Artificial Intelligence is rapidly transforming online shopping. The next major shift will involve AI agents making purchase decisions for consumers. As this era approaches, brands must reconsider how their products are discovered, understood, and recommended.

In this conversation with Jyotirmoy Dutta, Co-founder and CEO of Yarnit, we discuss the rise of agentic commerce, the growing importance of AI discoverability in ecommerce, and how brands can stay ahead by building smarter, AI-ready commerce experiences.

Q1. For readers hearing about Yarnit for the first time, what is the company building, and what problem are you trying to solve for ecommerce brands?

Jyotirmoy Dutta: We’re building the intelligence layer for the next generation of commerce.

For the past two decades, ecommerce has relied on consumers searching websites, reading product pages, and making their own decisions. This is now beginning to change.

Consumers are increasingly relying on AI assistants to research products, compare alternatives, and, in some cases, complete purchases. This represents a significant shift in product discovery.

However, most ecommerce infrastructure was not designed for this change. Product catalogues are often incomplete, content is inconsistent across channels, and much of the information AI systems need to make confident recommendations is unavailable or unusable.

At Yarnit, we help brands address this gap. We use specialised agents to continuously enrich product data, generate contextual content, and make products easier for both humans and AI systems to understand. In today’s landscape, brands are not only competing for customer attention but also for AI confidence.

Q2. You’ve spoken about the rise of “agentic commerce.” Why do you believe AI assistants will fundamentally change how consumers discover and buy products online?

Jyotirmoy Dutta: Every major shift in commerce has removed friction.

Search eliminated the need to browse directories. Mobile technology removed the need to shop from a desk. AI assistants now remove an even greater barrier: the effort required for research.

Instead of spending thirty minutes comparing multiple products, shoppers now describe their needs, and the assistant evaluates options, makes a decision, or narrows the choices. The buying journey is increasingly delegated.

This is the shift we refer to as agentic commerce, and it redefines the purpose of marketing. Previously, marketing focused on influencing people. Now, brands must also communicate clearly with AI systems that make recommendations on behalf of consumers. These systems respond to structured information, context, and trust signals, rather than to traditional advertising or design.

Protocols such as ACP and UCP are early efforts to standardise how an agent completes a purchase once a decision is made. Their development demonstrates the immediacy of this shift.

Q3. What are most retailers and D2C brands getting wrong today when it comes to preparing for this AI-first future?

Jyotirmoy Dutta: Most retailers still believe they are publishing webpages, when in reality, they are publishing data.

A human visitor can infer much from images, layout, and brand cues on a product page. An AI system, however, requires explicit information. Most product pages address basic questions but often omit critical details that influence buying decisions, such as compatibility, intended use, limitations, and comparisons to alternatives.

SEO trained brands to think in keywords. AI is forcing a shift toward thinking in terms of knowledge. The brands that win here aren’t just the ones producing the most content. Going forward, brands will actually have to produce the most trustworthy, complete product intelligence to rank on AI platforms.

Q4. Yarnit says that brands need to become discoverable to AI agents, not just customers. What does that mean in practical terms?

Jyotirmoy Dutta: Most people assume an AI assistant is just searching the internet the way Google does. It’s not. It’s reasoning over information to form a judgment.

If someone asks an assistant to recommend a trekking backpack for a week-long hike, it isn’t looking for the best-designed page. It’s about weighing capacity, durability, weather resistance, comfort, what past buyers actually said, return policy, and value for money, and most catalogues were never built to answer any of that directly.

Becoming discoverable to AI agents means giving your products enough structured depth that a model can confidently say when your product is the right answer, and just as importantly, when it isn’t.

That’s a different job from ranking on a search results page.

Q5. How do CatalogIQ and Creative OS help merchants improve both discoverability and conversions?

Jyotirmoy Dutta: There are really two conversations happening at once in commerce today. One is with the AI system as it tries to understand your product. The other is with the customer deciding whether to buy it.

CatalogIQ is built for the first one. It doesn’t just tidy up existing listings. It continuously and autonomously enriches the catalogue: filling gaps, generating the comparative and contextual content a model needs, keeping product data current as things like pricing, stock, and specs change, without someone manually reworking thousands of SKUs every quarter.

That continuous, self-updating layer is what most catalogue tools in the market don’t do: they’re built for a one-time cleanup, not an ongoing state of readiness.

Creative OS is built for the second: generating high-quality creative and campaign assets at scale, on-brand, fast enough to keep pace with how often things now need to change.

Put simply: CatalogIQ helps AI understand your products. Creative OS helps customers want them. Together they cover both ends of the funnel.

