Have you ever wondered why Google could become a $2 trillion giant while Wikipedia remains a nonprofit organization? The answer is simple: the magic of commercial search. When you search for 'how many protons are in a cesium atom', Google does not earn a dime. But when you search for 'the best tennis racket', it starts printing money. This asymmetry defines the very essence of the search economy. Now, with the rise of AI, this balance is being fundamentally disrupted.
I recently read a deep analysis by a16z partners Justine Moore and Alex Rampell, and their insights on how AI is reshaping the e-commerce field left me deeply shocked. They not only analyzed the threats Google may face but, more importantly, depicted a new picture of e-commerce in the AI era. In this picture, the traditional search-compare-buy model is being replaced by an AI agent-driven intelligent purchasing experience. I have spent a lot of time reflecting on their viewpoints and combining them with my observations of this industry to share some deeper thoughts.
Google's real crisis: not search volume, but value migration.
Justine mentioned a point in the article that impressed me: even if Google loses 95% of its search volume, its revenue may still grow as long as it can retain those commercially valuable queries. This point sounds counterintuitive but actually reveals the core secret of the search economy. After deep thinking, I found that a deeper issue lies behind this: AI is changing the locus of value creation.
In the traditional model, Google plays the role of an information intermediary. Users have purchasing intent, Google provides search results and advertisements, merchants gain traffic, and Google collects advertising fees. This is a relatively simple three-party game. But the emergence of AI agents breaks this balance. When ChatGPT or Perplexity can directly answer questions like 'what is the best tennis racket' and provide specific recommendations, why would users still need to click on Google's ad links?
More critically, AI is not just answering questions; it is redefining 'search' itself. Our previous search behavior was: pose a question → receive a list of links → click to view → compare information → make a decision. The AI agent's process is: describe needs → receive recommendations → purchase directly. The intermediate stages of comparison and research are significantly compressed or even disappear. This means traditional search engines not only lose query volumes but also lose their key position in the decision-making chain.
From the testimony of Apple Senior Vice President Eddy Cue at the DOJ antitrust trial in May 2025, we can see some clues. He stated that Safari's search volume had declined for the first time in over twenty years, and this news directly led to Alphabet's stock price falling nearly 8% in a single day, evaporating more than $150 billion in market value. Although Google's Q2 financial report shows that search revenue is still growing, indicating that what is currently lost are mainly low-value queries, the direction of this trend is clear.
I believe that Google is facing not just a simple competitive threat, but a structural challenge to its business model. When AI can directly complete the entire process from intent recognition to purchase decision, the traditional model of 'traffic → advertising → conversion' will become inefficient or even outdated. What Google needs is not a better search algorithm, but a brand new business model to adapt to AI-driven consumer behavior.
The AI transformation of five types of purchasing behavior: from impulse to deep reflection.
Justine categorized purchasing behavior into five categories, from impulse buying to life purchases, each of which will undergo varying degrees of change in the AI era. I find this classification framework very precise, but I want to analyze the underlying psychological mechanisms behind each type of purchasing behavior and how AI reshapes these mechanisms at a deeper level.
Impulse buying may seem like the area least affected by AI, as impulse implies a lack of rational research process. But I think this judgment may be overly superficial. The true power of AI lies in predicting and guiding impulse. Imagine when you see a funny T-shirt on TikTok, AI has already analyzed your browsing history, purchase records, social media activities, and even your emotional state, then pushes the product that best meets your current psychological needs at the most precise moment. This is not just simple algorithmic recommendation; it is a deep understanding and manipulation of human impulsive psychology. I feel that this personalized impulse guidance could make impulse buying more frequent and precise.
The AI transformation of routine essentials is the easiest to understand and implement. However, I have observed an interesting phenomenon: as AI begins to take over our daily purchasing decisions, our consumption habits may undergo subtle changes. For example, AI might adjust your purchasing timing and quantity based on price fluctuations, inventory levels, or even weather forecasts. A smart AI agent might notice a certain brand is on sale a week before your laundry detergent runs out, prompting an early purchase and suggesting you try it. This 'intelligent arbitrage' behavior may allow consumers to unknowingly achieve better value for money while forcing brands to rethink their pricing and promotional strategies.
Lifestyle purchases are the area where I believe AI will have the greatest impact. The characteristics of these purchases include: a certain price threshold, involving personal taste, and requiring a certain degree of research. Justine mentioned products like Plush, but I believe this is just the tip of the iceberg. The real revolution will come from AI's deep learning of individual styles and preferences. Imagine an AI assistant that not only knows what you've bought in the past but also understands your body shape, skin tone, lifestyle, social circles, and even your aspirations. It can recommend not just individual products, but entire outfits, or even an upgrade path for your lifestyle. This level of personalization is unattainable by traditional e-commerce platforms.
