Do you recall a time when grocery shopping meant visiting different stores for different needs? Vegetables came from the local market, clothes from another shop, footwear from somewhere else, and even puja essentials had their own dedicated stores. Shopping was not just about buying things, it involved travelling across crowded markets and spending hours finding what we needed.
Then came supermarkets, malls and hypermarkets where everything slowly started coming under one roof. A few years later, commerce evolved again with quick-commerce and delivery platforms. Suddenly, groceries, medicines, electronics and even food items for late-night cravings could be ordered within minutes through apps. Shopping transformed from a physical activity into a digital experience inside a smartphone screen.
For years, online shopping followed a familiar rhythm. Consumers opened apps, browsed listings, compared products, checked reviews and eventually made a purchase. But now, that journey may once again be entering a new phase.
AI assistants like Google Gemini and OpenAI’s ChatGPT are changing commerce away from search-led behaviour towards conversational, AI-assisted decision-making. Instead of opening multiple apps and comparing products manually, consumers may soon simply type or say, “Order my weekly groceries under `2,000 with delivery within 20 minutes,” while AI systems will compare platforms, evaluate pricing, delivery windows and preferences before placing the order automatically.
In many ways, commerce is now moving beyond clicks and carts into what the industry is calling ‘agentic commerce’. The transition has already begun. The strong step was when in 2025, Razorpay, along with the NPCI and OpenAI, launched a pilot for AI-driven payments directly inside ChatGPT. The collaboration introduced ‘Agentic Payments,’ allowing users to discover products and complete UPI payments without ever leaving the conversation window.
Globally, companies have already been experimenting with similar systems. Grocery delivery platform Instacart introduced ‘Instant Checkout’ inside ChatGPT, allowing users to plan meals, select groceries and complete purchases directly within conversational prompts. Meanwhile, Amazon has been steadily expanding AI shopping experiences through Rufus and newer AI-powered shopping assistants capable of comparing products, tracking prices and helping users make contextual purchase decisions inside the app.
Google has also been steadily building conversational shopping experiences into its ecosystem through Gemini, Lens, Circle to Search and AI Mode in Search. Consumers can now visually identify products, search conversationally instead of through keywords, and even use Virtual Try On features by uploading their own image. A Google spokesperson says, “Consumer behaviours have shifted dramatically as people now expect seamless, conversational experiences when they search and shop for products. AI is transforming shopping journeys by empowering consumers to search what they see, describe what they are looking for, and even try on outfits virtually.”
The larger ambition, however, goes beyond discovery. Google says its focus is increasingly on reducing ‘shopping grunt work’ while enabling seamless transactions through initiatives like the Universal Commerce Protocol. The company is also experimenting with agentic checkout experiences internationally, where AI systems can complete purchases automatically once certain user-defined conditions are met.
The same transition is visible across other ecosystems too. On Amazon, AI-powered shopping assistant Rufus is increasingly shifting shopping behaviour from search-led discovery towards contextual recommendation. Kishore Thota, Director, Shopping, India & Emerging Markets, Amazon, says, “Traditional search gives you a list. Rufus gives you an answer. The difference is personalisation at the moment of decision, your lifestyle, your price point, your brand preferences, all factored in.”
Instead of forcing consumers to refine searches repeatedly, conversational AI systems are trying to understand layered intent. A consumer asking for best phones for photography under `15,000 with no-cost EMI, is not treated as a keyword query. Instead, it becomes a contextual decision-making problem involving budget, financing, camera preferences, trust, use case and lifestyle.
India has also seen early examples of LLM-led commerce. In December 2025, Zepto introduced an AI agent for Zepto Café, allowing users to place food and beverage orders through conversational prompts instead of navigating the app manually. While these remain early use cases, they illustrate how AI assistants are evolving from recommendation tools into interfaces capable of guiding consumers through the shopping journey.
That contextual layer becomes especially important in India, where shopping decisions are often influenced by multiple variables simultaneously: exchange offers, language preferences, affordability, weather conditions, payment options and delivery convenience. But if AI assistants are becoming the starting point of shopping journeys, the next question becomes how exactly will these systems interact with commerce platforms in real time?
How will agentic commerce actually work?
Unlike traditional commerce platforms that rely heavily on consumers manually browsing products, agentic commerce depends on constant communication between AI systems and commerce infrastructure. For AI assistants to recommend products, compare prices, check delivery timelines and complete transactions automatically, they first need access to structured, machine-readable commerce data through real-time APIs. Unlike humans, AI systems rely on continuously updated inventory, pricing and fulfillment information to function reliably inside conversational interfaces.
