The pattern
India has never adopted technology the way the textbooks said it would.
While the rest of the world got landlines and computers before mobile phones, many Indians began their digital journeys mobile-first. The West built credit card infrastructure for decades before digital payments arrived. In India, UPI became the first mainstream digital payment system. The pattern is consistent: where legacy infrastructure is absent, India doesn't catch up. On the contrary, it arrives somewhere else entirely.
In 2026, we are now at the third inflection point, some would argue the most consequential one yet.
The first wave gave access to connection, through the mobile phone that reached the farmer, the first-generation migrant, and the small business owner who had never had a landline.
The second wave gave access to information and transactions through Google Search, followed by e-commerce, and then the explosion of D2C brands. Suddenly, a consumer in Coimbatore could discover a type of product that previously only existed in Mumbai's malls or Delhi's boutique stores.
The third wave is different in kind, not just degree. AI does not solely offer connection or information. Instead, it offers something that, until now, was only available to those with the right social capital: judgment.
The elder brother you never had
One of the participants in our research, titled 'AI and the Indian Shopper's Mind', a 25-year-old early adopter in Bangalore, described his AI tools as an "elder brother." Rather than a search engine or a catalogue, he saw AI as a figure who had seen more, could be consulted before a decision, and wouldn't judge him.
This revealed something powerful about the role of AI in India, where access to judgment has always been relational - for instance, through the uncle who had already bought the car or the older cousin who had navigated the bank loan. Information in India travels through trust networks, communities, and people who have already been where you are going.
For those inside those networks, this system works well enough, but for those outside such as the first-generation college student, the small business owner entering a new category, and the Tier 2 consumer buying a laptop for the first time, the network has always had gaps.
AI, at its best, fills the gap, acting as the elder brother that everyone can now access. It becomes the person who has always been there and does not require you to know the right people to reach.
This is the promise of AI. It is also, as yet, largely undelivered.
The Leapfrog that is not happening
63% of Indian online shoppers are from Tier 2 and Tier 3 cities. These cities drove roughly half of incremental e-commerce orders in 2025. E-commerce penetration in Tier 2 and beyond sits at around 4% against 14% in metros. This means the growth story of the next decade is not in Delhi, Mumbai or Bangalore, but rather far beyond the metros.
These are, in many cases, first-generation digital buyers. They came to smartphones before they came to laptops. They never learned to type keyword strings into a search bar - the behaviour that an entire generation of metro consumers had to unlearn before they could adopt conversational AI. Instead, they are arriving at the AI era without the baggage of the search era.
However, while the digital leapfrog should be clean, it is not.
When Tier 2 and 3 consumers encounter AI as they arrive at an e-commerce platform or open a shopping assistant, it is, almost without exception, built for someone else. It assumes English literacy, the ability to articulate a need in a typed prompt, and a familiarity with AI interfaces that comes from years of exposure, e.g. corporate mandates, tech-adjacent social circles, the privilege of curiosity over utility.
The tools that could serve this consumer, such as voice search, image search, and vernacular interfaces, exist. Meesho saw a 162% rise in vernacular voice search in 2024, driving 22% higher conversions. The infrastructure is being built. But the dominant AI entry point across most platforms remains a text box, in English, embedded in a product page. While it is the right product for the kind of consumer that platforms already know, it is the wrong product for the consumer platforms most need to reach.
This gap between who AI is built for and who needs it most is one of three frameworks we mapped in the full research.
Preview "AI and the Indian Shopping Mind" here
Visible AI, Invisible AI, and the gap between them
Our research maps a distinction that turns out to be more important than anyone in the industry seems to be acting on: the difference between visible AI and invisible AI.
Visible AI announces itself. It is the chat assistant on the e-commerce platform - Maya on Myntra, the style advisor on Nykaa, the shopping assistant embedded in Amazon's interface, the conversational agents increasingly appearing across D2C brand sites. It spells out: I am AI, and I am here to help you. It requires the consumer to initiate, prompt, and trust the interface enough to engage with it.
