By playing, you agree to uListen's Terms and Privacy Policy and YouTube's Terms of Service.

Silicon Valley Girl

Google’s AI Search Expert: How to Get Ahead Before AI Changes Everything

What makes AI recommend one business - and ignore another? In this interview, Robby Stein, VP of Product for Google Search, breaks down how AI now decides what gets discovered, ranked, and recommended — and what founders, creators, and…

Marina MogilkohostRobby Steinguest

Oct 30, 202528mWatch on YouTube ↗

Summaries, chapters, transcripts, and FAQs on this page are made by uListen, not YouTube.

CHAPTERS

Jump to publisher timestamps

Chapter summaries

Why AI recommendations can make (or break) your business overnight
Marina frames the new reality: AI systems like Google’s AI Mode and chatbots can drive massive demand by recommending a business. Robby Stein, VP of Product at Google Search, is introduced as the person shaping how modern search ranking and discovery work.
- •AI mentions can cause sudden spikes in bookings and demand
- •Search is shifting from links-first to AI-assisted recommendations
- •PR and online presence increasingly influence what AI surfaces
- •Robby Stein’s role and why his perspective matters
From “keywordese” to natural-language search: what Google Search is becoming
Robby explains how search behavior is expanding: people still use Google traditionally, but increasingly ask multi-sentence, natural-language questions. Google’s AI can pull from web knowledge and structured Google context to answer more effortlessly.
- •Search use isn’t shrinking—it's expanding into new query types
- •Natural-language questions replace rigid keyword phrasing
- •AI taps broader context about the web, world, and products
- •Goal: reduce effort while improving answer quality
Personalization roadmap: what Google may (and may not) use about you
Marina asks whether Google will use personal data (Gmail, Drive, YouTube) to personalize results. Robby describes opt-in personalization experiments in Search Labs and hints at deeper integrations later, while keeping timelines open.
- •Opt-in ‘enhanced personalization’ is being explored
- •Search Labs: personalization experiments for shopping and local recommendations
- •Potential future connections to services like Gmail (TBD)
- •Personalization framed as “more helpful,” not automatic by default
Getting better outputs: prompting limits, Gemini vs Search, and “one super app” tension
Marina shares how she uploads research to generate narratives and scripts, then asks why the system won’t generate thumbnails. Robby clarifies capability boundaries (Search AI vs Gemini app) and acknowledges the fragmented user experience across tools.
- •Search AI can follow prompts, but has capability limits (e.g., no native image gen)
- •Gemini app may handle tasks Search won’t (like working with images)
- •Practical prompting: be specific about what you want
- •User desire: unify powerful features into a single workflow
AI Mode for local discovery: browsing lunch spots with rich context
They walk through AI-driven local recommendations that already reflect Marina’s location. Robby highlights how AI reasoning pairs with Google’s business data—hours, menus, reviews—to make choosing faster and more informed.
- •Local recommendations incorporate location context (e.g., Los Altos)
- •AI reasoning + Google’s place data creates a guided browsing experience
- •Surfacing: hours, menus, review highlights, and visuals in one flow
- •Focus on decision support, not just listing links
Live demo: “Find and book it for me” — AI gathers reservations across platforms
Robby demonstrates AI Mode researching sushi options and pulling availability from multiple reservation sources. The product reduces manual comparison work by aggregating options and presenting bookable time slots.
- •AI Mode kicks off multiple sub-searches automatically
- •Finds options across OpenTable/Resy and other sources
- •Returns consolidated booking choices and time slots
- •Main value: saves ~15 minutes of manual checking
How AI Mode chooses what to recommend: query fan-out + Google’s knowledge systems
Marina asks how recommendations are selected—especially for businesses competing via SEO and ads. Robby explains the “query fan-out” approach: the model issues many related searches and uses Google’s web results plus structured knowledge bases (including local place data) to decide.
- •AI Mode uses ‘query fan-out’ to run dozens of related queries
- •Uses Google Search itself as a tool ‘under the hood’
- •Combines web results with Google knowledge bases and real-time systems
- •Local pulls from hundreds of millions of place entities and their attributes
Ads in an AI-first search world: not disappearing, but evolving formats
