Microsoft’s Azure language models and algorithms in a partnership to launch a new e-commerce chatbot.
Among Gupta’s clients is Indian alternative medicine products firm Kapiva, which has deployed a LimeChat bot for customer interactions over WhatsApp.
Keying in a prompt — “What kind of diet should I have to reduce weight?” — yielded an AI meal-plan creator. A follow-up query in English and Hindi about how a slimming juice differs from another item was also answered, with the chatbot eventually sharing links to Kapiva products with a smiling emoji.
Kapiva did not respond to Reuters questions.
LimeChat’s rivals include Reliance, the conglomerate chaired by Mukesh Ambani, which acquired Indian start-up Haptik in 2019.
Haptik said it offers “AI agents that deliver human-like customer experiences” that cost US$120 and can cut support costs by 30 percent. Revenue skyrocketed to almost US$18 million last year from less than US$1 million in 2020, disclosures show.
Haptik promoted an online seminar last month by posing the question: “What if you had a full-time employee who never sleeps and costs just 10,000 rupees?”
“We are seeing a huge shift,” Haptik product manager Suji Ravi said. “Brands are not investing in human agents, and they want to deploy AI agents.”
For LimeChat client Mamaearth, an Indian personal-care brand, the main attraction of AI chatbots is scalability, said Vipul Maheshwari, head of product and analytics at parent firm Honasa Consumer.
“Providing good customer support is make or break for us,” he said. “But can we infinitely scale my customer support team? Absolutely not.”
The chatbot used by Mamaearth could go beyond simple assistance such as order tracking, and help users with queries such as recommending the right products during pregnancy or, in some cases, handle an agitated customer, Maheshwari said.
The promise and perils of AI are evident at The Media Ant. The Bengaluru-based advertising agency cut 40 percent of its workforce to about 100 over the past year and vacated space in another building to save on rent, founder Samir Chaudhary said.
The firm eliminated 15 salespeople, replacing them with AI bots that identify leads and send e-mails to prospective customers, Chaudhary said.
A six-member call center was replaced with a voice agent called Neha that speaks in near-flawless, Indian-accented English.
When a Reuters reporter asked Neha about advertising on YouTube, she sought details about the budget and target markets, noted the requirements, and ended the conversation cheerfully: “I will e-mail you the details ... have a great day.”
“Ask her out for a coffee and she will laugh it off,” Chaudhary said.
Yet the race to embrace AI is not always smooth for companies. Take Sweden’s Klarna. Chatbots helped the fintech firm cut thousands of jobs last year, but its CEO told Reuters last month the company is now “trying to course correct” and use the technology to improve products rather than reduce costs.
Chatbots have limitations. While most generic e-commerce-related queries posed by a Reuters reporter were handled well by LimeChat bots, some stumped them.
When LimeChat client Knya’s bot was asked for proof of its claim that 1 million medical professionals trust its products, such as its stethoscopes, it replied: “I am sorry, I don’t have enough information to answer your question.”
Knya did not respond to a request for comment.
Customer surveys show chatbots are still disliked by many.
An EY survey of 1,000 Indian consumers last year found 62 percent made purchases influenced by AI recommendations, compared with 30 percent globally. Yet, “the desire for a human connection remains strong,” EY said, with 78 percent preferring online platforms that provide human support.
However, Gupta said well-trained AI agents could resolve queries faster than humans.
He said many standard bots pass conversations to a human agent when they encounter angry customers.
“You need a very small number of people to just handle negative experiences,” he added.
In the 1990s and 2000s, India’s tech boom fueled rural-to-urban migration. Cities such as Bengaluru became outsourcing hubs as domestic firms, including Tata Consultancy Services, Infosys and Wipro, grew into global juggernauts.
That expansion trickled through to Ameerpet, a Hyderabad neighborhood where university graduates fill classrooms to learn IT skills and earn certifications for tech jobs.
Ameerpet’s training centers traditionally offered courses in Microsoft Office and programming languages such as Java. These centers are increasingly focused on AI training.
Outside one, Quality Thought, a banner featured a robot overlooking a globe with the letters: “AI.”
The center was offering a nine-month course in AI data science and prompt engineering for about US$1,360, more than double the price of a traditional Web-development program.
“Recruiters are asking for students with basic AI skills,” staffer Priyanka Kandulapati said. “We are going to streamline our courses even further to suit the demand.”
In a discussion with start-up founders last month about the pace of change, venture capitalist Vinod Khosla, who cofounded Sun Microsystems, offered a stark view of the future for India.
“All IT services will be replaced in the next five years,” he said. “It’s going to be pretty chaotic.”
China’s supreme objective in a war across the Taiwan Strait is to incorporate Taiwan as a province of the People’s Republic. It follows, therefore, that international recognition of Taiwan’s de jure independence is a consummation that China’s leaders devoutly wish to avoid. By the same token, an American strategy to deny China that objective would complicate Beijing’s calculus and deter large-scale hostilities. For decades, China has cautioned “independence means war.” The opposite is also true: “war means independence.” A comprehensive strategy of denial would guarantee an outcome of de jure independence for Taiwan in the event of Chinese invasion or
A recent Taipei Times editorial (“A targeted bilingual policy,” March 12, page 8) questioned how the Ministry of Education can justify spending NT$151 million (US$4.74 million) when the spotlighted achievements are English speech competitions and campus tours. It is a fair question, but it focuses on the wrong issue. The problem is not last year’s outcomes failing to meet the bilingual education vision; the issue is that the ministry has abandoned the program that originally justified such a large expenditure. In the early years of Bilingual 2030, the ministry’s K-12 Administration promoted the Bilingual Instruction in Select Domains Program (部分領域課程雙語教學實施計畫).
Chinese Nationalist Party (KMT) Chairwoman Cheng Li-wun (鄭麗文) earlier this month said it is necessary for her to meet with Chinese President Xi Jinping (習近平) and it would be a “huge boost” to the party’s local election results in November, but many KMT members have expressed different opinions, indicating a struggle between different groups in the party. Since Cheng was elected as party chairwoman in October last year, she has repeatedly expressed support for increased exchanges with China, saying that it would bring peace and prosperity to Taiwan, and that a meeting with Xi in Beijing takes priority over meeting
Philippine Department of Foreign Affairs spokesman for maritime affairs Rogelio Villanueva on Monday said that Manila’s claims in the South China Sea are backed by international law. Villanueva was responding to a social media post by the Chinese embassy alleging that a former Philippine ambassador in 1990 had written a letter to a German radio operator stating that the Scarborough Shoal (Huangyan Island, 黃岩島) did not fall within Manila’s territory. “Sovereignty is not merely claimed, it is exercised,” Villanueva said. The Philippines won a landmark case at the Permanent Court of Arbitration in 2016 that found China’s sweeping claim of sovereignty in
RSS