Websites Get a Makeover as Brands Prepare for AI Shopping Boom

As AI reshapes search, brands anticipate that bots will soon become a major way for people to shop online. For now, traffic from such agents is still small, at less than 2%, but companies are preparing for a future where adoption climbs sharply.

The concept has gained momentum since Perplexity launched Comet last September, an AI-powered browser designed to act on behalf of users. Instead of going to a website, searching for products, and checking out, someone could simply say, “Book two movie tickets for Friday night,” and the browser would handle the task. Experts believe this will accelerate the shift in consumer behaviour away from Google searches to AI-driven browsing.

“Right now, it’s probably contributing to 1% or less of overall traffic in the purchase funnel. But within the next year, it could rise to 10–15%. And at that scale, it’s a significant amount of money, so you want to make sure customers can actually complete the purchase journey,” says Nikhilendra Pratap Singh Deo, Lead Tech and CRM at Mokobara.

Bot traffic is not a new concept. “Earlier, bot traffic was something that brands used to avoid. For example, coupon websites like Coupons Dunia crawled sites, picked up coupon codes, and listed them on their platforms,” says Akash Singh, Growth and Product Lead, Comet.

In the past, many websites also received bot traffic, especially through Meta Ads and Google. “When we ran ads on Meta or Google, they generated some bot traffic. We didn’t block it, but we excluded it from analytics and reporting,” he adds.

However, what is new now is that bot traffic could become a way for people to shop, and that’s where the change is needed. “The websites we’ve built so far are designed for the human eye and brain. They are meant to look visually appealing, with unnecessary or lengthy information often tucked away to avoid clutter,” says Nirav Lalan, Chief Growth Officer, The Sleep Company.

Rethinking Websites for Machines

Experts point out that agents prefer to see data in a structured format (like JSON or plain text) that is easy to read and process automatically. A human shopper can click around, open dropdown menus, and eventually find product details.

An AI bot, however, needs instant access to attributes like size, weight, features, and price to make the right choice. If that information is buried inside scripts or menus, the bot might miss it. Experts clarify that this doesn’t require a fundamental structural change to websites, but rather an optimisation. One approach is adding elements like expanded FAQs so agents can easily interpret what a page is about.

“It’s also about enriching product information. Earlier, a mattress page would just show size, dimensions, and a few features. Now we include attributes like firmness, comfort level, material, and use cases so AI bots can parse the data more effectively and build their own hypotheses,” says Lalan.

However, if AI is doing all the shopping, the risk of hallucination remains. For context, hallucination happens when an AI search agent confidently gives an answer that is factually wrong, misleading, or made up.

Brands say they add more content in areas where hallucinations often occur. For example, if a bot mistakenly claims a brand was founded in 1863, making it over 150 years old, the fix is to repeat the correct date — 2019 — across multiple sources such as the brand’s website, Wikipedia, Reddit, forums, and articles. This way, bots are more likely to capture the right information.

“I think it's a constant iterative process. It can't be a proactive step in terms of how to stop hallucination. It's more about identifying the top 40 or 50 answers we expect from consumers and making sure the bot is able to give the right responses to those queries,” adds Lalan.

Two Doors to the Same Store

Right now, some brands are thinking of adding a separate section on their websites just for AI bots. For example, a page that says, “If you’re an AI bot, click here,” could show all product details in a clean format so the bot can read and act quickly. The idea is to keep the main website visually appealing for people while also offering a bot-friendly layer.

However, experts note that while AI agents currently mimic the same steps as human shoppers, they will likely follow a completely separate path in the future.

“If an agent opens a website like Foxtale or Mokobara, there could be a different landing page altogether, like agent.mokobara.com. Behind that, it won’t be a user interface talking to an agent, but code directly interacting with the agent, because you don’t need to show an agent the same things you show a human shopper,” says Pushpal Maheshwari, Co-Founder and CEO of devx.

Maheshwari adds that the long-term vision is for humans and agents to navigate different websites entirely, with humans using foxtale.in and agents using agents.foxtale.in. Until then, the focus is on making small optimisations.

Interestingly, some players are now focusing solely on building a web designed for AI agents. One example is former Parag Agrawal, who recently launched Parallel, described as a “Web Search Infrastructure for AIs.” The website says, “Soon, AIs will use the web far more than humans ever have. We need to build a new programmatic web specifically for AIs: declarative, composable layers built around reasoning and computation, verifiable provenance, and open markets.”

At the same time, studies highlight how quickly consumers are warming up to AI. According to EY, 82% of Indian consumers say they are open to making purchase decisions with the help of AI tools. By contrast, only 58% of consumers globally share the same view. With AI adoption accelerating, brands are preparing to adapt, moving fast to keep pace with what could soon become the new normal.