Pankaj Srivastava, Founder & CEO – UnoSearch, an AI-first digital marketing agency with specialisation in performance-driven SEO, GEO (Generative Engine Optimisation) and paid-marketing says, “What we keep seeing, across the client categories we track, is that these aren’t one pipeline; they’re three distinct surfaces: discovery, comparison, transaction.”
The first layer—discovery—is where consumers express intent conversationally instead of through structured searches. Instead of typing fragmented keywords, consumers may communicate naturally with AI systems the same way they speak to another person. The second layer is comparison. Here, the AI starts analysing factors like platforms, products, reviews, prices, inventory availability, delivery timelines and user preferences simultaneously. Instead of consumers manually checking multiple apps, the AI could compare them automatically in real time. And finally comes the transaction—the actual completion of the order through payment systems, checkout infrastructure and fulfilment integrations. Pankaj says, “Platforms treating all three as a single funnel will struggle. The ones separating them and optimising each independently are already pulling ahead.”
That separation explains why commerce platforms are increasingly investing in backend infrastructure instead of just front-end user experiences. Rohit Sakunia, Founder of Art-E Mediatech, says, “Blinkit, Zepto, and Swiggy Instamart will need live product APIs that LLMs can query in real time.”
Ritika Taneja - Sr. Vice President- Ecommerce, WPP Media backs this, “Price, stock, delivery promise, platform fees, and substitutions can vary by location, dark store, time of day, and basket.” This makes real-time inventory accuracy significantly more critical in conversational commerce environments where AI systems are expected to make autonomous recommendations instantly.
For AI systems to function reliably inside commerce environments, they need access to continuously updated information instead of static product listings. Without that, conversational shopping begins breaking down very quickly. For example, an AI assistant recommending tomatoes, onions or milk that are already out of stock could instantly damage user trust. Similarly, if pricing changes between recommendation and checkout, consumers may become hesitant to delegate transactions to AI systems entirely.
The complexity becomes even sharper in q-commerce because fulfilment systems are deeply hyperlocal. “Price, stock, delivery promise, platform fees, and substitutions can vary by location, dark store, time of day, and basket,” Taneja notes. This makes real-time inventory accuracy significantly more critical in conversational commerce environments where AI systems are expected to make autonomous recommendations instantly.
Several respondents believe this operational layer may ultimately become more important than the AI layer itself. Summit Kapoor, Brand & Growth Marketing Consultant, says, “Your catalog is the new storefront. Your pricing API is the new checkout. If your inventory isn’t updating in real time, the agent has moved on before your page finished loading.”
Historically, platforms competed through interface design, app engagement and visual merchandising. But in agentic commerce, the systems most compatible with AI agents may become the ones most likely to surface during recommendation.
Vishal Rupani, Co-founder, Sprect.com, explains, “Your app may soon be the back-end. The AI your customer is already talking to becomes the front-end.” This could eventually reshape how commerce platforms position themselves entirely. Instead of functioning only as consumer-facing apps, they may increasingly become infrastructure layers powering transactions behind AI assistants. That interoperability layer is already beginning to emerge globally. Rupani mentions, “Google and Shopify launched an open standard the same month so any AI agent can transact with any merchant without a custom integration for every single store.”
Open standards and interoperability are emerging as critical building blocks for agentic commerce. Rather than remaining dependent on a single platform, commerce platforms may eventually need to integrate across multiple AI ecosystems simultaneously. With integrations emerging across ChatGPT, Gemini and Claude consumers could complete similar shopping journeys regardless of which LLM they choose. For merchants, this means discoverability may depend less on exclusive partnerships and more on ensuring their products, inventory and transaction systems remain accessible across different AI interfaces.
This is partly because consumers themselves are unlikely to remain loyal to a single AI assistant. Different users may prefer different platforms based on familiarity and habit or entirely new AI ecosystems depending on accessibility, devices, subscriptions, integrations, or most importantly, convenience.
This has been tried and tested by a quick commerce platform. Some commerce platforms have already started building the infrastructure for this future. In April 2026, Swiggy became India’s first large-scale commerce platform to adopt the Model Context Protocol (MCP), opening Swiggy Food, Instamart and Dineout to AI agents. Through its Builders Club initiative, the company has received over 2,000 developer applications and made available three MCP servers exposing 35 production-grade tools that enable AI applications to understand user intent, search products, create carts, complete transactions and track orders through conversational interfaces. The platform supports AI ecosystems including ChatGPT, Claude, Gemini, Cursor, VS Code, Windsurf, the OpenAI Agents SDK, Anthropic SDK, LangGraph and Google ADK.
Swiggy’s decision precisely proves how important it is for commerce players to ensure their systems remain visible wherever AI-driven consumer intent exists.
