Google’s VP of Product explains the technical mechanisms behind AI search recommendations and provides actionable strategies for businesses seeking visibility.
Robby Stein, VP of Product for Google Search
Robby Stein stood in front of a camera on October 30, 2025, ready to explain how the world’s largest search engine now decides which businesses appear in AI-powered recommendations. As VP of Product for Google Search, Stein leads the team responsible for ranking mechanisms that determine whether a restaurant in Los Altos gets listed for lunch queries or whether a pet groomer receives automated phone calls from…
Google’s VP of Product explains the technical mechanisms behind AI search recommendations and provides actionable strategies for businesses seeking visibility.
Robby Stein, VP of Product for Google Search
Robby Stein stood in front of a camera on October 30, 2025, ready to explain how the world’s largest search engine now decides which businesses appear in AI-powered recommendations. As VP of Product for Google Search, Stein leads the team responsible for ranking mechanisms that determine whether a restaurant in Los Altos gets listed for lunch queries or whether a pet groomer receives automated phone calls from Google’s systems.
The interview revealed technical details about query processing, recommendation algorithms, and visibility strategies that marketing professionals had been speculating about for months. Stein’s explanations provided the first comprehensive look at how artificial intelligence reshapes business discovery on Google.
Subscribe PPC Land newsletter ✉️ for similar stories like this one
Technical architecture behind AI recommendations
Google’s AI Mode employs what Stein described as a “query fanout technique” to process user requests. The system receives a question from a user and immediately breaks it into dozens of related queries. “It’s literally using Google search as a tool like doing googling under the hood,” Stein explained during the interview.
This approach differs fundamentally from traditional search algorithms. Rather than ranking individual web pages based on relevance signals, the AI reasoning model executes multiple simultaneous searches across Google’s information systems. For restaurant queries, the system taps into databases containing information about 250 million real-world places.
When a user requests Italian restaurant recommendations with specific parameters like “fun” and “date night,” the AI generates related queries such as “great experience” and “great for date nights.” The system then searches reviews and available information to produce a curated list of recommendations.
The infrastructure supporting these capabilities includes Google’s product graph containing 50 billion items and location databases with continuously updated business information. Many establishments have claimed their Google Business listings and provided menu details, operating hours, and other data that becomes eligible for AI processing.
Google’s systematic AI Mode rollout reached over 200 countries and territories by October 2025, processing queries typically three times longer than traditional searches. The conversational interface allows follow-up questions within the same session, eliminating the need for users to formulate separate searches.
Visual search drives adoption patterns
Stein revealed during the interview that visual searching on Google increased 65% year-over-year. This surge represents one of the most significant behavioral shifts since Google’s founding. Users photograph objects, documents, and environments to initiate searches combining camera input with conversational AI.
The 65% visual search growth reflects changing user expectations about information access. Shopping queries dominate multimodal usage, with people photographing clothing to “shop the look” or capturing product images to find similar items. Educational applications include students photographing geometry problems for step-by-step assistance.
Circle to Search technology, available on over 300 million Android devices, allows users to capture screenshots or circle objects on their screens to generate instant AI responses. Younger demographics naturally transition between photo, voice, and text inputs without conscious effort about modality switching.
The demonstration during the interview showed how users can photograph a cream container and request similar products. The AI analyzes the image and returns comparable items from multiple retailers, displaying stock status and direct shopping links. The system processes visual queries through the same reasoning architecture that handles text inputs, generating related searches based on image content.
Buy ads on PPC Land. PPC Land has standard and native ad formats via major DSPs and ad platforms like Google Ads. Via an auction CPM, you can reach industry professionals.
Automated calling transforms local business discovery
Perhaps the most striking capability demonstrated involved Google’s AI making phone calls to local businesses on behalf of users. During the interview, Stein showed how the system handles requests like “find and book a pet grooming appointment for an extra small dog under one year old.”
The AI processes the request, identifies relevant local groomers, and initiates phone calls to multiple establishments. Within 10 minutes of the demonstration, Google had contacted three grooming businesses and returned pricing information ranging from $74 to $105, along with availability details.
The automated calling feature launched July 16, 2025, affecting businesses throughout the United States except Indiana, Louisiana, Minnesota, Montana, and Nebraska. The system operates across three primary functions: appointment scheduling, restaurant wait time inquiries, and service pricing confirmation.
Business participation occurs through Google Business Profile settings with default opt-in status. Companies must proactively contact Google or modify profile settings to avoid receiving automated calls from the system. Google positions the service as available at no charge to encourage participation.
