In 2015, Bernard Marr, a renowned futurist, influencer, and business and tech thought leader, was working at an e-commerce company. Minutes after a major product launch at midnight, the website crashed due to unexpected traffic. Initially assumed to be customer demand, engineers later discovered the surge came from automated bots.
At the time, bots were viewed purely as threats that needed to be blocked. Today, that mindset has shifted dramatically.
Consumers are choosing AI assistants in droves to browse and shop online. McKinsey estimates that by 2030, nearly $5 trillion of global e-commerce gross merchandise value could be driven by AI agents, even before including services such as travel or subscriptions. This signals a structural change in digital commerce.
Agents like Reliance-owned Fynd’s AI Agent Builder or H&M’s Visual Merchandising Agent can upsell and cross-sell products, reduce cart abandonment, and handle routine support tasks such as order tracking, returns, and FAQs. Integrated with existing business systems like CRM, payments, and catalogues, the tools help companies automate workflows while maintaining seamless human handovers for complex issues.
Now, agentic shopping is beginning to challenge the foundational revenue strategies of traditional e-commerce.
Advertising is becoming less effective as AI agents operate with precise, goal-driven intent rather than browsing behaviour influenced by promotions or visual discovery. Instead of responding to emotional triggers or exploratory shopping patterns, these agents follow clearly defined user instructions—directly impacting cross-selling opportunities and significantly reducing impulse purchases.
“In offline settings, shopping has always been conversational. People describe what they need, ask questions, compare options, and seek reassurance before making a purchase. However, this behaviour has not translated easily to digital commerce,” Meesho said while launching its generative AI-powered voice assistant, Vaani, last month.
Falling Numbers
Talking to AIM, Sachin Kumar, Founder, BottleOpeners & ConTech Commerce, notes that agentic commerce has started reflecting on the P&L sheets.
“We are already seeing early signs of AI agents reshaping ad-driven metrics by taking over repetitive, rule-based tasks once objectives are clearly defined. In SEO and GEO, this is already visible—tasks like competitor analysis, keyword clustering, and content optimisation can now be executed in minutes,” he explains.
AI agents are fundamentally shifting behaviour from impulse-driven to decision-driven purchases. Nearly 40% of consumers already use AI for online shopping, according to CapitalOne Shopping Research.
“This indicates a clear move away from browsing-led discovery toward AI-curated decision-making. As a result, traditional levers like impulse buying and cross-selling are weakening,” Kumar states.
This is not a short-term shift, as AI agents reduce the cognitive effort involved in decision-making.
In B2B, this shift is even more pronounced than in B2C. While impulse buying has always been limited, browsing-led discovery, comparison, and vendor evaluation have historically driven the buying journey. Today, 73% of B2B buyers are already using AI tools during purchase research, and nearly 61% of the buying journey is completed before a buyer engages with a vendor, per research by Loganix.
Three Models of Agentic Commerce
AI shopping assistants are now embedded directly into websites, helping prevent cart abandonment, enabling automated product comparisons, and offering conversational search experiences that are gradually replacing traditional keyword-based navigation.
Speaking at MLDS 2026, Dipayan Chakraborty, Head of Indian Analytics Centre, eBay, explained that API-driven transactions represent a deeper level of integration, where AI agents interact directly with retailer systems to complete purchases without users needing to visit websites or apps.
Agentic commerce itself is progressing across six levels of shopping autonomy. It begins with programmed convenience such as subscriptions and auto-replenishment, moves to cognitive assistance where AI gathers and summarises information, and then to guided purchasing where AI builds carts, but humans approve transactions.
More advanced stages include delegated buying, where AI completes purchases within predefined rules; economic optimisation, where it manages budgets and timing for better value; and ultimately network autonomy, where multiple agents continuously negotiate services like insurance, renewals, and recurring needs. As Chakraborty noted, most enterprises today remain within the first three stages, where human oversight still plays a central role.
At the same time, retailers are deploying agents that act on behalf of the business itself—dynamically presenting discounts, personalised recommendations, and contextual offers in real time to improve engagement and drive conversions.
This shift is also expanding beyond metros, shaping the next phase of India’s e-commerce evolution.
Agentic AI Beyond Metros
Sanjeev Barnwal, Founder & CTO, Meesho stated that the company stands at an advantage in implementing agentic AI for value-conscious customers in India’s tier-2,3 markets. “We have a deep understanding of how India shops, and that allows us to anticipate and solve consumer needs in ways that very few platforms can,” he said while launching Vaani.
However, the task is steep. To enable agentic commerce in lower-tier cities, Meesho had to develop a voice agent, since the population there is more comfortable speaking than typing.
Vaani attempts to replicate the experience of interacting with a shopkeeper. Built using a multi-agent AI architecture, the system leverages edge computing for speech understanding and synthesis, enabling low-latency responses while optimising operational costs.
“We believe this is a meaningful step towards making e-commerce more accessible and more human, and towards shaping a future where technology adapts to people, not the other way around,” Barnwal states, adding that users engaging with the assistant show a 22% higher conversion rate.
Besides, Meesho, Fynd, an AI-native unified commerce platform which operates extensively in tier-2,3 cities, has an AI Agent Builder designed to help online businesses automate sales, customer support, and shopping experiences through intelligent conversational agents built specifically for commerce.
“We’ve built a strong AI-led portfolio with products such as Kaily, Pixelbin, Boltic, and GlamAR. This evolution is further strengthened by capabilities like AI-powered PIM and AI fashion design (Fynd Create),” founder Sreeraman Mohan Girija earlier told AIM.
The platform allows brands to deploy AI-powered assistants that act like virtual salespeople, stylists, and support executives—guiding customers through product discovery, recommending personalised items, answering queries instantly, and even enabling in-chat checkout.
Such AI agents can upsell and cross-sell products, reduce cart abandonment, and handle routine support tasks such as order tracking, returns, and FAQs, all while operating 24/7 in more than 50 languages. As agentic commerce grows, retailers must build machine-readable infrastructure with structured catalogues, APIs, and AI-ready payment systems while transforming websites into engaging discovery platforms.
Your reaction