Payments infrastructure also becomes critical in this ecosystem because conversational shopping only becomes viable if transactions themselves can take place seamlessly inside AI interfaces. Manas Mishra, Chief Product Officer, PayU & Wibmo, says, “As this evolution accelerates, we are actively preparing for a future of secure, compliant, and trusted AI-enabled transaction environments.” PayU has already begun integrating AI-led systems across onboarding, fraud detection, payment orchestration and conversational payment experiences.
The larger switch is being driven by changing consumer expectations around ‘real-time decisions, contextual recommendations, and minimal friction across everyday use cases like mobility, utilities, subscriptions, and grocery commerce,’ notes Mishra.
Grocery, in particular, may be one of the earliest categories to adopt agentic commerce because many shopping journeys are repetitive and context-driven. Kumar says AI assistants could simplify tasks such as replenishment, meal planning, smart substitutions and budget-based shopping by allowing consumers to express their intent naturally while the system handles discovery, comparison and purchasing.
Hence, the convenience layer may eventually become one of the strongest accelerators of agentic commerce adoption in India, especially because consumers are already deeply comfortable with facilities like UPI, instant delivery, conversational technology or app-led convenience ecosystems.
Q-Commerce and the Advertising Disruption
One particularly interesting challenge to witness in this transformation is how q-commerce platforms function. For years, platforms like Blinkit, Zepto and Instamart have depended heavily on consumer attention. The more time users spent browsing inside apps, the greater the opportunity for platforms to monetise visibility through sponsored listings, promoted products, homepage takeovers and recommendation placements.
But in an agentic commerce ecosystem, much of that browsing layer may begin disappearing.
Abhinay Bhasin, Executive Vice President – Product & Technology, dentsu India, says, “In my opinion when this happens, I feel commerce platforms fundamentally lose front-end control while gaining a new battleground in the algorithmic layer. Visibility will no longer be driven by app installs, search ranking, or banner ads, but by how well a platform’s catalog, pricing, delivery speed, and reliability are understood and prioritised by AI agents.”
Historically, brands competed for homepage banners, sponsored rankings, influencer visibility, app installs and search placements. But AI assistants do not interact with visual storefronts the way humans do. They process structured information, trust signals and operational reliability instead. Bhasin further says, “The ‘shelf’ shifts from UI to model preference. This weakens direct consumer relationships for players like Blinkit, Zepto, and Swiggy Instamart, as the interface owning the conversation becomes the primary touchpoint.”
Rupani adds, “The current ad formats were designed for human attention. In an agent-led flow, there’s no guarantee those placements even get surfaced.” This creates a difficult challenge for q-commerce platforms because advertising had become a major revenue stream for the industry. Brands like HUL, ITC, Marico, and D2C brands spend significantly here.
The pressure may not remain limited to advertising alone. Several experts believe AI-led comparison systems could eventually squeeze pricing power across q-commerce platforms themselves. Taneja says, “When an AI agent can instantly route a grocery basket to the cheapest provider, the premium platforms charge for convenience gets squeezed.” She believes this could create severe margin pressure for the q-commerce ecosystem as agentic shopping becomes more mainstream.
When we questioned q-commerce platforms about this, we did not receive any response. However, several media experts believe this may lead to the rise of sponsored AI recommendations, recommendation bidding systems, AI-prioritised placements and agent-native advertising formats.
Bhasin mentions, “Advertising or discovery-driven revenue models will be forced to evolve from in-app banners and search ads to ‘agent-native monetisation’ such as sponsored recommendations, dynamic bidding for inclusion in AI-generated shortlists, and even outcome-based pricing tied to conversions initiated by agents.”
In many ways, commerce visibility itself may slowly evolve from: “How do I get consumers to click?” to “How do I become the AI’s preferred recommendation?”
Brijesh Munyal, Co-Founder at Ethinos Digital Marketing, says, “Tomorrow visibility may depend more on fulfilment reliability, pricing consistency, delivery accuracy, reviews, repeat purchase behaviour and overall trust signals because AI systems will optimise for confidence and reliability.”
This creates a temporary opening for challenger and D2C brands that may not traditionally dominate digital shelf space. Unlike app-based marketplaces, where visibility is often shaped by advertising investments, this could allow smaller brands to compete more effectively, at least during the early stages of agentic commerce. Sakunia says, “Strong ratings and clean product data can surface ahead of legacy brands in early AI outputs.” However, other experts also believe large brands are likely to adapt quickly by investing aggressively in AI visibility optimisation.
Srivastava believes, “A brand with 300 reviews, a few editorial mentions, and a clean Wikipedia entry will surface ahead of a better product with weaker documentation.” That means visibility itself may increasingly become tied to semantic authority and machine readability.