The AI transformation of functional purchases is the most complex and challenging. These purchases typically involve significant expenditures and long-term use; consumers need not only product recommendations but also expert consultations. I believe a new category of AI applications will emerge: AI consultants. These AIs will not only possess extensive product knowledge but also engage in deep dialogues similar to human sales experts. They can inquire about your specific needs, usage scenarios, budget constraints, and even your future plans, then provide highly personalized suggestions. More importantly, these AI consultants will be cross-brand and will not favor any specific product due to commissions or inventory.
Life purchases may be the area least affected by AI but also the most important. Decisions like buying a house, getting married, and education are too significant and personal to be fully entrusted to AI. However, AI can play an important role in information gathering, option comparison, and risk assessment. The AI coach I envision is not supposed to make decisions for you but to help you make better decisions. It can organize vast amounts of information, identify potential pitfalls, simulate the long-term consequences of different choices, and even assist in contract negotiations. I believe the value of this AI coach lies in its neutrality and comprehensiveness, unlike human consultants who may have conflicts of interest.
Amazon and Shopify's moats: a dual advantage of data and infrastructure.
Justine pointed out in the analysis that Amazon has a stronger defensive capability compared to Google. I completely agree with this viewpoint, but I want to analyze the source and sustainability of this advantage at a deeper level. Amazon's advantage lies not only in its control over the entire chain from search to delivery but, more importantly, in its mastery of the most valuable behavioral data.
Amazon knows what you bought, when you bought it, how quickly you received it, whether you returned it, and whether you repurchased it, and so on. The value of this data far exceeds search history because it directly reflects real purchasing behavior and satisfaction. When AI agents need to make purchasing decisions for users, this data is the most valuable training material. Although Google knows what you searched for, it does not know what you ultimately bought or whether you were satisfied with the purchase outcome. This data gap will be further magnified in the era of AI.
More importantly, Amazon Prime's loyalty program creates a unique economic phenomenon: sunk cost bias. When you have already paid to become a Prime member, you tend to buy more items on Amazon to 'break even'. This psychological mechanism may become even stronger in the AI era. When AI agents seek the best purchasing options for you, they may naturally lean towards Amazon because they know you are a Prime member and can enjoy free shipping and other benefits.
Shopify's defensive logic is completely different but equally powerful. It does not build a moat by controlling consumers but by empowering merchants to create network effects. As more and more D2C (Direct-to-Consumer) brands choose Shopify, the platform becomes increasingly irreplaceable. In the AI era, this decentralized advantage may become even more apparent. AI agents may need to obtain information and complete purchases from hundreds of different brand websites simultaneously, and if these sites all operate on Shopify, it will create a standardized API ecosystem.
I believe Shopify has another underestimated advantage: it is closest to brand storytelling. In the AI era, the functional differences of products may be quickly recognized and compared by AI, but the emotional connection of a brand still needs to be felt by humans. Brands on Shopify often have unique stories and cultures, and these soft values are difficult for AI to fully quantify but are important factors influencing consumer decisions.
Four foundational infrastructure challenges for AI commercialization.
Justine mentioned four foundational conditions needed for AI to realize its full potential in business, and I believe each is worth exploring in depth, as they are not only technical challenges but also opportunities for business model innovation.
First, there is the issue of better data. The current product review system indeed has serious problems: fake reviews, polarization, and lack of background information. But I believe the root of the problem lies in the misalignment of incentive mechanisms. Consumers typically write reviews because they are extremely satisfied or extremely dissatisfied; very few people document their neutral experiences. Moreover, existing review systems cannot capture the usage scenarios of products, users' expectations, and changes over time.
The ideal data system I envision is as follows: AI agents not only collect users' subjective evaluations but also monitor the actual usage of products through IoT devices. For example, a smartwatch should not only check whether users gave a five-star rating but also look at the frequency and duration of actual use. A coffee machine's evaluation should not only consider textual feedback but also the user's actual usage frequency, cleaning and maintenance conditions, etc. This combination of objective usage data with subjective feedback can form a truly valuable product evaluation system.
The challenge of unified APIs is more political than technical. Each e-commerce platform has its own API structure, data format, and authentication mechanism, and these differences are largely intentional, aimed at creating platform lock-in effects. But in the AI agent era, this fragmentation may become an efficiency bottleneck for the entire industry. I predict that specialized API aggregation services will emerge, similar to the global distribution systems in the travel industry. These services will standardize the interfaces of different platforms, allowing AI agents to seamlessly compare and purchase across platforms.
Identity and memory are the most complex challenges because they involve balancing privacy, accuracy, and adaptability. I believe future AI shopping assistants need to establish a multi-layered preference model. This model should not only record your purchase history but also understand your values, life stage, financial constraints, etc. For example, it needs to know that you seek convenience during weekday lunches but focus more on quality and presentation during weekend gatherings. This situationally aware recommendation requires AI to possess a nearly human-like social understanding ability.
Embedded capture may be the area with the most innovative potential. Traditional data collection is passive and delayed: reviewing after purchase, providing feedback after use. But AI agents can achieve real-time preference learning. For example, when you are browsing a particular product and linger on a specific feature for a long time, AI can infer that you are particularly interested in that feature. When you quickly skip over certain color options, AI can learn your color preferences. This micro-interaction analysis can enable AI to have a more detailed understanding of your preferences.
