The question has moved past whether coding agents can produce code that works. The harder question is whether they can complete real software work safely, measurably, and repeatedly — and how we supervise it. Great piece by @realjustinbauer on the evolution of coding agents and

→ View original post on X — @snorkelai