The Death of Search: How Shopping Will Work In The Age of AI

Unknown Source September 17, 2025 45 min

artificial-intelligence generative-ai startup investment google apple openai

🎧 Listen to Original

85 Companies

103 Key Quotes

4 Topics

3 Insights

🎯 Summary

Comprehensive Summary: The Decrapification of Commerce in the Age of AI Agents

This A16z podcast episode, featuring General Partner Alex Brampell and Partner Justine Moore, dives into the profound disruption AI agents are poised to bring to the world of e-commerce, focusing heavily on the current “crapification” of the web due to SEO spam and the resulting shift in consumer behavior away from traditional search engines like Google.

1. Main Narrative Arc and Key Discussion Points

The central narrative revolves around the unhealthy state of the current web, flooded with low-quality, SEO-optimized content, and how AI agents represent a potential “decrapification” mechanism. The discussion moves from historical context (affiliate marketing, the dominance of Google) to analyzing where AI agents will insert themselves into the commerce journey—from high-consideration purchases to impulse buys—and the strategic implications for incumbents (Google) and startups.

2. Major Topics and Subject Areas Covered

• Web Quality & SEO Junk: The saturation of the internet with low-value, search-engine-optimized content.
• The Future of Search & Discovery: How AI agents will replace traditional search for many queries, especially informational ones.
• E-commerce Attribution: The difficulty of accurately crediting marketing channels (last-click attribution is heavily criticized).
• Platform Shifts: Comparing the current shift to previous platform changes (e.g., the rise of e-commerce over physical retail).
• Consumer Purchase Ontology: Differentiating between impulse buys, considered purchases, and high-ticket research.
• Business Model Vulnerability: Assessing the threat to Google’s advertising revenue model.

3. Technical Concepts, Methodologies, or Frameworks Discussed

• Affiliate Marketing Mechanics: Detailed explanation of the traditional cookie/pixel tracking system used in affiliate models (dating back to TrialPay).
• Last-Click Attribution: Heavily criticized as the “most pervasively corrosive business model on the internet,” where the final touchpoint unfairly captures all credit.
• Consumer Surplus: Discussed in the context of dynamic, personalized pricing—a concept that is economically sound for producers but often unpopular with consumers and potentially regulatory challenging.
• PageRank/H-Index Analogy: Used to explain how Google initially established authority based on linking structures.

4. Business Implications and Strategic Insights

• Google’s Imperilment: Google’s freemium model is under threat. They are losing the free, informational queries (e.g., “Who won the Oscar in 1977?”) to LLMs like ChatGPT, while their premium, monetizable queries (commerce-related searches) remain relatively stable for now. The “tax” Google levies on GDP via clicks may shift elsewhere.
• Aggregators vs. Single-SKU Retailers: The discussion suggests that aggregators (like Amazon, Shopify) are better positioned to survive AI disruption than single-SKU, trend-dependent brands (like Casper or Allbirds), as aggregators can ride any emerging consumer trend.
• The Subscription Advantage: Businesses attached to recurring revenue (like Dropcam/Nest) have more durable models than pure commodity resellers who rely on constantly acquiring new, one-time customers.

5. Key Personalities, Experts, or Thought Leaders Mentioned

• Alex Brampell: Shared experience founding TrialPay, an early affiliate marketing company.
• Justine Moore: Highlighted observations from consumer behavior, such as the use of CamelCamelCamel and viral examples of teens using ChatGPT for visual product discovery.
• Bill Gross: Mentioned as the originator of the Overture business model that Google copied for AdWords.

6. Predictions, Trends, or Future-Looking Statements

• Agentic Commerce: AI agents will increasingly handle the entire purchase circle for considered items, moving beyond just providing information to executing transactions when prices are right.
• Inculcating Demand is Hard for AI: AI agents are currently poor at generating new demand or recognizing ephemeral social trends (like the “Bama Rush” shoe trends), which still requires human observation and social context.
• Dynamic Pricing Challenges: While technically feasible to charge different prices based on inferred consumer wealth (e.g., iPhone vs. Android users), this faces significant regulatory and customer backlash hurdles.

7. Practical Applications and Real-World Examples

• CamelCamelCamel: Cited as a prime example of “inefficient AI” where consumers actively monitor price drops, demonstrating a clear demand for automated price action.
• Teenage Visual Search: Viral examples of young consumers using ChatGPT to identify and find alternatives for fashion items seen in street style photos, often leading to recommendations for more affordable options.
• Impulse Buys: AI is unlikely to drive impulse purchases (like checking out at a supermarket) as these are emotionally driven, not rationally researched.

8. Controversies, Challenges, or Problems Highlighted

• Attribution Complexity: The rise of AI agents interacting with ads, social media, and search will make existing attribution models (especially last-click) even more inaccurate and corrosive.
• Commodity Reselling Failure: The internet killed the long tail of local commodity resellers, and direct-to-consumer brands that simply rebrand OEM products struggle because they lack barriers to entry and are vulnerable to trend shifts.
• LLM Hallucinations: Early issues with LLMs hallucinating product recommendations initially slowed down their adoption for direct commerce use cases.

9. Solutions, Recommendations, or Actionable Advice Provided

*

🏢 Companies Mentioned

Android ✅ tech

Gemini ✅ tech

New Balance ✅ apparel/tech

On running shoes ✅ apparel/tech

Adidas ✅ apparel/tech

Nike ✅ apparel/tech

Walgreens ✅ retail

Delta ✅ tech

Prime team ✅ tech

Universal Product Code ✅ unknown

Buy It For Life ✅ unknown

TikTok Shop ✅ unknown

Jeff Bezos ✅ unknown

Now I ✅ unknown

And AliExpress ✅ unknown

💬 Key Insights

"the most consideration end of the purchase, I think it's hard to have that before AI end-to-end because while you may start doing your research online on ChatGPT or Gemini or any sort of new AI-native property that shows up, the purchase is so significant that you're probably going to want to have some sort of in-person experience where you're seeing the thing, touching the thing, experiencing the thing, talking to another human expert about it, right?"

Impact Score: 10

"I think the—so both ends of the spectrum, I think, are harder for AI to disrupt."

Impact Score: 10

"But it is a very, very unique business model that will stand the test of time, and I think it's AI-proof."

Impact Score: 10

"normally it doesn't work because it's like, "Hey, just trust me because I'm the best." Well, you have to have many, many decades of trust such that Justine's mom is like, "I don't know what it is, but if it's sold at Costco, it's good.""

Impact Score: 10

"Why would they refuse [to take a high gross margin]? ... Because it degrades the value of the membership. They make money from the membership."

Impact Score: 10

"No matter how, like no more hallucination, like everything is awesome, but most of the things on the internet are crap, and they're crap, and we know that they're crap, but the SEO-optimized crap in order to earn affiliate commissions. And like summarizing that crap is not helpful. So, how do you decrapify that?"

Impact Score: 10

📊 Topics

#artificialintelligence 70 #generativeai 27 #startup 2 #investment 1

🧠 Key Takeaways

💡 turn all of those high-quality videos into transcripts, and an LLM can then read and review and make recommendations