Creating software is no longer a team sport
When implementation labor is commoditized, the practitioner's bottleneck moves from feature prioritization to change management and judgment.
Creating software is no longer a team sport.
In the traditional SDLC, producing software relied on a team of specialists — separate skill sets, authorities, and swim lanes. Each work stream was tightly orchestrated to extract as much parallelization and efficiency as possible from an inherently inefficient system. In agentic software development, those work streams collapse into a team of orchestrated agents. The specialist work becomes AI’s domain; the practitioner’s role becomes orchestration.
When the labor of implementation is commoditized to the speed of AI-generated text, the primary concern is no longer which version of features we build, or which insignificant slices of business value to prioritize. The constraint instead becomes the change management of the new capabilities we ship, and judgment about what’s best for the business. The size of the work becomes trivial.