Are Superhuman Agents Here?

A discussion on the current state and future trajectory of AI coding agents. The key insight is that agents face the same fundamental challenge as human developers: breaking down big ideas into small, actionable chunks of work. While agents still struggle with vague, high-level tasks and require constant clarification, they are expected to make dramatic progress on more concrete tasks like picking up issues, fixing bugs, and navigating codebases. Trust and day-to-day reliability are identified as critical milestones before agents can be considered median-programmer equivalents.