The capability addresses offline businesses lacking easy web access. Many local establishments operate without sophisticated online booking systems, relying instead on phone communication. Google’s AI acts as an intermediary, gathering information that previously required manual calling by consumers.
Advertising integration remains experimental
Stein addressed questions about whether businesses could pay to receive AI recommendations. “We don’t think that there should be any barrier to people finding information,” he stated. The AI Mode system currently processes queries without incorporating paid advertising signals.
However, Google has begun experimenting with ad formats within AI experiences. “We started some experiments on ads within AI mode and within Google AI experiences,” Stein confirmed. The company maintains focus on building consumer products first while exploring how advertising might integrate with conversational search interfaces.
Traditional Google Ads continue operating separately from AI recommendations. When asked whether ads would disappear, Stein indicated that search usage patterns are expanding rather than changing. Users still need insurance quotes, tax filing assistance, and local business information that advertising can address.
The company envisions future ad formats tailored to complex queries. Users providing detailed parameters about home remodeling projects might receive relevant contractor recommendations or service deals matching their specific constraints and price ranges. These scenarios remain theoretical as Google finalizes how ads might appear in AI systems.
Traffic analysis shows that while some metrics indicate reduced click-through rates when AI Overviews appear, Google executives maintain that organic click volume remains relatively stable year-over-year. The company defines quality clicks as those where users demonstrate genuine interest by not immediately returning to search results.
Visibility strategies for AI search
When asked how businesses can optimize for AI recommendations, Stein drew parallels to traditional search optimization. “Interestingly the AI thinks a lot like a person would in terms of the kinds of questions it issues,” he explained.
Businesses mentioned in top business lists or featured in public articles become discoverable to AI systems searching for recommendations. The reasoning models issue Google searches as tools, meaning traditional search engine optimization principles remain relevant. Websites providing helpful, clear information for specific topics position themselves to appear when AI systems conduct research.
“So now you’re investing in PR not for people to see it, but for AI,” the interviewer noted. Stein acknowledged this framework as accurate, explaining that AI models work by issuing Google searches as tool calls. Optimizing website content for search queries means optimizing for AI discovery.
The importance of public relations visibility extends beyond human readership to algorithmic consumption. Articles, press coverage, and online mentions create signals that AI systems incorporate when evaluating businesses for recommendations.
Reviews play a complex role in recommendation algorithms. Stein compared the process to how a person might evaluate information when making decisions. “Imagine something is scanning for information and trying to find things that are helpful,” he suggested. Helpful reviews could contribute to visibility, though no single factor determines inclusion.
The kinds of questions people ask AI are increasingly complicated and differ from simple keyword searches. Users submit detailed, multi-sentence queries about purchase decisions, how-to instructions for complex tasks, and advice about life situations. Content creators addressing these areas should study AI use case patterns to understand emerging opportunities.
Search behavior transformation metrics
AI Mode users ask questions nearly three times longer than traditional searches, according to data shared during the interview. This behavioral shift reflects the conversational nature of the interface, which accommodates multi-part questions that previously required multiple separate searches.
The global expansion brought AI Mode to more than 35 languages on October 7, 2025. The custom Gemini model designed specifically for Search enables comprehension of local language subtleties rather than simple translation of English results.
Google Trends provides real-time information about search patterns that businesses can leverage for optimization. The tool reveals exactly what users are searching for, with keyword value data that remains underutilized according to Stein. The Google Ads platform also offers traffic estimates for various queries, providing insight into potential reach.
Search Console and related tools provide data about search behavior that becomes increasingly valuable as users shift toward multimodal queries. Voice conversations, image uploads, and text combinations create search patterns that differ from historical keyboard-based queries.
The transformation affects how businesses approach visibility strategies. Traditional approaches focused on ranking for specific keywords prove insufficient when AI systems synthesize information from multiple sources to answer complex questions. Companies must ensure their information appears in contexts where AI systems conduct research.
The shopping demonstration during the interview showed how AI Mode connects to Google’s product graph. Users can photograph items and request similar products, with results displaying real-time inventory status, pricing, and direct purchase links.
Google maintains 50 billion products in its shopping graph with live updates occurring two billion times per hour. Merchants worldwide update inventory status, pricing, and availability information that flows directly into AI-powered shopping recommendations.
When Stein demonstrated the feature using a photographed cream container, the system returned products from retailers like Sephora with in-stock indicators. Users can tap products to view details or proceed directly to merchant sites for purchases.
The capability extends beyond simple product matching to understanding ingredients, features, and comparable alternatives. Users can specify requirements like “similar ingredients” to refine results beyond visual similarity.