The reshuffling of e-commerce platforms: who will emerge victorious?
After reflecting on Justine's analysis, I have formed some of my own judgments about the future landscape of the e-commerce industry. I believe AI will trigger a new platform reshuffle, but the logic for winning will be different from before.
Competition in the traditional e-commerce era mainly revolves around three dimensions: richness of choice, convenience, and price. Amazon wins in terms of choice with the 'Everything Store' concept while establishing advantages in convenience through Prime. However, in the AI era, the importance of these advantages will change.
When AI agents can automatically compare prices across the web and act on purchases, the price advantages of a single platform will be diluted. When AI can intelligently process and fulfill orders across platforms, the definition of convenience will also change. True competitive advantages will shift to data quality, AI capabilities, and ecosystem integration.
I predict several new platform players will emerge: AI-native e-commerce platforms, vertical AI agents, and commercial infrastructure providers. AI-native platforms will be designed from the ground up, focusing on the needs of AI agents, providing structured product data, standardized APIs, and AI-friendly user experiences. Vertical AI agents will focus on specific categories, such as fashion AI, digital product AI, or home improvement AI, establishing competitive advantages through deep specialization. Commercial infrastructure providers will offer underlying technical services to help traditional e-commerce platforms become AI-enabled.
I also believe a new business model will emerge: AI agent subscriptions. Consumers may no longer shop directly on various e-commerce platforms but subscribe to one or more AI shopping agents that make all purchasing decisions on their behalf. These agents will charge a subscription fee rather than a commission, thus avoiding conflicts of interest and truly standing in the consumers' shoes. This model could redefine the value chain distribution in e-commerce.
The AI reconstruction of brand marketing: from mass marketing to individual dialogues.
AI's changes to business extend beyond purchasing behavior; they will fundamentally reshape the logic of brand marketing. In the era of AI agents, the effectiveness of traditional mass marketing will decline significantly, as consumers no longer actively search for and compare products but rely on AI agents' recommendations.
This means brands need to learn to communicate with AI rather than with humans. AI agents will evaluate products in a more rational and data-driven manner; they will not be influenced by beautiful packaging or emotional advertising but will focus on objective performance metrics, cost-effectiveness, and user satisfaction scores.
But this does not mean that brand storytelling becomes unimportant. On the contrary, I believe authentic brand narratives will become even more important, as AI agents will deeply analyze the consistency and credibility of brands. If a brand communicates contradictory messages across different platforms and at different times, AI can easily identify this and reduce the recommendation weight.
I predict that a new marketing role will emerge: AI Relationship Officer. The job of these officers will be to ensure that all aspects of a brand's product information, pricing strategies, inventory management, etc., can be correctly understood and assessed by AI. They need to optimize product data, manage API integrations, monitor AI recommendation patterns, and so on.
Another important change is the extreme personalization. When AI agents have a deep understanding of each consumer, brands can offer customized products to everyone. This is not just personalized recommendations but personalized products themselves. Imagine when your AI agent tells a clothing brand your exact size, color preferences, material requirements, and budget range, that brand can customize a unique item just for you. This kind of large-scale customization becomes economically feasible in the AI era.
The next ten years: What are we witnessing?
After deeply reflecting on Justine's analysis and my own observations, I feel that what we are witnessing is not just a transformation of the e-commerce industry but a deeper shift in economic behavior.
Traditional economics assumes that consumers are rational actors who actively gather information, compare options, and make optimal decisions. But in reality, we all know that human decision-making is full of biases, emotions, and cognitive limitations. The emergence of AI agents may make consumers more 'rational' because AI can process more information, avoid emotional biases, and consistently apply decision-making criteria.
The proliferation of rational consumption could have far-reaching effects. First, market efficiency will significantly increase as consumers can more accurately assess product value. Second, product quality will become more important than marketing ability, as AI agents will not be misled by flashy advertisements. Finally, price transparency will increase, as AI can easily compare prices across the web.
But I am also concerned that this 'hyper-rational' consumption may bring some negative consequences. The joy of discovering while shopping may diminish because AI agents always recommend the 'optimal' choice rather than options that are surprising or delightful. Impulse buying, while not rational, is also part of the enjoyment of life. If everything is optimized by AI, life may become overly predictable.
From a more macro perspective, I believe the application of AI in the commercial field will accelerate economic digitization. More and more commercial behaviors will be digitally recorded and analyzed, providing an unprecedented data foundation for economic planning and policy-making. Governments may be able to predict economic trends more accurately, identify market failures, and design targeted interventions.
I predict that within the next decade, we will see AI-driven commerce evolve from experimental applications to mainstream practices. Early adopters will gain significant competitive advantages, but as technology becomes more widespread, these advantages will gradually be commoditized. The true long-term winners will be those companies that can redefine customer value in the AI era.
