Presenters
Source
The AI Revolution in Search: How Google is Building the Future, One Iteration at a Time 🚀
Get ready to dive deep into the fascinating world of AI-powered search! We recently caught up with Robby Stein, VP of Product at Google Search, a true veteran who has navigated the product landscapes of both Google and Instagram. With Google Search serving billions of users and holding over 90% of the global market share, Robby is currently steering its most significant transformation: the shift to AI.
In this insightful conversation, Robby pulls back the curtain on how Google is tackling this monumental challenge, discussing everything from product-market fit metrics to learning from colossal failures, and the exciting future of personalized, proactive search.
💡 The AI-Powered Evolution of Google Search
Google’s journey into AI-powered search began with AI Overviews, evolving into more sophisticated models and culminating in the new AI mode. This conversational AI experience has already scaled to an impressive 75 million daily active users (DAU).
Robby emphasizes that AI mode isn’t just a new feature; it’s a game-changer for complex queries. Imagine planning a trip where one friend has an allergy, another has a dog, and you all want to sit outside near your hotel. AI mode can effortlessly handle these multi-faceted questions by leveraging real-time information from Google Maps and Travel.
While AI mode excels with intricate requests, traditional search remains vital for quick, direct queries. Google’s vision isn’t to replace one with the other, but to create a seamless, integrated experience. Users can simply type anything into the search box, and if AI is helpful, a preview will appear, allowing them to effortlessly transition into a conversational AI mode. This approach ensures that AI enhances, rather than complicates, the user’s search journey, offering an expansionary opportunity for new use cases like visual search with Google Lens and Circle to Search.
📈 Unpacking Product-Market Fit: Beyond the Billions
Building a product at Google’s scale might seem daunting, but Robby reveals that every great innovation starts small. For AI mode, the journey began with just 500 trusted testers (friends and family). The first true sign of traction? When testers shifted from complaining to proclaiming its usefulness.
Quantitatively, Google looks for a distinct signature of product-market fit: flat retention or a J-curve retention. This means that when you track a cohort of users, the probability of them returning to use the product flattens out over time, indicating consistent engagement. If the product continuously improves, this probability can even intensify. The goal is to see organic growth in users and usage week over week without artificial boosts.
For AI mode specifically, Robby highlights the depth of engagement (more follow-up questions) and recurring engagement as key non-obvious metrics. Unlike consumption-based products where a quick scroll counts as engagement, initiating a search query is a higher user commitment, making these metrics particularly telling.
🏆 Winning the Race, Not Just Starting It: Lessons from Instagram & AI
Google wasn’t the first to market with AI search, just as Instagram wasn’t the first with stories or short-form video. Robby explains that success isn’t about being first, but about solving a problem for users and aligning with shifting user expectations.
- Instagram Stories: Initially, people posted only “highlights” on their feeds. Stories provided a “pressure release valve” for casual, disappearing content, aligning naturally with what users wanted to share. Integrating Instagram’s creative tools and DM system made it uniquely Instagram.
- AI Search: Google isn’t aiming to build another general-purpose chatbot. Instead, it focuses on informational AI tailored to Google’s strengths: shopping, location, travel, and understanding the vastness of the web. Robby notes a latent demand where users were already adding “AI” to their queries, signaling their desire for more sophisticated answers.
The core idea is to discern how AI makes your product amazing for its specific use case, rather than just adding AI for AI’s sake.
🔮 The Future of Search: Personalized, Proactive, and Ubiquitous
Robby envisions a short-term future for search defined by three exciting themes:
- Deep Knowledge Integration: Search will become incredibly knowledgeable, performing dozens of traditional searches for you. Imagine buying a TV: AI could analyze price history, compare deals, show creator reviews, and even recommend a local installer, all in real time. This “query fanout” makes complex decisions effortless.
- Ubiquitous Modalities: Search will adapt to your environment. Whether you’re driving and asking live questions in the Google app, using Circle to Search on your Android phone, or getting context-aware help in Chrome, AI will be your seamless companion.
- Personalized & Agentic Experiences: Search will deeply understand you and proactively get things done. Instead of generic restaurant recommendations, it will suggest places based on your tastes, allergies, and even the size of your living room when buying a TV. The human remains in the loop for critical decisions, but AI handles the burdensome research and planning, providing “super leverage.”
🛠️ Building at Scale: Speed, Focus, and “Colossal Disasters”
Operating at Google’s immense scale requires a unique approach to speed and focus. Robby points out that advanced models now make it easier to add new functionality, as they require less custom tuning. Leaders must identify “five or ten-year opportunities” and focus on areas where they can uniquely intervene – often those that are risky or cut across organizational boundaries.
Robby, drawing from his founder background, emphasizes that everyone, from leaders to individual contributors, must adopt an “owner’s mindset.” This means having a clear vision, driving momentum, and taking full responsibility for outcomes.
Crucially, Robby shares invaluable lessons from “colossal disasters” at Instagram:
- Close Friends: Initially a confusing mess, it failed because it tried to do too much (feed, stories, private profile) and had poor localization. The lesson: Users wanted a safe space to share casually without judgment. By simplifying it to a stories-only feature with a distinctive green ring and encouraging larger lists (20-30 friends), it found its purpose and became a success.
- Reels: Launched integrated into stories in Brazil, it failed because creators wanted their content to live longer and go viral, not disappear in a day. The lesson: Understand the user’s core motivation. Reels needed to be a persistent format.
These stories highlight the importance of conviction before the numbers. Initial failures are inevitable, but the willingness to iterate, learn from losses, and keep pushing—like a golfer chasing that “perfect golf shot”—is what ultimately leads to breakthrough success.
✨ The Continuous Pursuit of Value
Robby’s insights paint a picture of Google Search as a dynamic, evolving entity, constantly pushing the boundaries of what’s possible with AI. The mission remains timeless: organizing the world’s information and making it accessible. Now, with AI, that mission is being realized in ways that were unimaginable just a few years ago. It’s an exciting time to be a user, as search transforms from a query box into an intelligent companion that deeply understands and proactively helps us navigate our complex world.