What Indian D2C Brands Must Do Before AI Agents Become the Buyer - Anant Bhatt, CTO Paperflite

Here's a number that should make every D2C founder sit up: 4,700%. That's how much AI traffic to retail sites has increased year-over-year according to Adobe. And by 2028, roughly a third of all online sales will flow through AI agents. Not humans shopping with AI assistance - actual AI systems making purchases on behalf of users.

The Indian D2C market is racing towards $300 billion by 2030. But here's the problem: brands are still optimizing for human buyers. Human eyeballs, human scroll patterns, human impulse purchases. Meanwhile, AI-assisted buying has already started, and full autonomous AI purchasing is around the corner. The question isn't whether this shift will happen. It's whether brands will be ready when it does.

Anant Bhatt is the co-founder and CTO of Paperflite, an AI-native revenue enablement platform. Before starting Paperflite, he spent over a decade at Cognizant where he filed multiple patents and was part of the first product initiative within the company. He's been watching this shift from both sides - building AI-powered sales tools while observing how enterprises and SMBs are reacting to the AI wave. His perspective on content, discovery, and the future of buying is refreshingly practical and sometimes uncomfortable.

This conversation covers why AI traffic converts better than human traffic, what content chaos actually looks like, why email is dead, and whether India is actually ahead of the West in adopting agentic commerce.

Key Takeaways: The AI Buyer Is Already Here

The Shift Is Real and Measurable:

• AI traffic to retail sites is up 4,700% year-over-year per Adobe's report
• AI-driven traffic has significantly higher conversion rates because visitors come with clear purchase intent, not just window shopping
• Today an AI model might pick your store, but tomorrow it might decide not to - and nobody knows why

Content Chaos Is the New Problem:

• Creating content used to be hard and expensive. Now it's cheap and often useless
• People generate articles from two lines of thought via ChatGPT, then receivers summarize it back to two lines
• The real challenge shifted from stale content to unverified, hallucinated, AI-generated content flooding every channel
• Technical writers who meticulously calculate every word will survive. Generic content creators won't

India's Unique Advantage:

• India has consistently closed the technology adoption gap - computers were 10-15 years behind, internet was 5 years, mobile phones were 2-3 years, AI is essentially same-day
• Cash on delivery, dark commerce, and 5-minute delivery are innovations that didn't exist in the West first
• India's language diversity is actually an opportunity for AI agents - language barriers vanish when an AI can translate Tamil to Hindi to English seamlessly

Q: You spent a decade at Cognizant before starting Paperflite. What made you stay that long, and what finally made you leave?

Anant Bhatt: Honestly, by nature I like to build things. That's mostly my motivation. At Cognizant I got freedom and that was a very powerful motivator. They would just give me and my team a free hand to do whatever we wanted, and we did a lot through those 10 years. But at the end of the day, as much as I learned, at its core Cognizant was not a product company. It's always going to be a services company. Whatever we built would reach a certain volume and then get stalled because that's not really what they are. The focus shifts.

There was a long time we had this feeling that if we really built something and marketed it properly, we would have the joy of building and really do it right. I'm happy I took that plunge because at the end of the day it's very satisfying when somebody buys your thing. Even now, even if it's a $300 sale, I still feel like - oh yes, that's something I built, these people are using it. Negative feedback, I don't mind. People come and say something's wrong, cool, I'll fix it.

🔥 ChaiNet's Hot Take: In services, you build for someone else's roadmap. In product, you build for your own obsession. The difference between satisfied and fulfilled.

Q: Did your Cognizant experience give you an unfair advantage that first-time founders straight out of college wouldn't have?

Anant Bhatt: Yes. One thing we didn't realize back then was that Cognizant really trained us into selling enterprise. The enterprise language is completely different - they start with security documents, policies, compliance requirements. That never bothered us because we were used to it. Very early in our journey, like one year in, we landed a huge enterprise company. We only realized later that that doesn't happen easily.

