For a decade, Indian e-commerce has been built on a simple bargain. Platforms invest in getting shoppers to open an app, and brands pay handsomely to be seen once they do. India’s eRetail advertising reached INR 176 billion in CY25, up 56% year on year and roughly 25% of digital advertising spends, according to Dentsu. That sum rests on one assumption, that a human being is browsing, scrolling and clicking.

AI with the capacity to search, compare, and purchase on one’s behalf challenges this presumption. As per Karan Taurani from Elara Capital, a 15% fall in Blinkit’s grocery ads will affect EBITDA for FY29E of Eternal by roughly 5%. This risk appears to be relatively small; however, the risk does not appear equal. Platforms that thrive on habit, like Blinkit and Instamart, are most exposed to this disruption. Players that thrive on discovery, like Nykaa and Meesho, are relatively less exposed to this risk. The competition is increasingly becoming about delivering quickly, compared to owning the interface for making the decision as we move closer to the festival season.

The thesis begins to play out

The potential disruption by agentic commerce was first flagged by experts in April 2026. Since then, the industry’s major players have begun to stake out positions. Meta launched Muse in the US in September, a personalised agent that can browse, compare and purchase products for a user. Amazon has blocked Muse from shopping on Amazon.com, citing the lack of authorisation, agent identification and customer security concerns. Shopify has gone the other way, partnering with Meta to enable agentic checkout through Shop Pay across Shopify stores.

The split matters for India, which may not take long to follow. According to OpenAI, ChatGPT already has over 100 million weekly users in the country, as of February, a sign of how quickly Indian consumers migrate to a new interface. Taurani believes agentic commerce could follow a similar curve once payments, merchant integration and consumer trust mature.

Momentum in marketing departments can be felt as well. According to industry reports, 48% of marketing operations teams use AI agents in pilot programs, while 19% use at least one agent in production, mostly for scoring, content creation, and workflow automation. Grand View Research says that the AI in marketing market in India was worth $756.4 million in 2023 and may hit $4.38 billion in revenue by 2030 with a CAGR of 28.5%. India’s share in the global market was 4.8% in 2023. The global market for AI in advertising is expected to grow ninefold from 2020 to 2028, from $12 billion to $107 billion.

Not every category is equally vulnerable. Taurani argues that agents will matter most where consumers already know what they want, such as groceries, staples and household essentials, and far less in categories built on discovery, such as beauty, apparel and consumer durables. That makes quick commerce structurally exposed. A shopper could simply instruct an agent to recreate the grocery basket at the lowest price and deliver it within 15 minutes, and no app would need to be opened.

Nykaa and Meesho, having had more visibility on discovery-led beauty and fashion and on non-grocery products, are likely to be impacted much less. The reason becomes evident when looking at the holiday shopping season. The same consumer who replenishes food and cleaning supplies before the holiday season also looks out for skincare and gift products. The current scenario sees the consumer switch between different apps. It would be possible to get an agent involved to divide the journey between the platforms so that the predictable basket gets completed by whichever platform provides the most price advantage and speed of delivery, whereas browsing goes on in the place where the inspiration exists.

Where Blinkit and Instamart are concerned, according to Taurani, the main problem is not that of the reduction in the demand for fulfillment. Rather, it is the deterioration in the consumer touchpoint as well as the first discovery point. The nature of competitive differentiation now lies in the product assortment, fill rates, inventory, pricing, and lead times. Platform with better back-end operations will definitely win out, but a challenger that is pricing driven will be able to capitalize on the opportunity, because the agent can compare SKU and get the cheapest one right away. Zepto, a price driven company, could increase market share.

What happens to the ad rupee

Advertising is where the stakes are highest. Taurani’s analysis assumes 90% of Blinkit’s EBITDA is ad-driven and that grocery and FMCG account for about 60-70% of GMV. Under those assumptions, a 15% fall in grocery ad revenue would shave about 5% off Eternal’s FY29E consolidated EBITDA. Elara rates Eternal a Buy with a target price of INR 400.

Fewer search impressions and less sponsored-product inventory could push some performance budgets toward AI agents and branding spend toward social and video. Yet the economics may not deteriorate sharply. Taurani points out that agents concentrate activity among high-intent consumers. Reaching 50 high-intent consumers instead of 100 could deliver similar purchases if conversion doubles from 10% to 20%, supporting higher CPMs, which would be positive for Affle. E-commerce and q-commerce ad growth may therefore moderate rather than structurally decline.

Rajiv Dingra, Founder and CEO of ReBid, says the currency of commerce advertising will gradually shift from attention to outcomes. An agent may weigh hundreds of products while the consumer sees almost none of them. In that world, he says, the relevant units could become cost per qualified intent, cost per inclusion in the agent’s consideration set, cost per transaction, or a share of the GMV the agent generates. Transaction or outcome-based pricing, he believes, will ultimately prove strongest, because advertisers will ask whether the agent chose their product and whether that choice created incremental business.

As shopping moves from app-led discovery to agent-led transactions, platforms could lose their grip on the customer interface and with it a key monetisation layer. The offset is higher agent-led conversion and potentially greater fulfilment volume.

He also expects a new contest around what he calls “agent share of recommendation”, meaning how often a brand appears in an AI’s consideration set and how often it is finally selected. Brands will have to optimise product data, pricing, availability, reviews, fulfilment and offers for machine decision-making as much as for human eyes.

Dingra sees CPC weakening most in habitual, replenishment-driven commerce. “Advertising will increasingly need to exist inside the decision layer rather than only inside the interface layer,” he says. Where consumers still browse and compare themselves, particularly in high-consideration and discovery-led categories, CPC will remain relevant.

Siddharth Kelkar, Managing Director of India/MENA and Performance Business at AnyMind Group, urges caution about rushing to new metrics. The market, he says, is likely to move closer to outcome-based models for low-consideration repeat purchases, built around a completed transaction, an incremental sale, a new-to-brand customer or a repeat order. In beauty, an agent may only narrow choices or direct a shopper to a platform, so value may attach to qualified intent or a verified downstream action. “The absence of a click does not mean the absence of advertising influence,” Kelkar says, adding that a recommendation has value only if it changes the commercial outcome. The real test is whether the agent brought a new brand into consideration, won a switch from a rival, raised basket value or merely automated a purchase the consumer had already planned.

The deeper strategic question is control. Taurani cautions that as shopping moves from app-led discovery to agent-led transactions, platforms could lose their grip on the customer interface and with it a key monetisation layer. The offset is higher agent-led conversion and potentially greater fulfilment volume. Dingra frames the dilemma as whether a platform stays part of the agent’s decision-making layer or becomes only the fulfilment pipe behind it.

Taurani’s conclusion is that the long-term winner will need best-in-class commerce execution along with proprietary agentic capability that protects customer ownership, without opting out of an increasingly agent-led ecosystem. For marketers, that means the next festive season may be the first in which part of the shelf, and part of the buyer, is no longer human.