Dukaan CEO Suumit Shah says the Indian enterprise e-commerce platform used an AI chatbot called Lina to transform customer support. According to Shah, the company laid off 90% of its support team after deploying the assistant, while support costs fell by about 85%.
He also said Lina made first responses instant and brought resolution times down from more than two hours to just over three minutes. The claims offer a striking example of how a company may weigh faster service and lower costs against the role of human support staff.
What Dukaan says Lina changed
Lina handled both general questions and account-specific inquiries, according to the account of Shah’s announcement. That scope suggests the chatbot was used for more than directing customers to basic information: it also responded to questions tied to individual accounts.
Shah described the change in two measures of service speed. First responses became instant, he said, while the time to resolve an issue fell from more than two hours to just over three minutes. The distinction matters: an immediate reply does not necessarily mean a customer’s problem is solved, but Shah’s claim covers both the initial response and the resolution time.
The article describes Lina as presumably based on GPT-4. It does not provide details about how the system was built, what types of account questions it could answer, or how the company handled inquiries the chatbot could not resolve.
Cost savings came with a workforce change
The reported results came alongside a substantial staffing reduction. Shah claimed that Dukaan had laid off 90% of its customer support team after the chatbot was put in place. He also attributed an approximately 85% reduction in customer support costs to the AI system.
Those figures frame the decision as a business trade-off: the company says it achieved quicker service at much lower cost, while sharply reducing the number of people on its support team. The announcement does not explain how many employees were on the team before the cuts or what work remained for the staff who stayed.
Nor does the account provide independent measurements of Lina’s performance. The time and cost figures are Shah’s claims, so readers cannot use the article alone to determine how the chatbot performed across different types of customer problems or whether its results held over time.
A shift from growth toward profitability
Shah placed the decision in the context of startups prioritizing profitability over becoming “unicorns.” He said Dukaan was doing the same, adding that the move was less magical but helped pay the bills. In that framing, automating customer support was presented as a way to address a longstanding operational challenge while controlling costs.
He also described customer support as a problem the company had struggled with for a long time, and said fixing it felt like an opportunity. That gives a sense of why the company pursued the chatbot: the stated goal was not only to lower expenses, but also to address a service function Shah said had been difficult to manage.
What the announcement leaves open
Dukaan’s claims point to potential gains in response speed and operating costs, but they leave important questions unanswered. The source does not say how customer satisfaction changed, whether customers preferred Lina, or how often questions had to be handed off to a human representative.
Those details would help show what the reported improvement meant in practice. A chatbot can answer quickly, but the quality of support also depends on whether it resolves the customer’s issue. For now, the announcement offers Shah’s account of Lina’s speed and cost impact, alongside the reported reduction in support staff.