@swyx Similar Git+GitHub moment happening for AI agents with constraint layers. The 'aligning thousands of stakeholders' bit resonates — we're finding ontology-style guardrails do for agent reliability what version control did for code collaboration.