Kapoor highlights another point, “Paid presence does not disappear in this shift; it changes format. Traditional splash screens and hero banners give way to sponsored, intent-aligned suggestions in the agent’s response layer.”
But it also fundamentally changes how visibility itself gets built inside AI ecosystems. Taneja explains, “Complete objectivity from LLMs is a myth. Because they are trained on existing market data, their inherent bias naturally favours established brands with larger digital footprints.” She adds that smaller brands may increasingly need newer discoverability strategies, or what she describes as ‘LLM optimisation,’ to remain visible inside AI-led recommendation environments.
Trust and the Accountability Factor
But even if AI systems become capable of comparing products, evaluating prices and completing transactions automatically, one major question still remains: will consumers trust AI enough to let it shop on their behalf?
That question may ultimately determine how quickly agentic commerce might evolve. Unlike content recommendations or AI-generated summaries, commerce carries real-world consequences. If an AI assistant recommends the wrong medicine, confirms unavailable inventory, makes poor substitutions or triggers payment failures, consumer trust can collapse almost instantly. And in q-commerce especially, where expectations are already built around speed, convenience and reliability, even small operational mistakes can have outsized consequences.
Srivastava believes, “The problem isn’t technology; it’s accuracy.” One aspect of that operational complexity, as discussed earlier, is that q-commerce environments are extremely dynamic. Experts say operational consistency itself may become one of the biggest competitive advantages in an agentic commerce ecosystem.
That trust deficit becomes even more important because AI systems are expected to operate at scale. As Kapoor rightly says, “One human getting the wrong item is a bad review. One agent placing that wrong order ten thousand times before anyone notices is a front-page story.”
Completely autonomous commerce may still take time despite rapid technological progress. In the near term, most AI-led commerce systems are expected to continue functioning with human confirmation layers, spending limits, approval checkpoints, transaction verification and override controls.
Achint Setia, CEO of Snapdeal, believes the evolution of agentic commerce will happen in stages rather than through an immediate leap to autonomous shopping. He describes the journey as moving from assistive AI that helps consumers search, discover and make informed decisions, to guided systems that proactively recommend products and surface relevant choices, before eventually reaching autonomous commerce where AI can execute transactions on a user’s behalf. According to Setia, the industry is still largely in the assistive phase, making robust infrastructure, explainable recommendations and consumer confidence essential before greater autonomy becomes viable.
Trust over payments infrastructure becomes especially important in this ecosystem. AI-led transactions involve financial trust, authentication and compliance layers beyond simple product recommendations. Mishra explains, “Growth will depend not just on speed or convenience, but on building confidence across users, merchants, banks, and regulators in AI systems that enable secure and compliant movement of money.” He says payment ecosystems will continue requiring authentication, fraud prevention, governance frameworks, user consent and grievance redressal systems even as conversational commerce environments become more seamless.
Questions around accountability like who becomes responsible if an AI orders the wrong product, who handles disputes when an autonomous agent completes the transaction, and whether accountability sits with the AI provider, the commerce platform or the merchant still remain unresolved. OpenAI has not shared official responses around these queries when approached by IMPACT.
At the same time, consumers themselves may not fully care about the technical architecture powering these transactions. Rupani mentions that the customer does not blame the AI. They blame the app, the delivery platform, or the brand they interacted with. That creates additional pressure on commerce platforms because even if AI assistants become the primary interaction layer, consumer trust may still remain attached to the fulfillment ecosystem underneath.
This also explains why platforms may compete on what some describe as ‘machine trustworthiness.’ Historically, commerce platforms competed through interface design, app installs, advertising visibility, consumer engagement and brand recall. But in an agentic commerce ecosystem, platforms may compete through fulfillment reliability, clean structured data, accurate inventory systems, pricing consistency, API readiness and operational predictability.
The Road Ahead
For now, agentic commerce still remains in its early stages, and consumers may first begin trusting AI systems with smaller, repetitive tasks like grocery refills, household essentials or utility payments before gradually moving towards more autonomous shopping experiences.
Kumar believes AI assistants and retailer platforms are likely to coexist rather than replace one another. While consumers may rely on AI for speed and convenience in routine, repeat purchases, retailer platforms are expected to remain important for browsing, discovery, offers and situations where shoppers want greater control over their choices.
While the technology is evolving rapidly, the industry is still exploring what the next phase of AI-led commerce will eventually look like, from trust and governance frameworks to newer discovery, advertising and transaction models. The next generation of commerce winners may not simply be the platforms with the loudest advertising, the biggest interfaces or the most aggressive discounting strategies. They may instead be the platforms that become the most reliable for machines themselves, as AI assistants increasingly begin shaping how consumers discover, compare and purchase products.