Marina presses on whether Google Ads will fade as AI answers become more prominent. Robby argues search needs are expanding, and ads will likely adapt into new formats within AI experiences rather than vanish; AI Mode doesn’t rank based on ad spend, but ads experiments exist.
- •AI recommendations aren’t based on ad information for ranking
- •Ads aren’t going away; usage of Search is broadening
- •New ad formats may emerge for complex tasks (shopping, remodels, services)
- •Google is prioritizing consumer product quality while testing ads in AI experiences
Live demo: agentic calling—Google makes real phone calls to local businesses
They demo an “offline agent” that calls multiple pet groomers on Marina’s behalf and returns pricing and availability. The segment highlights a major shift: AI can complete real-world coordination, especially for small businesses with limited online booking.
- •User specifies needs (pet, breed, size, service) and AI initiates calls
- •Targets offline/local businesses without easy web workflows
- •Returns a summary: prices, availability, and unreachable businesses
- •Privacy note: calls aren’t recorded; results arrive via email
How to get recommended by AI: PR, authority signals, and classic ‘helpful content’ fundamentals
Marina asks what business owners should do now to be considered by AI systems. Robby says AI ‘thinks like a person’: it relies on reputable mentions, lists, and articles—making PR and strong content/SEO practices even more valuable because AI is effectively ‘Googling’ when it answers.
- •AI surfaces what it can reliably find—similar to human trust signals
- •PR and credible third-party mentions can heavily influence visibility
- •AI answers come from tool-using search + context windows
- •Traditional best practices still matter: clear, helpful, discoverable content
Reviews, fraud, and the new SEO: optimizing for longer, more complex AI questions
They discuss the impact of bought reviews and how hard it is to reduce ranking to a single factor. Robby notes the key change: AI queries are longer and more nuanced (how-to, purchase decisions, life advice), so creators should study emerging AI use cases and align content to those needs.
- •Bought reviews complicate trust; no single ‘one trick’ determines ranking
- •There’s overlap with SEO, but queries are becoming more complex
- •AI use cases skew toward how-to, decisions, and detailed constraints
- •Recommendation: study how people actually use AI and build for those patterns
AI shopping + multimodal search: product graph scale, visual lookup, and why checkout is cautious
Robby explains Google’s shopping advantage: a massive product graph with constant inventory and price updates, now connected to AI reasoning. They test visual shopping (photo → similar products) and discuss why Google still routes users to merchants rather than fully automating purchases.
- •Google shopping graph: tens of billions of products, frequent live updates
- •AI can answer with more comprehensive product/price context
- •Visual search: take a photo and ask for ‘similar’ (or ‘similar ingredients’)
- •Purchasing automation is risky—mistakes matter more in checkout scenarios
Google vs ChatGPT: differentiation through trusted knowledge, the open web, and inspiration tools
Marina asks about competition as people ‘GPT’ their questions. Robby argues Google’s edge is combining Gemini reasoning with Google’s deep knowledge systems (Finance, local, shopping) and direct connection to the web—plus fast-growing multimodal behaviors like Lens and live camera Q&A for inspiration.
- •Users increasingly demand AI-style interactions inside Search
- •Google’s advantage: tool access to high-quality vertical knowledge (Finance, local, shopping)
- •Connection to the web supports verification and deeper reading
- •Multimodal growth: Lens/visual search and live camera Q&A for inspiration
Building standout products in the AI era: find gaps, interview deeply, and optimize for stickiness
Robby shifts to product strategy: when building becomes cheap and fast, the differentiator is idea quality and user insight. He recommends hunting for ‘gaps,’ doing deep interviews (even a dozen), and identifying the moments users ‘hire’ or ‘fire’ a product—then designing for daily value when appropriate.
- •AI democratizes building—competition rises, idea quality matters more
- •Be a ‘student of gaps’: what do people wish tech did better?
- •Use deep qualitative research: observe, prototype, and interview ~12 people
- •Key interview prompts: when users loved it forever vs stopped using it
- •For mainstream consumer products, optimize for daily value and stickiness

Publisher timestamps

Get more out of YouTube videos.

Read along on your phone, search it beside the video, or ask your favorite AI app what was said.