Artificial intelligence replacing real people — and no, that is not a metaphor.
In 2023, Indian CEO Suumit Shah made a decision that shook up the business world in a way few people saw coming.
At the helm of Dukaan, one of the standout companies in India’s e-commerce sector, he laid off nearly 90% of his entire staff and put AI-powered chatbots in their place.
This was not some cautious experiment.
It was a radical pivot, done all at once, straight out of Bengaluru — and the shockwaves traveled far.
The reaction was immediate and global, drawing international coverage and sparking a debate that still has not ended: how far does automation go when it comes to replacing human work?
A year later, it is worth asking — what actually happened?
The numbers improved, sure.
But what about the people who got left out of that equation?
That is exactly what we are going to dig into here 👇
Receive the best innovation content in your email.
All the news, tips, trends, and resources you're looking for, delivered to your inbox.
By subscribing to the newsletter, you agree to receive communications from Método Viral. We are committed to always protecting and respecting your privacy.
What Dukaan did — and why everyone was stunned
Dukaan is not just any startup. Founded in 2020, it grew fast within the Indian e-commerce ecosystem, helping small merchants set up digital stores in a simple and affordable way. It did not take long for the platform to gain traction and catch the attention of investors. But it was an operational decision — not a new product or funding round — that put the company’s name in headlines around the world.
Suumit Shah publicly announced on social media that he had let go of most of his team — roughly 90% of the workforce — and replaced them with an artificial intelligence-based chatbot system. The reasoning was straightforward: efficiency and cost reduction. The primary target was the customer support operation, an area that, according to Shah, had response and resolution times well below what he considered acceptable for a fast-growing digital platform.
The impact was swift. Newspapers, tech outlets and business publications around the globe picked up the story and turned it into a symbol of something bigger — the definitive arrival of automation in the labor market at real scale. This was no longer theory. It was no longer futurist chatter at a tech conference. This was a company, with a name and an address, that had actually traded humans for machines in customer support — one of the most sensitive areas of any consumer-facing business.
The question hanging in the air was inevitable: if Dukaan did this, how many other companies are about to do the same?
Suumit Shah shared data that seemed to back up the bet. According to figures released by the company itself, first-response time to customers dropped dramatically — what used to take about two minutes with human agents became virtually instantaneous with the chatbots. On top of that, the average time for full resolution of support tickets plummeted from over two hours to just a few minutes. Monthly operational costs for customer service reportedly saw a significant reduction as well. For the CEO, the numbers spoke for themselves.
But what the numbers did not show was the human side of that equation — people who woke up one day and found out their jobs had been replaced by lines of code and a language model trained to simulate conversations.
The operational results — what the data actually shows
From an operational standpoint, the results Dukaan posted over the first year were undeniably positive when looking at the company’s internal metrics. Efficiency went up, costs went down and service speed improved significantly across several key indicators.
The most eye-catching figure was the transformation in initial response time. What previously took around two minutes with a human agent started happening almost instantly with the AI chatbot. For anyone who has ever sat waiting in an online support queue — and let us be honest, who has not? — that leap is pretty noticeable from a user experience perspective.
Full problem resolution also sped up in an impressive way. According to data released by the company, the average time needed to fully resolve a support ticket dropped from over two hours to just a few minutes. Shah stated that this new speed translated into higher satisfaction scores among the platform’s customers.
Those numbers, taken on their own, are the kind of results any investor or financial analyst would love to see on a spreadsheet. The operation became leaner, faster and cheaper. But is that enough to call the experiment an outright success? That is where the conversation gets more complicated.
Real efficiency or short-term illusion?
There is a layer that rarely shows up in performance reports: the quality of the customer experience over time, especially when cases start falling outside the standard script the chatbot was trained to handle. Artificial intelligence systems are exceptional for predictable scenarios and frequently asked questions — but they still stumble when a situation requires empathy, contextual interpretation or decision-making in unusual circumstances.
In e-commerce, most support interactions involve repetitive questions about delivery timelines, exchanges, returns and how the platform works. For that kind of demand, a well-trained chatbot genuinely delivers superior results in speed and consistency. The problem shows up at the edges — in the atypical cases, the more complex complaints, the frustrated customers who need something beyond an automated response to feel heard.
There is also the trust factor. In e-commerce, especially in emerging markets like India where consumers are still building their relationship with digital shopping, human support carries an emotional weight that goes beyond the technical resolution of a problem. When someone reaches out to a store’s support team, they want to feel like there is actually someone on the other side who is genuinely trying to help. A well-trained chatbot can simulate that in many situations — but not all of them. And when it fails, the negative perception can be bigger than the failure itself, because the customer feels abandoned by a company that chose technology over a real person.
