Meesho’s AI expansion is moving beyond product recommendations. The e-commerce platform says its systems now influence nearly 75 per cent of orders, handle 62 per cent of user queries and support sellers across smaller Indian cities. The figures point to a broader shift in how digital commerce is being organised: artificial intelligence is becoming part of the basic infrastructure through which consumers discover products, sellers manage businesses and platforms control operating costs.
The numbers were shared by Pratik Kumar, Meesho’s head of engineering, in an interaction with Business Standard. Kumar said close to 75 per cent of the platform’s orders come through personalised ranking via Meesho’s Prism system. The platform’s AI customer-support agent, Chorus, handles 62 per cent of user queries, according to the company.
Taken together, these claims describe a marketplace where AI is present at several points in the transaction. It influences what a shopper sees, helps answer questions, anticipates what sellers may need to stock, assists with voice-led purchases and supports operational tasks such as fraud prevention, address prediction and logistics. The significance is not only technical. It concerns who can participate in e-commerce, how much human intervention is required and how digital businesses function across India’s uneven urban geography.
## Meesho AI and the problem of product discovery
Prism is designed to recommend products using signals including demographics, geography, device usage, browsing behaviour, product clicks and listings that users ignore. Kumar said the system is intended to identify preferences even when shoppers arrive without a clearly defined purchase intention.
That distinction matters for a platform whose users may not begin with a precise product search. Instead of relying entirely on a customer to type the right words or know the exact product category, the platform attempts to infer what may be relevant from earlier behaviour and contextual signals. Product discovery therefore becomes a platform-managed process rather than a simple response to a direct search.
For shoppers in smaller cities and towns, this kind of interface can reduce the effort involved in navigating a large catalogue. But it also means that visibility is increasingly shaped by an automated ranking system. A product is not only competing on price or availability; it is also being placed within a recommendation structure that determines what a particular user is likely to see.
The supplied account does not establish how Prism ranks individual sellers, how frequently its models are audited or how the company handles errors in personalisation. Those questions are important because ranking systems can influence demand distribution across a marketplace. The available evidence confirms the scale of AI-led ordering, but not the detailed rules governing the system’s outcomes.
## From seasonal demand to regional signals
Meesho is also using an AI system called TrendPulse to anticipate demand and help sellers prepare inventory. Kumar cited Navratri to explain how demand can vary by region and day, including differences in the colours of clothing shoppers seek because of local customs.
This example illustrates why a nationwide digital marketplace cannot be managed as a single, uniform market. Demand is shaped by local practices, seasonal calendars and regional preferences. A seller who prepares inventory based only on a national trend may miss a short-lived opportunity in a particular location. Conversely, a system that identifies regional signals may help sellers respond more quickly.
The platform’s use of AI is therefore tied to the geography of consumption. The technology is not simply processing a larger volume of transactions; it is attempting to interpret differences between markets that may otherwise be difficult for individual sellers to track. In this model, data becomes a way of connecting dispersed sellers to geographically varied demand.
However, the source material does not provide figures showing whether TrendPulse has improved inventory turnover, reduced unsold stock or increased seller income. It establishes the intended function of the system and the company’s use of regional examples, but not its measured commercial impact. That distinction is essential when assessing the difference between deployment and demonstrated performance.
## The access question is broader than language
Meesho’s Vaani AI agent allows users to discover products, understand product information, add items to carts and complete transactions through voice-led interactions. The company is deploying this capability for customers who may be less comfortable typing search queries or navigating text-heavy interfaces.
Voice commerce can be understood as an access layer for digital retail. It reduces dependence on a keyboard-based search process and may make the platform easier to use for people who find text-heavy interfaces difficult. The supplied material does not identify the languages supported by Vaani or provide user adoption figures, so its precise reach cannot be determined from the report.
Even so, the direction is clear. The platform is treating the customer interface as a problem that can be addressed through AI, rather than expecting every user to adapt to the same text-based shopping process. This places product discovery, product information and checkout within one voice-led interaction.
The same logic is visible on the seller side. AI agents respond to questions about financial journeys, payouts, monetisation and campaign creation. By reducing dependence on human intervention, these systems are intended to help sellers resolve issues more quickly. For small businesses, the value of such a system may lie less in the novelty of the technology than in the time saved while managing routine platform processes.
## Small-city sellers and digital urban economies
Meesho reported approximately 960,000 transacting sellers on its platform, while its annual transacting seller base has grown 87 per cent year on year. The company is using AI to simplify seller onboarding, particularly for sellers in Tier-II and Tier-III cities.
These figures place the technology story within the urban economy. Digital commerce is increasingly connecting businesses outside the largest metropolitan markets to customers across regions. The platform’s role is not limited to providing an online storefront. It also mediates discovery, payments-related journeys, logistics, customer support, promotion and demand forecasting.
That concentration of functions can lower the operational burden for a seller who may not have separate teams for marketing, customer service, inventory planning and platform administration. AI tools can make those activities available through a single commercial system. At the same time, the seller becomes more dependent on the platform’s rules, rankings and automated support channels.
The report does not provide a breakdown of sellers by city, business size, product category or income. It also does not establish whether the growth in the transacting seller base has been evenly distributed across locations. What it does show is that Meesho regards smaller-city seller participation as a significant area for AI-enabled onboarding and support.
## The cost of operating AI at scale
AI adoption creates its own infrastructure problem: the cost of running models across a large user and seller base. Meesho says its in-house Bharat-MLStack has reduced AI inference and model-related costs by around 60 to 70 per cent compared with external providers, according to Kumar.
This claim reveals an important part of platform economics. AI is valuable only if it can operate at a cost compatible with the business. Personalised ranking, voice interactions, automated customer support and forecasting all require computing resources. As these tools are used across more transactions, even small per-interaction costs can become significant.
An in-house technology stack gives Meesho greater control over how these systems are deployed and optimised. It also indicates that the company sees AI capability as an operating requirement rather than an experimental add-on. The available report does not provide the absolute cost base, the methodology behind the claimed reduction or a comparison across specific models. The percentage should therefore be read as a company-reported efficiency measure, not as an independently verified industry benchmark.
## What the evidence confirms—and what it does not
The evidence supplied by Business Standard confirms that Meesho is using AI across product ranking, customer support, demand forecasting, voice-led shopping, fraud prevention, address prediction, logistics, seller assistance and onboarding. It also provides three central operating figures: nearly 75 per cent of orders are linked to Prism’s personalised ranking, Chorus handles 62 per cent of user queries and the platform has approximately 960,000 transacting sellers.
These figures indicate breadth of deployment. They do not, by themselves, prove that AI has improved consumer satisfaction, seller profitability, delivery performance or marketplace fairness. Nor do they show whether automated systems produce different outcomes for sellers or users in different locations.
That gap between adoption and outcome is the central issue to monitor. A platform can automate more decisions without necessarily making those decisions more transparent or more beneficial for every participant. The company’s comments establish how Meesho is using AI and where it expects value to come from. They do not provide independent performance measurements across the platform.
Meesho’s AI strategy nevertheless offers a useful view of the next phase of India’s digital urban economy. The relevant infrastructure is no longer only warehouses, roads, delivery networks or payment rails. It also includes ranking models, voice interfaces, automated support systems and forecasting tools that organise how goods move between dispersed sellers and consumers.
For Tier-II and Tier-III businesses, that infrastructure could determine how easily they enter and operate within online markets. For consumers, it could shape how products are found and purchased. For the platform, it is a way to scale service while controlling costs. The company’s next challenge will be to demonstrate not only that its systems are widely used, but also how reliably, fairly and effectively they work across the diverse markets they are designed to serve.
