Source: Nate’s Substack
Rather than wait for perfectly reliable AI systems, this piece proposes borrowing institutional scaffolding from medieval guilds—QA functions, review boards, appeals processes—to make unreliable agents accountable through structure rather than capability. Consumers don't need to trust the technology itself if they trust the organization operating it, which inverts how most AI companies frame the adoption problem. The near-term competitive advantage belongs to platforms that can layer traditional institutional practices around AI outputs, not those chasing alignment or interpretability breakthroughs.