Our peers, even the pioneers in the market doing 50-60 million, would proudly come and say "now we've started cracking the enterprise game" and I didn't realize it was that hard. We used to talk to CEOs of multi-million Fortune 500 customers regularly because that's what the Cognizant brand gave us access to. That enterprise language, that comfort level, is something you can only learn through experience.

🔥 ChaiNet's Hot Take: Most startups spend years climbing from SMB to mid-market to enterprise. Anant skipped the line because he already spoke enterprise fluently. Corporate jobs aren't just paychecks - they're language schools.

Q: AI traffic to retail sites is up 4,700%. What does that actually mean on the ground for brands?

Anant Bhatt: The shift is phenomenal. In the Google era, it was slightly harder to capture the intent of why a user came to your site. Mostly you'd be targeted through social media and ads, but when they land on your website, it's very hard to realize what the intent is. Whether we like it or not, at this point we're pouring our souls into AI platforms - these are my pain points, these are my problems, this is what I'm looking for. The amount of intent we're giving AI platforms is huge.

So when an AI platform brings someone to your website, they're not there for window shopping. They've come with an intent to buy. And if you're able to satisfy what they're looking for, the sale will increase. I did check that report and the conversion rates of AI-driven traffic are highly high. It's literally what the SEO game was during the golden age of Google. It's just slightly more black-box because we don't know what decisions the AI makes. Today a model might pick your store, but just like that, tomorrow it might decide not to.

🔥 ChaiNet's Hot Take: SEO optimized for Google. The next game optimizes for AI. Same idea, completely different playbook, and nobody has the rulebook yet.

Q: Are enterprises and SMBs reacting differently to this AI shift?

Anant Bhatt: Completely. Enterprises tend to play it safe. They have appetite for AI but when it hits the corporate security layer, last year it was "no AI." This year it's "guardrail AI" - you can't send stuff to ChatGPT but maybe OpenAI APIs are fine. There are a lot more guardrails now.

SMBs are like mavericks. It's the wild west. Take everything, just make it work, if it doesn't, fine, we'll figure something else out. But there are so many distractions too. Everyone is selling everything and the promises are huge. Everyone says you can fire 90% of your staff and AI will do everything. I don't think that's being delivered at any front right now. The general appetite is there, people want it, they just don't know how to use it. And honestly, ChatGPT kind of spoiled it because there's no ramp-up time. You open it, you type, you're done. Every other AI solution has a ramp-up time. There's no true AI product. AI-enabled products where you configure it for your company, understand how your function works, tailor it - that's where you get the best results.

🔥 ChaiNet's Hot Take: Enterprises fear AI. SMBs adopt everything blindly. Neither approach works. The winners will be the ones who adopt selectively and configure deeply.

Q: Will AI agents eventually replace influencer recommendations?

Anant Bhatt: Yes and no. Yes because it's new tech and the general AI response is better than an average person's recommendation. That's hands down, no doubt. But what's really come out is that we as humans value scarcity. Anything in abundance has no value.

Before 1900, before cameras, you had paintings and those were the most valuable because it was very scarce. When photography came, art didn't die. It transformed into things that couldn't be photographed - modernism, cubism, postmodernism. Same with printers. When printers came, people started creating things you couldn't print. Right now AI suggestions are scarce so we give them value. But once everyone is getting AI recommendations, you'll be starved for a human thing. You'll look for something different. Does it mean AI won't work? No, obviously it will. But eventually we'll go back to a human touch. It might not be influencers on Instagram. It might be something completely different. But the human influence element would still be there.

🔥 ChaiNet's Hot Take: Scarcity creates value. Right now AI recommendations feel special. In two years they'll feel like spam. And we'll crave the human voice again, just in a different form.

Q: You mentioned self-checkout as an example. What happened there?

Anant Bhatt: When I was in New York in 2015-16, every store had self-checkout. At one point there were more self-checkout queues and only one human queue. When I left in 2016-17, they shut it down because people did not want to do self-checkout. They wanted the human touch. It's not like the systems were faulty, they worked quite well. But people wanted to interact with a person. I still think there will always be that need for a human somewhere in the process.

