Five Durable Verticals of the Web

DISTILLED SUMMARY

Source: Video transcript (~26 min) · Topic: What AI cannot replace on the web

AI app builders (Lovable, Replit, Bolt, V0) are thin wrappers around the same foundation models. Their moats evaporate as fast as a UI can be replicated — about a week with current tools. Training a proprietary model is not the escape. The companies that survive own something structural that model providers cannot replicate.

The web organizes around five durable verticals that persist regardless of how capable models become:

1. Trust. AI floods the web with indistinguishable apps and storefronts, including malicious ones. Verification layers — Stripe, Shopify, Apple's App Store, Vercel — become the routing infrastructure for responsible traffic. Autonomous agents will refuse to transact with unverified services.

2. Context. The most valuable resource online is situation-specific data: company records, customer relationships, domain knowledge. Companies that store and permission that context (Notion, Salesforce, Snowflake, Palantir) own the choke point every agent must flow through. An agent without context is a chatbot; an agent with context is a junior employee.

3. Distribution. When supply of software is infinite, curation is the scarcest resource. Gatekeepers (Google, Apple, Amazon, YouTube) grow stronger as the flood grows. Agent-to-agent discovery — an agent-native app store — is an unsolved and massive emerging problem.

4. Taste. When production is free, what you choose to produce is the entire game. Design sensibility, editorial judgment, and product-market fit intuition are human skills not derivable from training data. On the agentic web, taste manifests as orchestration quality — the thousand small decisions about how an agent should behave.

5. Liability. "The AI did it" will not survive in court. Regulated industries sell accountability. As AI grows more plausible-sounding, the stakes of mistakes rise. Liability management becomes a governance layer for autonomous agent systems.

The test: Does a 10x better model make your product obsolete or more valuable? If obsolete, reposition now. If more valuable, you're building on durable ground.