The Unicommerce ‘Catalyst’: How Bilingual AI Voice Agents Are Ending Cart Abandonment

Imagine the situation. After browsing for fifteen minutes, adding a pair of shoes to their cart, and entering their delivery address, a Jaipur shopper stopped. Perhaps the delivery schedule was unclear. Perhaps they wanted to inquire about returns, but no one was available. Perhaps they were just anxious about making an online payment and thought COD was a safer option, but they were also uncertain about that. The tab is closed. The sale disappears. And until the numbers arrive the following morning, no one in a warehouse is aware that it took place.

This is how millions of transactions occur simultaneously in Indian e-commerce on a daily basis. Since the beginning of online shopping, the global cart abandonment rate has increased steadily; it was approximately 59% in 2006 and is currently over 70% in 2025. That’s not a passing fad. It’s a structural aspect of online shopping, and for the brands looking at those figures, it represents a huge and obstinately persistent revenue gap that WhatsApp nudges and emails have never quite been able to close. What really works when text doesn’t is the question that Convertway by Unicommerce has been attempting to address.

Their response is a bilingual AI voice agent called Catalyst, which was introduced in January 2026. The idea is simple enough to sum up in a sentence, but it can be challenging to implement effectively: when a customer abandons their checkout, the system immediately initiates an automated outbound call, initiates a real-time conversation in either Hindi or English (automatically switching between them based on the customer’s response), and resolves whatever issue caused the hesitation. When will it be delivered? Responded. Perplexed by a discount? described. Concerned about COD versus prepayment? Within minutes of the drop-off, the customer will receive direct communication in their preferred language. There is no human on the other end. Don’t wait. Just a voice that sounds remarkably human by most accounts.

It’s important to remember that this is not an audio-only chatbot. The difference is important. The majority of follow-up tools used in Indian e-commerce are one-way; a push notification is swiped away, an SMS goes off, and a WhatsApp message remains unread. They are all unable to react to what the customer actually says. catalyst is able to. In real time, it listens, interprets the response, and makes adjustments. The ability to have that trust conversation right away, before the customer’s uncertainty has had time to calcify, is a significant operational shift for a nation where a large percentage of online orders still flow through cash-on-delivery, a payment method that tends to generate higher return-to-origin rates, slower cash flow, and compounding logistics costs.

Over the past year, Unicommerce’s CEO and Managing Director, Kapil Makhija, has positioned the company’s overall direction as an AI-first move. That framing is appropriate for the Catalyst launch. At the time of the announcement, he stated that as India’s digital commerce market moves past its early-adopter stage, real-time communication has become crucial to conversion. That observation appears to be accurate, and international platforms have been addressing it in different ways. Amazon has invested a lot of money in tools for conversational shopping. With AI-assisted upsell and recovery features, Shopify’s ecosystem has exploded. Building for Hindi-English code-switching, which is how tens of millions of Indian consumers actually communicate, rather than treating one language as primary and the other as a backup, is what sets the Unicommerce approach apart.

In this market, Unicommerce is a major player. Serving more than 7,500 customers, including FabIndia, Lenskart, boAt, Mamaearth, and Healthkart, the company operates throughout India, Southeast Asia, and the Middle East. In Q3 of FY25, its Uniware platform surpassed an annualized run rate of more than one billion order items. Because of its scale, Catalyst is being implemented throughout a live, high-volume commerce infrastructure where the stakes are real and the feedback loop is instantaneous, rather than being tested in a lab.

The extent to which the stated conversion increases—up to twice the typical rate, according to the company—hold true across various product categories and consumer demographics is still unknown. Fast-moving consumer goods might react differently than luxury goods. Urban millennials may drop off differently than older consumers who are less accustomed to automated calls. These are the factors that typically yield more complex outcomes than what launch announcements imply. Another issue is saturation: if every brand calls consumers who have abandoned their carts within sixty seconds, will the novelty wear off and consumers just learn to ignore the calls? That’s a serious risk, and the outcome will likely depend more on how beneficial each conversation is than how fast it happens.
As this develops throughout the Indian D2C market, the bilingual component has a subtle but significant impact. It serves as a reminder that a market’s linguistic architecture determines what technology must do to function there. Only a small percentage of Indian online shoppers would be reached by a cart recovery tool designed exclusively for English speakers. Although it’s more difficult, it’s the only version that can function on a real scale when it comes to building for how people actually speak—fluently, across two languages, and occasionally within a single sentence. The fact that Catalyst was created with that reality in mind reveals something about the direction and speed of Indian commerce technology.