Even with websites - the static website might get replaced by dynamic, personalized, AI-recommended experiences. A lot of those parts will change. But that one component where you want to talk to somebody, you would want to talk to a human.

🔥 ChaiNet's Hot Take: Self-checkout worked perfectly and still failed. Technology adoption isn't about capability - it's about whether humans actually want it replacing the things they enjoy.

Q: Content creation is now cheap thanks to AI. So what's the actual problem?

Anant Bhatt: Creating content has become cheap but also useless. This year I saw the pattern - I have two lines to tell you, but I can't send two lines, so I put it in ChatGPT to generate a big article and send it to you. You don't have time to read it, so you use AI to summarize, which gives you the two lines back. Unnecessary content everywhere.

It's really becoming a thing. I've personally started penalizing people who send ChatGPT-generated content. The worst part now is WhatsApp with Meta's AI. You message your friend "are we meeting at 8?" and they send back an AI-generated reply - "Hi, thanks for asking about our meetup, yes we are indeed meeting at 8 PM." Just say yes, man.

At this point, organizations can't guarantee if the content they have is actually correct. It's not about stale content anymore. It's "my sales guy just generated content, didn't read it, and shared it." That's content chaos. How do we solve for that?

🔥 ChaiNet's Hot Take: We went from "not enough content" to "too much garbage content" in under two years. The new skill isn't creating - it's curating and verifying.

Q: Is email dead?

Anant Bhatt: Literally. Email is dead. I don't meticulously follow it anymore. What's happened is if I want to get your attention, I would Slack you or WhatsApp you. Email has no more guarantees. I'm probably using email for only two use cases now. One is the calendar for meetings. The other is transaction notifications - what happened, what updates came through. Very few people are still using email for actual communication.

🔥 ChaiNet's Hot Take: Email isn't dying slowly. It's already a notification inbox masquerading as a communication tool. The real conversations moved to Slack and WhatsApp years ago.

Q: India versus the West - who's more ready for agentic commerce?

Anant Bhatt: Look at the adoption curves. Computers - India was 10 to 15 years behind. Internet was shorter, maybe 3 to 5 years. Mobile phones were literally two to three years, I never expected them to catch up but they caught up crazy fast. And India cracked e-commerce with a model unique to India - cash on delivery, which doesn't really exist in the US. A country that's typically not that trusting figured out trust mechanisms for online buying.

Now with AI, India has surpassed even the US in user numbers and volume. Major AI players are looking at India as their number one market. That was never there before. Same-day launches, same pricing. The gap is done. And India has something unique - language barriers. You have these mega brands that only exist in Tamil Nadu or only in the North. AI agents could solve that because language is no longer a barrier. I can speak in Hindi and the product listing could be in Bengali, and the agent just handles it.

The other thing I really see is a festival-driven buying pattern. AI knows your preferences, what you bought last Diwali, what you want this time, takes your Instagram activity, and makes a purchase 10 days before the festival. Your dress is just there. I completely see that happening in India.

🔥 ChaiNet's Hot Take: India went from "always 10 years behind" to "sometimes ahead" in one generation. The next leap is AI agents buying your Diwali outfit before you even think about it.

Q: If you were advising a new Indian D2C founder today, what should they focus on - marketing or data infrastructure?

Anant Bhatt: For all the marvelous things AI does, it still starts with someone asking something. That's the trigger. Discovery is still a thing. Google will still be relevant. Social media will still be there. Hence marketing will always be there. The persona might change - instead of attracting humans, you could attract agents. That's completely possible. But marketing will always stay. So yes, focus on marketing. Don't let that go.

On data infrastructure - yes you should, that's going to be your moat. But in the age of AI, data has to be vast, and as a founder without massive funding, building that much data is hard. What you really have to do is find a segment that really loves you and serve that need. Nobody else is doing it, you do it. And I truly think the age of AI will solve the inventory problem for D2C - like cloud did for SaaS. I sell a concept before I buy inventory. That could fundamentally change D2C economics.

