Voice AI could become the next UPI-like movement for India: Akhil Gupta, NoBroker

A year ago, when Akhil Gupta, Co-founder and CTPO of NoBroker, spoke about the launch of Convogen, the company’s conversational AI platform, the idea was simple but ambitious: transform how enterprises understand and manage customer conversations. Twelve months later, that idea is beginning to evolve into something far bigger. Voice AI, Gupta believes, is not merely another interface, it could become the next major digital shift in India.

“The last one year has been amazing,” Gupta says, reflecting on the journey since Convogen’s launch. “You were there last year and you are here today, so you would have seen the kind of innovations we have brought in. When the NASSCOM data came out recently, it reassured us that we are absolutely on the right path. Voice AI has grown leaps and bounds.”

The rise of voice as the next digital interface

The growing momentum around voice interfaces is no coincidence. As digital services penetrate deeper across India, the nature of user interaction is gradually shifting from screens and text to natural conversation. Gupta points to the larger narrative being discussed across the industry, that voice could become the next foundational layer of India’s digital infrastructure.

“With voice AI, you don’t even need a smartphone,” he explains. “Even a basic ₹1,000 feature phone can access enterprise AI agents that can talk to you, guide you, and solve problems. When payments can already happen through systems like UPI Lite, voice combined with AI becomes extremely powerful.”

The belief in voice as the next frontier is not limited to India. Gupta recalls conversations with engineers from global technology companies at the India AI Summit where the direction seemed unmistakable.

“Everyone believes voice will take over,” he says. “And for a country like India, with so many languages and dialects, voice becomes even more important.”

India’s linguistic diversity presents both a challenge and an opportunity for conversational AI systems. Dialects shift significantly across regions, and even within the same language, expressions and pronunciation vary widely. According to Gupta, building AI systems that understand these nuances is essential for creating meaningful engagement with users.

“Even within Kannada there are multiple ways it is spoken,” he says. “We saw similar patterns when we deployed solutions in the UAE. Customers wanted Arabic in different dialects, Jordanian Arabic, Saudi Arabic, and so on. When the AI speaks in your dialect, the connection with the customer becomes much stronger.”

From internal tool to enterprise platform

For NoBroker, conversational AI was initially not about building a new product for the market. It was about solving an internal challenge. Over the years, the company has evolved from a simple rental marketplace into a broad ecosystem of real estate services. Today, it spans renting, buying, rental agreements, sale deeds, home loans, interiors, cleaning, painting, and relocation services.

Gupta describes the company’s ambition in straightforward terms. “You can think of NoBroker as the Amazon of real estate. Anything related to real estate, we want to provide it for our customers.”

Operating across such a wide range of services inevitably introduces complexity. Managing large teams of sales and support agents while maintaining consistent service quality becomes increasingly difficult. That operational challenge led to the creation of Convogen.

“We built Convogen internally to solve our own problems,” Gupta says. “When you have so many facets of business, you need strong monitoring and analytics to ensure conversations are happening the way they should. Once we saw how powerful the system was, we realised it could become a product in itself.”

Why conversational AI needs to behave like a digital employee

While conversational AI has been discussed for years, Gupta believes the industry is only now entering a phase where the technology truly lives up to its name.

“AI itself is still very new,” he says. “I don’t think anyone can say they have established themselves yet. The pace of innovation is so fast that if you don’t keep up, you can get left behind very quickly.” He also believes many earlier automation attempts failed because they relied on rigid rule-based systems.

“What we saw over the last five to seven years were mostly flow-based chatbots,” Gupta says. “Those were not real conversational AI. They were essentially guided menus where users had to click through options to get answers. Anyone who has used those systems knows how frustrating that experience can be.”

The next generation of conversational systems, he argues, must behave far more like humans, capable of understanding context, remembering interactions, and responding naturally.

“You should not think of conversational AI as a tool,” Gupta says. “You should think of it as a digital employee, an AI employee who is as capable as a human employee.”

To achieve that, the technology must function across multiple communication channels. “It has to be omnichannel,” Gupta explains. “Customers might start a conversation on chat and continue it somewhere else. The AI needs to maintain that context and memory across all those interactions.”

The measurable impact of AI on business outcomes

Within NoBroker itself, Convogen is already playing a significant operational role. The company has integrated the platform deeply into its customer engagement processes, both to automate conversations and to enhance the productivity of human agents.

“About 35 percent of our voice calls are now handled by AI agents,” Gupta reveals. “At the same time, 100 percent of our conversations, whether human or AI, go through our supervisor agents. Our human agents also use AI copilots during conversations.”

The insights generated by conversational analytics are beginning to translate directly into business impact. In one instance within NoBroker’s interiors business, the platform helped identify gaps in how agents were following standard operating procedures.

“Before Convogen, the conversion rate for a certain group of agents was around 16.25 percent,” Gupta says. “When we scored them against our SOPs, their average score was about 3.5 out of 10. After working on those insights for a week, the score improved to about 7.5 and the conversion rate increased to nearly 19.5 percent.”

That improvement translated directly into revenue growth. “That alone added a few crores to our top line,” he says.

AI-driven conversations are also transforming how NoBroker manages large volumes of customer inquiries. With millions of users browsing properties on the platform, manually qualifying every potential lead is simply not feasible.

“In our primary sales business, we get a huge number of inquiries from people looking at properties,” Gupta explains. “It is humanly impossible to speak to every one of them. Voice AI agents now handle a large part of that process.”

According to Gupta, about 12 percent of visits in that segment originate from voice interactions, while nearly 8–9 percent of the revenue is generated through those conversations.

The long road from pilots to scale

Although Convogen emerged from NoBroker’s internal needs, its potential extends far beyond real estate. Organisations from banking to automotive are exploring how conversational AI can transform their customer engagement strategies.

“There are three core problems we are solving,” Gupta explains. “First, how do you empower human agents? That is where AI copilots help. Second, how do you understand what your agents, human or AI, are saying? That is where supervisor agents come in. And third, how do you interact with customers at scale? That is where frontline conversational agents operate.”

Because these challenges exist across sectors, Convogen is designed as a horizontal platform rather than a vertical solution.

“We are not a BPO,” Gupta clarifies. “We are providing the platform. Businesses own their processes and knowledge, and they use our platform to build and manage their AI agents.”

Still, Gupta acknowledges that enterprise adoption will take time. Data readiness remains the biggest hurdle.

“The most important ingredient for any AI system to work is data,” he says. “And every organisation has its own challenges when it comes to preparing and organising that data.”

Many organisations are currently experimenting with proof-of-concept deployments before moving towards full-scale implementation.

“POCs are happening with a lot of customers right now,” Gupta says. “But moving from a POC to deployment, then to production, and finally to scale, that is a long journey. That journey is just beginning.”

Convogen currently supports nine languages and can adapt to different dialects through training data and recordings. But the bigger transformation, Gupta believes, lies in how conversational AI will reshape digital interaction itself.

If the past decade belonged to apps and smartphones, the next one may well belong to conversations. And if that shift unfolds as Gupta expects, the future of digital services may not be typed or tapped, it may simply be spoken.