Then there is the question of scalability. What worked for a mid-sized company’s support operation may not work the same way for larger organizations with more complex service volumes and more diverse audiences. Automation is not a one-size-fits-all solution — it is a powerful tool when applied correctly, but it requires constant calibration, human oversight and, often, the presence of specialized people to keep the system running smoothly.
In other words, jobs do not disappear completely — they transform, and that transformation has an adaptation cost that is not always factored in at the time of the announcement. 🤖
The debate Dukaan ignited — and it is far from over
The Dukaan case became a reference point in discussions about the future of work because it was not hypothetical. It was real, it was public and it was fast. And that forced companies, governments and workers to confront it head-on, without the comfortable distance that future scenarios usually provide.
The debate that followed was intense and multifaceted. On one side, advocates of artificial intelligence arguing that automation frees humans from repetitive tasks so they can focus on more creative and strategic work. That line of reasoning points to efficiency gains, a reduction in human error and the potential to reallocate talent to higher-value roles. In this view, AI does not replace people — it complements what they do and amplifies results.
On the other side of the discussion, workers and experts point out that this transition rarely happens in a fair or organized way. The employees laid off from Dukaan did not receive, according to public reports, robust reskilling programs, retraining or any structured support for this new reality. And that is a pattern that repeats itself in many cases of aggressive automation around the world — the bill for corporate efficiency ends up being paid by those least equipped to absorb the impact.
For critics, the Dukaan story represents the most concerning side of the tech revolution: the risk of mass technological unemployment, growing social inequality and the progressive dehumanization of services. When a company decides to swap people for machines purely based on cost-cutting logic, without considering the broader social impact, the outcome can be efficient from a financial perspective — but problematic in virtually every other dimension.
The skills gap is the real crux of the problem
The impact of automation on the global job market is a topic that has been gaining more and more attention on the agendas of international organizations. The World Economic Forum estimated, in a report published in 2023, that automation could displace around 85 million jobs by 2025 while potentially creating about 97 million new roles. The net difference looks positive on paper, but there is one critical detail: those new roles require different skills than the ones most displaced workers currently have.
That skills gap does not resolve itself overnight. Reskilling professionals takes time, investment and educational infrastructure — and the pace at which artificial intelligence advances is considerably faster than the pace at which workforce training systems can adapt. Companies that choose to replace entire teams with AI systems without investing in transition programs for their employees are, in practice, shifting the social cost of that change onto the government and society as a whole.
The future of humans and AI coexisting at work
In the context of e-commerce specifically, the pressure for efficiency is constant and intense — and it is going to stay that way. Tight margins, fierce competition and increasingly demanding consumers create an environment where adopting artificial intelligence is almost inevitable. The question is not whether companies will use AI, but how they will use it.
If the technology is implemented as a complement to human work, enhancing what teams already do and freeing people up for higher-value tasks, the outcome can be genuinely positive for everyone involved. Hybrid models — where AI handles repetitive and predictable demands while humans step in for more complex and sensitive cases — are already being tested by various companies around the world with promising results.
But if the logic is simply mass replacement with no planning and no social responsibility, cases like Dukaan will multiply — and the consequences will become harder and harder to ignore. Will Suumit Shah’s model become the market standard or will it be remembered as an extreme, isolated case? That is a question the coming years will answer, but the most likely scenario points to something in between: companies adopting AI progressively but running into practical and ethical limits that will require maintaining human teams in strategic roles.
What sticks from this whole story
Dukaan’s trajectory in 2023 is, at the same time, a case study on the real gains that automation can deliver and a powerful reminder of what gets lost when technology is treated as an end in itself rather than a means to an end. The chatbots worked — that is a fact. Costs dropped, response speed improved and the operation became leaner.
But e-commerce, at the end of the day, is a business about people buying things from other people. And when that human layer is completely removed from the equation, something subtle but important ceases to exist in the relationship between a brand and its customers.
Artificial intelligence is going to keep advancing — that much is certain. New models, new capabilities, new applications. What remains an open question is how society will handle the pace of this transformation and who will make sure the jobs that disappear do not leave people behind with no prospects. Companies that manage to find that balance — between the efficiency technology offers and a commitment to the people who are part of their operation — are likely to build something more solid and lasting than those that simply go with mass replacement as the easy fix.
In the end, the Dukaan case is not just about an Indian company and its former employees. It is about a choice every organization will need to make in the coming years — and about the kind of future we want to build alongside technology, not in spite of it. 💡