Invisible AI, on the other hand, does not announce itself. It is the replenishment reminder that appears before you realise you're running low on something you need. The refund chatbot that resolves a damaged order in two minutes. The recommendation engine that surfaces a product you didn't know to search for. The virtual try-on that a 25-year-old in Lucknow used to choose her eyeglass frame, and described to us as "not an AI tool."
The irony is almost too neat: the consumers who have the most to gain from AI are the ones who engage most with invisible AI and the least with visible AI. They are experiencing the benefits without building a relationship with the technology. The platform earns the loyalty, while AI, as a category, remains foreign.
Meanwhile, visible AI, designed for the already-converted early AI adopter, sits unused by the consumers who most need what it could offer. This is because it speaks in a register - and through an interface - that was never designed for them.
Context Flattened
There is a deeper problem underneath the language and interface barriers. Even where AI is accessible, it tends to flatten context.
Neither does it know whether this is your first laptop or your fifth, nor does it know that you are shopping for a wedding and not a Tuesday. It does not know that the category is unfamiliar to you, and that you need someone to explain not just what the product's specifications are, but what the specifications mean for someone with your lifestyle, budget, and level of prior experience.
For a metro-based early AI adopter who cross-checks across ChatGPT, Gemini, and Perplexity before a considered purchase, this is a manageable limitation. They know how to prompt, interrogate, and treat AI as a tool they operate with confidence.
For a first-generation digital buyer in Lucknow or Coimbatore, the context problem is not just unmanageable. When AI returns a comparison table of two laptops with specifications the buyer cannot evaluate, or offers a shortlist without a framework, it ends up missing entirely the texture of what they actually need to know.
When it comes to AI, the gap is not capability, but comprehension. Vernacular AI is not a translation problem, but rather a cultural fluency problem. A consumer in Coimbatore wants AI tools that think in the register and idiom of how they actually speak when they shop, and understand the question behind the question.
No platform has built that yet.
What this means
For shopping platforms, the implications are structural:
The visible AI investment is real, but it is being made for the wrong consumer first. The consumers who most need judgment - not just search or filters but actual guidance in unfamiliar territory - are being served by AI that was designed for those who need it least. Voice-first, image-first, vernacular-first interfaces don't just enhance accessibility - they are the architecture of the next hundred million users. Platforms that treat them as add-ons will find themselves building for a consumer who is already gone.
Invisible AI, when done well, is also a competitive moat. The quick commerce platforms have understood this instinctively: the AI that works best for the cautious consumer is the AI that never asks for trust because it never asks to be seen. Every platform with routine, habitual, low-consideration moments in its journey has the same opportunity and most are leaving it on the table.
We go deeper on this in a dedicated Quick Commerce Edition - covering habit formation, platform benchmarking across Blinkit, Zepto, and Swiggy Instamart, and what we're calling the category's agentic horizon.
For brands, the question is adjacent but distinct:
If AI is becoming the layer through which consumers discover, evaluate, and decide, then brand visibility in that layer is not a digital marketing question but a brand-building question. The D2C brand that exists richly in the AI's training and retrieval architecture will be recommended, while that which doesn't will be invisible regardless of quality.
More broadly, the consumer who is using AI to navigate a purchase category for the first time is in a formation moment. Their mental model of the category, their sense of what is trustworthy, and their default brand is all being shaped in this window. The brand that earns the AI layer earns this first-time buyer, and in a country where hundreds of millions of first-time buyers are still arriving, that window is open for a very short time.
Therefore…
India has leapfrogged before and it will leapfrog again. The tech of the next leap - voice, image, vernacular, invisible AI embedded in the flow of daily life - is being built right now.
The question, therefore, is not whether the Tier 2 and Tier 3 consumer will adopt AI. Rather, it is whether AI will meet them where they are, or whether they will arrive at a door that is built for someone else.
The elder sibling everyone deserves hasn't been built yet.
That's the gap we set out to map.
If this raises questions about what it means for your category specifically, reply to this email, happy to talk it through.
If you're a platform or brand team wanting a confidential read on your own AI position, consumer perception data, and competitive benchmarking - we offer this directly, ahead of wider release. → Request a preview at hello@plum-insights.com
Want to go deeper? 🎥
Watch the webinar recording | Preview the full report
More on our work → www.plum-insights.com
