Google puts conversational AI inside YouTube ads as Demand Gen bets on friction removal over landing pages

Google announced on 24 September 2026 that Business Agent for YouTube Ads is now available, enabling viewers to engage with conversational AI alongside video ads with product feeds to get immediate answers without leaving YouTube. Consumers can now visit landing pages with a single tap on image ads on YouTube Shorts and Gmail.

The updates mark a clear bet: that the biggest obstacle to conversion isn't the ad creative but the number of steps between seeing a product and getting an answer about it. By embedding AI into the ad unit itself and collapsing the image-ad journey to a single tap, Google is effectively removing the landing page as a mandatory piece of the funnel.

What Business Agent for YouTube Ads actually does

The US beta is available to selected advertisers running eligible Demand Gen campaigns connected to Merchant Center product feeds. Viewers can engage with conversational AI alongside video ads with product feeds, getting immediate answers about products or brands without leaving their viewing context.

Business Agent was first introduced by Google in January as a shopping assistant in Search, and the YouTube test extends the experience into video advertising. Google's Search version uses Gemini models, Merchant Center information and retailers' websites to respond to shoppers' questions.

The beta starts with eligible Demand Gen campaigns, a $10-a-day minimum and no Shorts placements. The additions are rolling out gradually in the US, with Australia and Canada expected to follow in early 2027.

The format sits in direct competition with Meta's Muse agent, which positions AI between your audience and your brand, and OpenAI's Sponsored Agents, which let users converse with brands after clicking ChatGPT ads. Platforms are moving brands' conversational agents into ads or the immediate path to conversion, turning the chatbot into part of the ad experience.

One-click Shorts and Gmail image ads remove the intermediate step

Consumers can now visit landing pages with a single tap on image ads on YouTube Shorts and Gmail, and on Shorts this applies when advertisers upload 9:16 full-bleed image ads. Previously those placements required an extra interaction before a visit.

Google states that advertisers adding Gmail to their Demand Gen campaigns on average see their image creatives achieve a 40% increase in conversions at the same ROI. That figure comes from Google's own performance data and should be treated with the usual caution, but it is a strong enough signal to justify testing the placement for brands that have been excluding it.

The one-click mechanic is a direct response to what works on Meta and TikTok, where vertical image ads already dominate. Static image ads can now serve inside YouTube Shorts as full-screen placements, and that's a big deal for direct-response advertisers looking to expand reach without rebuilding creatives from scratch.

For brands already producing vertical images for Meta, especially direct-response style ads with hooks, offers and UGC, the shift is low-friction. The creative format is proven, the placement is massive, and the specs are nearly identical to what's already being produced.

What this means for social media marketers

Google is not asking whether conversational AI will be part of the ad experience. It is testing where in the funnel it delivers the most value, and for which products. The answer will vary by vertical, but the implication is the same: the ad unit is becoming the destination.

This changes workflow. If the ad is the place where product questions get answered, you need:

• Product feeds that can actually answer detailed questions. Sparse catalog data will fail the moment a viewer asks something the AI can't extract. Attribute richness, semantic clarity and contextual relevance signals matter more in a conversational system than in a keyword-based one.
• Creative that sets up the conversation, not just the click. A vertical image or video ad that drives someone to ask a question is now more valuable than one that just looks good. The call to action is implicit: "Ask me more."
• Alignment between what the ad shows and what the feed can support. If the creative highlights a feature the product data doesn't include, the AI can't help. The creative and the catalog have to speak the same language.

For media buyers, the beta's $10-a-day minimum and Demand Gen-only eligibility mean this is not a test-it-on-£50 experiment. It is a format for advertisers already running structured, feed-based campaigns with conversion tracking in place. If you are not yet running Demand Gen with product feeds, you cannot access Business Agent yet.

The strategic question is whether this format produces more qualified traffic. Google describes the intended result as more qualified traffic; its beta application provides no performance results to substantiate that promise yet. If a viewer asks three questions about shipping, sizing and returns before clicking through, are they more likely to convert than someone who clicked straight to the landing page? Possibly. Are they also more likely to have already decided not to buy? Also possibly. The data will decide, but the hypothesis is defensible.

The competitive context: everyone is embedding commerce into AI

Google is not alone. OpenAI announced catalog-based product ads inside ChatGPT, allowing retailers to connect product feeds directly to the platform. In 2025, Amazon Ads announced the integration of Sponsored Products campaigns with Alexa+, Amazon's next-generation conversational AI assistant, enabling brands to appear during organic product discovery moments as part of natural shopping conversations on Echo Show devices.

AI players that offer ads include Perplexity (sponsored follow-up questions), Microsoft Copilot (ads inside conversational AI), Google AI Mode (ads inside conversational AI), and Amazon Alexa and Rufus (conversational and agentic ads). Starting December 16, 2025, Meta began using conversations users have with Meta AI to personalize ads across Facebook, Instagram, WhatsApp and Messenger, affecting 1 billion monthly Meta AI users.

The pattern is consistent: platforms are turning conversational AI from a feature into a monetisation layer. The question for advertisers is not whether to participate, but which platforms deliver actual performance and which ones are simply repackaging existing inventory with a chatbot wrapper.

Google's advantage is infrastructure. It already has Merchant Center, product feeds, conversion tracking, and millions of advertisers running structured commerce campaigns. Extending that into conversational AI is an incremental step, not a rebuild. For platforms without that foundation, conversational commerce is harder to execute at scale.

What to do if you run YouTube ads right now

If you are already running Demand Gen campaigns with product feeds and meet the beta eligibility requirements, apply. The format is live, the inventory is material, and being early gives you better learning before saturation.

If you are not yet running Demand Gen with product feeds, that is the prerequisite. Business Agent is not a standalone format. It is an enhancement to an existing campaign type. Get the underlying structure right first: clean product data, conversion tracking, and campaigns that already deliver at an acceptable return.

If you are producing vertical image ads for Meta, test them in Demand Gen Shorts placements with the one-click experience enabled. The lift Google is claiming for Gmail (40% more conversions at the same ROI) is material if it holds. The format is proven elsewhere, the specs are nearly identical, and the incremental cost of testing is low.

And if you are building product feeds, assume they will be interrogated by AI, not just matched by keyword. Every attribute you leave empty is a question your product can't answer. Enrich for detail, clarity and the kind of natural-language questions a real person would ask. That is what the system will try to match against.

The shift Google is making is not about better targeting or smarter bidding. It is about collapsing the path between attention and action. Whether that produces better-qualified traffic or just shifts where the drop-off happens is still an open question. But the format is live, the placement is massive, and the competitive set is moving fast. If you run YouTube ads, this is worth understanding now, not in six months when the learning window has closed.

As platforms continue to embed AI into the ad experience, Google's broader AI workspace tools for Ads and Analytics show this is part of a wider automation play, and YouTube's Made On 2026 event made it clear the platform sees AI-assisted creation and monetisation as the default future, not an experimental add-on.