"AI made producing software cheap, but understanding it is still expensive." That one sentence, from a 20-year engineering veteran, captures a tension every team using AI coding tools is feeling right now.
The argument comes from a proposed addendum to the Agile Manifesto published by a VP of Engineering. The original Manifesto, written in 2001, optimized for a world where writing software was the bottleneck. Developers were expensive, iteration cycles were slow, and the whole framework pushed toward producing working software faster. That made sense when a feature took weeks to ship.
The Core Problem: Code Is Easy, Comprehension Is Not
AI coding assistants like Cursor, Claude Code, and Cody can now generate hundreds of lines of functional code in seconds. The bottleneck has flipped. Teams aren't struggling to write code. They're struggling to understand what was written, verify it does what they think it does, and maintain it six months later when the person who prompted it has moved on.
The addendum proposes four updated values and three refined principles that shift emphasis from production speed to comprehension. The details are worth reading if you manage a development team, but the headline takeaway is this: "working software" as the primary measure of progress made sense when humans wrote every line. When an AI can produce working software in minutes, the harder question is whether anyone on the team actually understands what it built.
A Real Problem, Even If You Skip the Manifesto
You don't need to care about the Agile Manifesto specifically to feel this shift. Anyone who has used an AI assistant to generate a complex function, then spent 30 minutes trying to figure out why it broke in an edge case, has lived this problem. The code arrived fast. The understanding didn't.
The practical implication for teams: code review, documentation, and architectural clarity matter more now than they did two years ago, not less. Speed of production is no longer the constraint worth optimizing for. Speed of comprehension is.