Code Is Clay: How Agentic AI Reshaped Development

The traditional software development paradigm—where code was inflexible, difficult to modify, and required extensive upfront planning—has fundamentally shifted with the advent of agentic coding. What was once laboriously carved like stone can now be rapidly reshaped and refined like clay. This dramatic reduction in the cost of change enables developers to sketch out rough implementations, observe how they behave, and iterate within minutes rather than days or weeks.

This new malleability has spawned two complementary development approaches. The artisan method keeps developers in tight feedback loops, enabling them to manually refine every iteration and apply their judgment directly to both user-facing and code-level design. The factory method automates as much as possible through pipelines of specs, generators, and verification systems, only escalating to human oversight when necessary. Both require human judgment—the difference is timing: artisans apply it continuously, factories embed it upfront and in their automated systems.

Neither approach eliminates the need for engineering rigor. Code may now be easy to change, but knowing which changes matter still demands taste, context awareness, and architectural thinking. The choice between artisanal and factory workflows—or a hybrid blend—depends on the problem at hand. Large products benefit from both detailed craftsmanship on novel features and automated factories for repetitive work, while individual developers can use factories to handle scaling challenges while focusing their judgment on what genuinely needs it.