Implications for marketing professionals
The interview clarified several aspects of AI search that impact marketing strategies. First, traditional SEO practices remain foundational since AI systems use Google searches as tools. Websites optimized for relevant queries position themselves for AI discovery.
Second, public relations and media coverage gain importance beyond human audiences. AI systems scanning for authoritative information incorporate press mentions, industry recognition, and editorial coverage into recommendation algorithms.
Third, businesses must consider how AI interprets their information. Clear, structured content about offerings, capabilities, and unique value propositions helps AI systems understand and accurately represent businesses in recommendations.
Fourth, the shift toward multimodal queries means businesses should optimize for visual discovery alongside text. Product images, location photos, and visual content become searchable elements that drive recommendations.
Research indicates that 24.3% of marketers receive consistent referral traffic from AI tools, while 39.3% report occasional traffic. This 63.6% combined rate suggests widespread integration between AI search platforms and traditional websites.
Google maintains that clicks from AI-enhanced search results demonstrate higher quality, with users spending more time on destination sites. The company argues that while some queries generate fewer clicks when AI provides immediate answers, increased overall query volume compensates for reduced clicks on individual searches.
The interview concluded with Stein emphasizing that businesses should focus on the same fundamentals that have always mattered: creating valuable content, building authority in their domains, and ensuring accurate, accessible information about their offerings. AI systems evaluate the same factors when determining which businesses to recommend.
Subscribe PPC Land newsletter ✉️ for similar stories like this one
Timeline
- March 2025: Google launches AI Mode in Search Labs, introducing conversational interface processing queries twice the length of traditional searches
- June 24, 2025: AI Mode expands to India with multimodal search capabilities including voice, image, and text inputs
- July 2, 2025: Google extends AI Mode to Workspace accounts across United States, reaching millions of business users
- July 9, 2025: Circle to Search gains AI Mode integration across 300 million Android devices worldwide
- July 16, 2025: Google launches automated calling feature enabling Search to contact local businesses for pricing and availability information
- July 24, 2025: Web Guide experiment launches using AI to organize search results by grouping pages according to query aspects
- July 29, 2025: Search Live with video input announced, enabling real-time conversations while pointing camera at objects or documents
- August 30, 2025: Google addresses content quality amid AI-driven changes at WordCamp US 2025
- October 7, 2025: AI Mode reaches over 200 countries and territories with support for more than 35 languages
- October 10, 2025: Google’s head of search discusses transformation in Wall Street Journal interview
- October 30, 2025: Robby Stein explains AI recommendation mechanisms in Silicon Valley Girl interview
Subscribe PPC Land newsletter ✉️ for similar stories like this one
Summary
Who: Robby Stein, VP of Product for Google Search, leads the team responsible for ranking algorithms and AI-powered search features affecting billions of users globally. The interview targeted marketing professionals, business owners, and content creators seeking to understand AI search visibility.
What: Google’s AI Mode employs query fanout technique breaking user questions into dozens of related searches processed simultaneously across databases containing 250 million places and 50 billion products. The system makes automated phone calls to businesses, processes visual searches, and provides conversational interfaces for complex queries. Visual searching increased 65% year-over-year, with Circle to Search reaching 300 million Android devices. Businesses optimize visibility through traditional SEO practices, public relations coverage, and accurate Google Business Profile information rather than paid advertising within AI recommendations.
When: The interview occurred October 30, 2025, after AI Mode launched in March 2025 and expanded globally throughout the year. Automated calling features became available July 16, 2025, while visual search capabilities and multimodal interfaces rolled out systematically across markets. The discussion addressed nine months of AI search deployment affecting business discovery patterns.
Where: AI Mode operates in over 200 countries and territories with support for more than 35 languages. The automated calling feature functions throughout the United States except Indiana, Louisiana, Minnesota, Montana, and Nebraska. Circle to Search technology works on Android devices globally, while various AI features roll out based on geographic availability and regulatory considerations.
Why: Google transforms search from keyword-based ranking systems to conversational AI interfaces responding to complex, multi-part questions. Users ask queries three times longer than traditional searches, combining text, voice, and visual inputs. The company invests in AI infrastructure to maintain search dominance amid competition from ChatGPT and alternative platforms. Marketing professionals need understanding of these mechanisms because AI systems now mediate business discovery, requiring optimization strategies addressing both traditional search signals and AI-specific recommendation factors. The transformation affects how consumers find restaurants, book services, shop for products, and research purchase decisions.