#e4mXplains:  Is ChatGPT's Agent the true beginning of the end of Search behaviour & advertising?

A funny thing happened last night. I was in the kitchen contemplating why people ate chia seed puddings even as I made one, while nearby simultaneously in my room, on my laptop, forex was being purchased for an upcoming trip as well as foreign train tickets booked for the same trip. In the best mystery novel tradition, I was at two places at once, witnesses and eccentric detectives, or the lack thereof, notwithstanding.

That’s because I got access to ChatGPT’s Agent, finally, and I was putting it through its paces. Let’s be clear. ChatGPT’s Agent, or at least this early iteration of it, isn’t the Jarvis of Tony Stark’s dreams or even Hal 9000 of Kubrickian nightmare. It doesn’t have a butter smooth transatlantic accent (yet), and it won’t save you from or eject you into the vacuum of space (hopefully), but it will book your tickets for you, buy your forex, or purchase shoes that best fit your needs for them. Slowly, clunkily, and succeeding after multiple tries. But more often than not, it WILL succeed.

And talking about types of succeeding, another fun(ny) thing happened last night. After my forex was purchased and my train tickets booked, my phone remained quiet. I was not bombarded afterwards by ads from innumerable forex providers, ticket sellers, and other merchants of travel and commerce. My purchase journey was non-intrusive, its search minimal if not non-existent, and my participation minimal.

What is ChatGPT Agent anyway?

Think of it less like a chatbot and more like a digital butler who doesn’t need to be micromanaged. Agent can perform multi-step tasks across the web. It can open new tabs, fill forms, click buttons, and navigate websites in real-time. You give it a goal, not a prompt. Yesterday, I asked it to book a train ticket from foreign point A to foreign point B. It did. I asked it to buy USD. It did that too, after a few wobbly detours through payment portals. It’s early, and the wheels are still greased with duct tape and prayer, but the thing works.

What it doesn’t do, yet, is everything. When I asked it to find and purchase a Chromebook from Amazon India, it failed. It couldn’t parse product filters. It couldn’t click the right listings. It got stuck. Amazon’s interface (or maybe Amazon’s architecture) is still Agent-proof. For now. But that limitation only makes the wins more telling. Because when Agent can act, it acts independently, cutting the marketer and the middle-click out of the picture entirely.

Read more on the Agent

The big picture

The digital advertising landscape, for all its dazzling promise, has always had a foundational premise: that users will click on a link, visit a website, and, in doing so, expose themselves to the advertisements that fuel the system. But what happens when that premise is upended? What happens when a new technology arrives that gives users the ability to get what they need without ever leaving the comfort of a single response? That is the game-changing question posed by the rise of ChatGPT-like AI agents, and it's a question that threatens to dampen the high-octane growth of an industry that, until now, seemed unstoppable.

To understand the magnitude of this disruption, let's look at the numbers. According to the Pitch Madison Advertising Report 2025, digital advertising in India is expected to reach a market size of ₹52,992 crore in 2025, growing by 17%. The Dentsu-e4m Digital Advertising Report 2025 projects an even higher figure, predicting the market will hit ₹59,200 crore by the end of the year, a growth rate of 20.2%. Globally, the search advertising market is a colossal beast, valued at over $265 billion in 2025.

Read more on OpenAI

This isn't just a market; it's the engine room of the modern internet. Yet, this engine is powered by an old fuel: our attention. It’s a resource we’ve been giving away freely, often without realizing the true cost. Reports suggest the average person is bombarded with anywhere from 4,000 to 10,000 ads every single day. We’ve become a population numb to the constant noise, a passive audience to an endless slideshow of sponsored content.

This is precisely where the “zero-click” phenomenon, supercharged by AI agents, enters the chat. A recent study found that search results featuring an AI Overview had a 34.5% lower average click-through rate. Think about that for a moment. An AI agent is designed to be a concierge, a personal assistant that synthesizes information for you. It bypasses the dozens of sponsored links and content farm articles, providing a direct, concise answer.

It's a frictionless, ad-free experience that, for the user, feels like a massive upgrade. For the advertiser, however, it’s a direct hit to the revenue stream. Why would a user click on your carefully crafted ad for “best smartwatches” when the AI has already given them a definitive list, complete with pros and cons, from its curated knowledge base?

The search advertising industry's titans are already feeling the tremors. Alphabet, Google's parent company, reported a strong Q2 2025 with Google Search & other revenue hitting $54.19 billion, a 12% increase year-over-year. Microsoft also saw its search and news advertising revenue (excluding traffic acquisition costs) rise by a significant 21% in the same period. However, these figures come with a crucial asterisk. While the top lines are still growing, both companies are actively integrating AI into their core search products, like Google's AI Overviews.

This is a double-edged sword: it boosts user engagement but also creates a “zero-click” environment that could cannibalize future ad revenue. Forecasts for the search advertising market are becoming increasingly nuanced, with analysts noting that while overall ad spend may continue to grow, the traditional, click-based model of search advertising is facing an existential threat. The conversation is shifting from “how to get more clicks” to “how to monetize a conversation,” a challenge that AI agents will bring to the forefront.

That being said…

None of this means the system is perfect. If anything, it’s terrifyingly incomplete. Agent isn’t just reshaping how we interact with the web, it’s raising a host of unanswered questions. What happens when it books the wrong ticket or buys from a scammy vendor? Who’s liable: OpenAI, the user, or the invisible plugin partner under the hood? The trust model hasn’t caught up to the task model.

Then there’s data. The more you use Agent, the more it knows about you. Your spending habits, your browsing preferences, your travel plans, your indecision. This isn’t just a cookie following you around the web. This is a high-context digital entity that builds a behavioural dossier in real time. How that data is stored, protected, or monetized remains vague at best. OpenAI insists privacy is a priority, but like every other major tech platform, it’s bound to the same incentive structure of scale first, regulate later.

Financially, we don’t yet know if this Agent will finally be a free concierge or a freemium trap. Will OpenAI eventually charge per action? Will access be tiered? Will brands be able to sponsor preferred outcomes? If the Agent recommends one hotel over another, is it algorithmic wisdom or quiet commerce? Today’s free tool could become tomorrow’s opaque auction engine.

And finally, there’s the larger cautionary tale. The more we delegate to Agents, the less friction we encounter. But friction wasn’t always the enemy. Friction forced decision-making, comparison, and sometimes reconsideration. When an AI does the thinking and the doing, we lose not just the banner ad, but the nudge to ask: is this what I actually want?

Final thoughts

Of course, the advertising industry isn't just going to roll over. They are already responding by developing new AI-driven strategies that are even more personal and seamlessly integrated. The future of advertising isn't about the banner ad; it's about the sponsored recommendation buried so deep in an AI's response that you don’t even recognize it as an ad. The battle for our attention is evolving, and the new battleground is not the website, but the very mind of the AI agent itself.