Q6. Do you see AI optimisation becoming as important for brands as SEO and social media marketing once were?

Jyotirmoy Dutta: I don’t think it replaces SEO. I think it becomes the next layer on top of it.

SEO was fundamentally about helping search engines understand your content. AI optimisation is about helping reasoning systems understand your business. Every retailer today has someone thinking about rankings and paid media.

In a few years, every commerce team will also be asking: can an AI assistant confidently recommend our products? That becomes a real discipline, just like SEO did.

Q7. What kind of early results or customer feedback have you seen from businesses already using Yarnit’s solutions?

Jyotirmoy Dutta: Content generation was never really the hard part. Every model today can write a reasonable product description. The hard part is knowing what should change, when, and why. That’s where we’ve put most of our effort.

Data, not intuition, drives every enhancement our platform makes. We’re continuously monitoring signals such as search behaviour, competitor activity, seasonality, campaign performance, and customer questions. Our agents identify what needs to change as it happens, rather than waiting for a team to do a manual review cycle every few months.

Depending on the customer’s governance setup, the system either recommends the fix or executes it autonomously.

In enterprise engagements where a brand has gone through this properly- catalogue enriched, content structured, the feedback loop running continuously- we’ve seen AI-driven discovery and citation of the brand go up by more than 50% in under three months.

That’s not a universal number; it depends heavily on how incomplete the starting catalogue was. But it’s the kind of shift that convinces a brand this isn’t a nice-to-have.

What we hear most from teams using this is that they stop treating AI as a writing assistant and start treating it as a colleague that’s always watching the business and getting better at it.

Q8. How does Yarnit differentiate itself from the growing number of AI tools being built for ecommerce and retail?

Jyotirmoy Dutta: Most of the AI industry is focused on models. We’ve always been focused on enterprise systems.

A large language model is capable, but an enterprise doesn’t run on prompts. It runs on product catalogues, digital assets, merchandising calendars, retailer specifications, and years of accumulated context about how the business actually works.

The hard problem is stitching all of that into a system that consistently makes good decisions, not generating a single good piece of content in isolation.

Two things set us apart there.

First, our agents aren’t point solutions. They’re mapped across the entire consumer journey, from catalogue and discovery through campaign, sales conversation, and post-purchase service. Hence, a decision made at one stage informs the next stage rather than resetting each time.

Second, underneath all of it sits Memora, our contextual intelligence layer, which gives every agent persistent memory of the brand, from its products, its past decisions, its tone, to its business context. That means the system isn’t generating in isolation each time; it’s learning and getting better with every cycle.

Before building the commerce platform, we spent years delivering enterprise AI for large retailers and consumer brands, and that taught us businesses don’t need another content generator. They need AI that understands how the business actually operates, integrates with what already exists, and can be trusted to run real processes autonomously.

Q9. Beyond helping merchants sell more products, what larger opportunity do you see AI creating for the future of ecommerce?

Jyotirmoy Dutta: I think AI has the potential to democratise capability that used to be available only to the largest retailers.

Building a genuinely world-class commerce operation used to require dedicated merchandising teams, SEO specialists, copywriters, marketplace experts. Smaller brands couldn’t compete on that axis.

AI changes that. Now, a brand with fifty products can operate with the same quality of merchandising intelligence as one with fifty thousand. That shifts competition back to product quality and customer experience rather than organisational scale, which I think is a healthier place for the category.

Q10. Five years from now, what does success look like for Yarnit, and how do you envision the shopping experience evolving in an agentic commerce world?

Jyotirmoy Dutta: I don’t think ecommerce websites or physical stores will disappear. What changes is who does the legwork.

Today, consumers spend real time researching, comparing, and reading reviews. Increasingly, they’ll delegate that to AI systems they trust, and step in only to approve the final decision. That changes what it takes for a brand to win: it needs to communicate as clearly to AI as it does to people.

Our ambition for Yarnit is to be the intelligence layer that powers that shift by helping enterprises build commerce experiences that keep learning and keep improving, for humans and AI agents alike.

If, five years from now, brands think about AI discoverability the way they think about SEO today, and Yarnit had a hand in defining that category, that’s success.

Conclusion

As ecommerce continues to evolve, AI is set to play a much larger role in how products are discovered and purchased. For brands, success will increasingly depend not just on attracting customers but also on earning the confidence of AI systems that influence buying decisions.

In this conversation, Jyotirmoy Dutta shares valuable insights into the future of agentic commerce, the growing importance of AI optimisation, and how businesses can prepare for a commerce ecosystem in which intelligence, trust, and structured product information become key competitive advantages.

-Interview By Shivani Solanki