While the internet was losing its mind over the viral @Citrini7 post "The 2028 Global Intelligence Crisis" (citriniresearch.com/p/2028gic?...…), I was obsessing over another one. Economist Christian @ccatalini's "Some Simple Economics of AGI" tweetstorm (x.com/ccatalini/status/2026311...…) was one of the most thought-provoking things I've read about the second and third-order effects of AI. My key takeaway: as AI drives the cost of execution toward zero, value will accrue to verification, trust, and the bearing/underwriting of risk. As we at @FJLabs have written ad nauseum, we believe that marketplaces will thrive in the AI era. Anyone can now vibe-code the Uber app, but you can't replicate the ability to call a driver in cities all around the world in three minutes or less. The public markets echo this: the FJ Labs Marketplace Index has outperformed the IGV software index by ~25% since January 2025. Catalini concurs, but frames network effects in three tiers: -Execution-grade effects (listings, content, matching) that are vulnerable because agents can simply replicate or route around them. -Verification-grade effects (dispute histories, fraud logs, identity infrastructure) that get stronger as synthetic activity floods the internet. -And coordination/community/identity effects that are arguably the most durable because agents can't manufacture the shared history that grants a network legitimacy. This maps onto what we see across hundreds of marketplace investments, but I would go further than that. Many of the best marketplaces are deepening their moats through physical infrastructure layers that no agent can spin up overnight. @Vinted didn't become a €10B GMV behemoth solely on depth of listings, but rather through three integrated businesses: a marketplace, a payments platform (Vinted Pay), and a shipping network (Vinted Go) that operates robotic warehouses with proprietary software across Europe, handling tens of millions of parcels in 2025 alone. Clutch underwrites and reconditions every used car through a physical facility before it's sold. Amazon, Mercado Libre, and Coupang have invested billions into logistics, payments, and fulfillment that compound over years. And Catalini's points about underwriting and risk transfer are crucial as well. The reason Kirkland and Ellis and other white-shoe law firms will be fine in the AI age isn't because of their superior legal advice, but rather because they are a risk transfer product, not a law firm. Their key offering is CYA. We're exploring startups that serve as next-gen versions of risk transfer in the AI era. AIUC (Artificial Intelligence Underwriting Company) is a great example (sadly not an investor yet!). In sum, the cost of code is trending toward zero. The cost of trust is not.