By Google has unveiled a wave of AI-driven advertising tools for the Indian market, centred on a feature that embeds a conversational Gemini agent directly inside a Search ad. The rollout, announced at Google Marketing Live 2026 on July 9, pushes Google’s ad business further into agentic territory and raises a central question: how far can an ad go before it starts sounding like advice? It also arrives alongside a separate requirement that advertisers disclose AI-generated ad content.
An AI agent living inside the ad itself
The centrepiece is Business Agent for Leads, now in beta in India. It places a Gemini-powered chatbot trained specifically on an advertiser’s own website inside a Search ad unit. Instead of clicking through to a landing page, a user can chat with the agent about the brand’s products or services right there in the search results, and the interaction is designed to convert into a qualified sales lead. Google says ed-tech platform upGrad is already piloting the tool to handle round-the-clock student queries, freeing its counsellors to focus on higher-value conversations.
Google frames it as capturing intent at the earliest possible moment: It cites an Ipsos survey it commissioned, which found that Google and YouTube are present in roughly 93% of consumer journeys in India at the moment someone discovers a brand, product, or retailer.
The zero-click problem, now inside the ad
The launch lands amid a broader industry anxiety over “zero-click” search, where AI-generated answers satisfy a user’s query without sending any traffic onward. MediaNama’s report points to Info Edge’s Shiksha, an education platform that had to rework its business model after Google’s own AI answers began cutting into the referral traffic it depended on. In this story, the same concern now moves from search results into paid ads.
Business Agent for Leads effectively extends that dynamic into paid search: if a user’s question gets fully answered by the in-ad chatbot, there’s no click to the advertiser’s website at all. A trade-off advertiser may accept if it means better-qualified leads, but one that changes what a “click” or a “visit” even means for measurement purposes.
A parallel move: disclosing when ads are AI-made
In a related announcement a day earlier, Google said it will start requiring advertisers to disclose when generative AI was used to create or edit an ad. The disclosure will appear in a new “How this ad was made” panel in My Ad Centre, across Search, YouTube and Discover.
The mechanism differs by origin: ads made with Google’s own AI tools get labelled automatically, while ads built with outside AI tools rely on the advertiser to flag that themselves when submitting the ad. Depending on local rules, a visible label may also appear on the ad itself. Google says this sits atop existing safeguards like SynthID watermarking and its long-standing disclosure rules for election ads, but enforcement of third-party AI tools still largely depends on advertiser self-reporting, and Google hasn’t specified a rollout timeline or how “edited with AI” will be defined technically.
Who’s liable when an ad sounds like advice?
The more interesting tension is regulatory. Nikhil Pahwa, founder of MediaNama, has argued that conventional search ads respond to explicit commercial queries, while a conversational agent can just as easily be asked something that sounds like a request for advice rather than a product. That distinction matters in India, where regulators like the RBI and SEBI draw a hard line between advertisement and advice and where, historically, unpaid professional advice via SEO was treated differently from paid third-party endorsement.
Pahwa’s two questions frame the emerging problem neatly: at what point does an AI agent’s suggestion become a recommendation, and who is liable if that recommendation steers someone toward a bad product? Those questions get sharper once the “advice” comes from a chatbot in a paid ad on a platform in a regulated category like loans, health or trading.
Google is already fighting a related legal battle on ad liability in India. The Delhi High Court ruled in May that Google infringed on Hindware’s trademark by letting rival advertisers bid on “Hindware” as a keyword; Google has since appealed, arguing the ruling makes India a global outlier and warning of broader consequences for competition and consumer choice.
Why Google is leaning in harder on ads
The financial logic is straightforward. Alphabet reported $109.89 billion in revenue for Q1 2026, with $77.25 billion of it — about 75% coming from advertising, including $9.88 billion from YouTube alone. That ad revenue is what funds Google’s AI infrastructure spending, which makes deeper Gemini integration into the ad stack less a side experiment and more a core business priority.
Taken together, these moves signal where Google is headed: ads that don’t just target intent but actively engage it, and a parallel push to keep that engagement visibly labelled as AI. Whether India’s regulators are ready to draw the line between an ad and an answer is the question that will decide how far this goes.
