This episode breaks down how AI agents are set to dismantle Google's long-standing "tax on GDP," fundamentally reshaping e-commerce by automating everything from product research to the final purchase.
The Genesis of AI in E-Commerce
- Affiliate Marketing: This is a model where a partner earns a commission for marketing another company's products. It has historically relied on cookies (small data files stored on a user's computer) and tracking pixels (invisible images on a confirmation page) to attribute sales.
- Alex notes that this cookie-based system, which powers much of the internet's commercial layer, seems ill-suited for the new paradigm of AI-driven purchases.
- Justine adds that despite the massive opportunity in online shopping, few startups are tackling it with AI. She was motivated to explore the system's complexity and identify where AI could play a role, hoping to uncover how founders are approaching the challenge.
Observing Present Behavior to Predict Future Trends
- Alex points to Camel Camel Camel, a price-tracking tool for Amazon, as a prime example. Users manually set price alerts for products they intend to buy, effectively acting as their own inefficient agents. He states, "The consumer is the agent and this is like a very, very inefficient AI." This behavior signals a clear demand for automated purchasing agents that can act on pricing information.
- Justine highlights another trend: teenage girls using ChatGPT to identify clothing from photos of celebrities like Taylor Swift. This demonstrates a user-driven push for AI to handle complex, visual-based product research, moving beyond simple text queries. This demographic is often a leading indicator of broader consumer shifts.
The Limits of Dynamic Pricing and E-Commerce Growth
- Alex explains that while dynamic pricing—charging different prices to different customers—is theoretically efficient for producers, it often faces regulatory and consumer backlash. Past attempts included charging iPhone users more than Android users, based on the assumption of different price elasticities.
- E-commerce currently accounts for only 16% of total retail sales. Alex attributes this to two factors:
- The demand for immediacy: Consumers will pay a premium for instant gratification, like buying toothpaste from a local Walgreens at 10 PM instead of waiting for a 7 AM Amazon delivery.
- The experiential nature of shopping: Many people shop at malls for entertainment or to physically interact with products, especially for aspirational purchases like a Rolex.
- Justine adds that the 16% figure may undercount the influence of online activity, as many consumers research products online before making an in-person purchase (e.g., comparing laptop weights at an Apple Store).
The Attribution Problem: The Internet's Corrosive Business Model
- Alex criticizes last-click attribution, a model that gives 100% of the credit for a sale to the final touchpoint before purchase. He argues this model is fundamentally flawed because it ignores the entire customer journey, from seeing a Super Bowl ad to reading a Reddit post.
- He singles out coupon services like Honey as a form of "theft," as they insert themselves at the very end of the purchase funnel to claim affiliate commissions without having driven the initial demand.
- Strategic Implication: As AI agents become intermediaries, the attribution model will be further complicated. An AI might be the "last click," but it's summarizing information from numerous other sources. Investors should watch for new attribution technologies that can more accurately map the AI-driven customer journey.
Why Aggregators Won and DTC Brands Stalled
- Alex explains that many DTC brands are essentially commodity resellers who don't manufacture their own products. They rely on buying traffic from Google and Facebook, making the tech giants the true victors. With no real barrier to entry, competitors can easily source the same products from the same manufacturers and undercut them on price.
- Justine adds that consumer product markets are trend-based. A brand like Allbirds might be popular one year, but trends shift quickly to other brands like On Running. Aggregators like Amazon can ride any trend, whereas single-brand companies are vulnerable.
- Key Insight: AI agents are likely to favor aggregators or go directly to the source, further challenging the viability of undifferentiated DTC brands that rely heavily on paid marketing to create demand.
Google's Vulnerability and the "Crap" Internet Problem
- Alex frames Google's business model as a "tax on GDP," capturing a percentage of consumer spending that begins with a search query.
- Currently, AI tools like ChatGPT are siphoning off Google's free informational queries (e.g., "Who won the Oscar in 1977?"), but not yet its premium commercial queries. Google's ad revenue remains strong because purchase-intent searches still happen there.
- However, a major obstacle for AI is the poor quality of online content. Alex states, "Most of the things on the internet are crap... they SEO optimize crap in order to earn affiliate commissions and like summarizing that crap is not helpful." This pollution of the open web with affiliate-driven, low-quality articles makes it difficult for an LLM (Large Language Model)—an AI trained on text data—to provide trustworthy recommendations.
The Search for Trust: Costco as the Ultimate AI-Proof Model
- The speakers contrast the polluted, affiliate-driven web with the business model of Consumer Reports, a publication that built trust by refusing advertising.
- Alex identifies Costco as the ultimate modern example of a trust-based commerce model. Costco's primary revenue comes from memberships, not product margins. This incentivizes them to curate high-quality products at the lowest possible price to preserve the value of the membership.
- Justine notes her mother's unwavering trust in Costco for everything from glasses to flights, demonstrating the power of this long-term strategy.
- Strategic Implication: Businesses that act as trusted curators for their customers, like Costco, are inherently more resilient to AI disruption. Their value proposition is not just about price or selection, but about a guarantee of quality that an AI summarizing the open internet cannot replicate.
How AI Will Reshape Different Purchase Types
- Impulse Buys (e.g., a t-shirt on TikTok Shop): AI will have minimal impact here, as these purchases are immediate and not research-driven.
- Highly Considered Purchases (e.g., a house or car): AI will be a powerful research tool, but the final transaction will likely remain in-person due to the significance of the purchase.
- The Middle Ground (e.g., handbags, laundry detergent, laptops): This is where AI agents will thrive.
- For known items, agents will perform price optimization, automatically buying a product like laundry detergent when it hits a target price.
- For research-heavy items like a travel handbag, agents can synthesize reviews from Reddit and TikTok to provide a tailored recommendation, potentially leading to a direct purchase.
- Alex adds another lens: whether a product has a UPC (Universal Product Code), a scannable barcode for tracking. If an item has a UPC, an AI agent can easily run an algorithm to find the lowest price across all retailers. If not, the process remains more discovery-oriented.
Where New Opportunities Will Emerge
- Specialized Shopping Agents: While ChatGPT may remain a horizontal player, there is a significant opportunity for specialized agents focused on optimizing purchases for consumers who value money over time. These agents would integrate credit card rewards, coupon codes, and affiliate cashback to automate finding the absolute best deal.
- Merchant-Side Infrastructure: Justine points out a massive opportunity on the merchant side. As AI agents begin browsing and purchasing on behalf of users, websites and payment infrastructure will need to be rebuilt. New companies will emerge to provide the tools for merchants to make their sites "agent-browseable" and to securely handle AI-initiated transactions.
- Key Insight: The next wave of innovation won't just be consumer-facing agents. A parallel ecosystem of B2B infrastructure will be required to support this new form of commerce, creating a fertile ground for investment.
Conclusion
This episode reveals that AI agents will fragment search and automate purchasing, shifting value from incumbents like Google. Investors and researchers should focus on specialized agent startups and the new merchant-side infrastructure required to support them, as these are the areas poised for significant growth.