🔥 ChaiNet's Hot Take: Marketing finds your customers. Data keeps them. But if you only have budget for one, pick marketing. You can't analyze customers you never acquired.

Q: As a CTO, should founders build proprietary AI or plug into existing models?

Anant Bhatt: It's like cloud all over again. When cloud came out, should you build your own proprietary cloud? No, it's too expensive. The same thing is happening with LLMs. It's consolidating into four or five players. But we've been pushed into this game of picking the most intelligent, most expensive model.

What I see happening is that most needs aren't that complex. If I'm doing data extraction, there are open-source models that run on my laptop, not even a GPU. They work fine. Just like founders say "I'd love an IIT grad as my developer but I can only hire at this rate," the same will happen with AI. We'll pick cheaper, simpler models to do simpler jobs. The super core tech will stay expensive, but the practical stuff will be more like picking the right tool for the job.

🔥 ChaiNet's Hot Take: You don't need GPT-4 for everything. Sometimes a screwdriver works better than a power drill. Match the model to the task, not the hype.

Q: Rapid fire - one thing that's massively overhyped?

Anant Bhatt: Agents that do everything.

🔥 ChaiNet's Hot Take: The "one agent to rule them all" pitch is this year's "one app to replace everything." It didn't happen then. It won't happen now.

Q: One thing that's underhyped that people should pay more attention to?

Anant Bhatt: Use agents to plan your tasks, not to execute everything. Planning - that's the real value.

🔥 ChaiNet's Hot Take: Stop asking AI to do your work. Start asking it to organize your thinking. The ROI on planning is 10x the ROI on execution.

Q: Best piece of advice you got, ignored, and later realized was gold?

Anant Bhatt: Don't ask people what solution they want. Ask them for the problem and then you give them the solution. It's a whole different perspective. I initially started saying "tell me what you want to solve, how you want me to solve it, I'll solve it for you." Completely wrong approach.

🔥 ChaiNet's Hot Take: Customers describe symptoms. Founders diagnose the disease. If you let the patient prescribe the medicine, you'll build the wrong product every time.

Final Thoughts: The Human Touch in an AI World

Anant's take on what survives: "We as humans value scarcity. Anything in abundance has no value. When everyone is giving you AI recommendations, you're going to be starved for a human thing."

The bottom line: The 4,700% increase in AI traffic isn't just a stat for boardroom decks. It's a signal that the way people discover and buy products is fundamentally changing. AI-driven visitors convert better because they come with intent, not curiosity. But the flip side is that your store could be picked by an AI model today and dropped tomorrow with zero explanation. Brands need to start thinking about how they show up not just to human eyes, but to AI systems.

The content problem is real and getting worse. AI made content creation trivially easy, which means everything is drowning in generic, unverified material. The brands that win will be the ones that produce content with real information, real data, and real human perspective. Technical accuracy and authentic voice will be the new differentiators when every competitor can spin up a blog post in 30 seconds.

And maybe the most interesting insight from this conversation is about scarcity. AI will handle discovery, recommendations, and maybe even purchasing. But the moment AI recommendations become ubiquitous, human curation becomes scarce, and scarcity creates value. The self-checkout lesson from New York applies here too: just because technology can replace something doesn't mean people want it replaced.

Q: How can people connect with you and learn more about Paperflite?

Anant Bhatt: You can find me on LinkedIn, I'm always happy to chat about AI, product building, or anything around revenue enablement. And of course check out Paperflite - we're helping sales teams figure out this whole content chaos problem and make sure the right information gets to the right buyer at the right time.

Final words: AI agents will buy your products. AI will write your content. AI will discover your brand. But in a world where everything is generated, curated, and optimized by machines, the human elements - real stories, authentic imperfections, genuine expertise - become the scarcest and most valuable things in commerce. Don't fight the AI wave. Ride it. But never forget that on the other side of every transaction, even one initiated by an agent, there's still a human who wants to feel like someone real made this for them.