Why is investing in the build loop so often wasted?#
The build loop is where agentic coding tools do their best work, and speeding it up is real progress. But the constraint usually sits elsewhere: in review, approval, release and measurement, inside the delivery loop. Faster code arriving at the same queue only makes the queue longer.
What does a misplaced investment in the build loop look like?#
Low flow efficiency: a wide gap between active time and elapsed time. If a change takes forty minutes of work and eleven days to reach production, another coding assistant will not close the gap. Reliable data is rarer than folklore; one study of 63 teams at ASOS found flow efficiency from 9% to 68%, so map your own value stream first.
The same mechanism drives the AI productivity paradox, and it is why writing code was never the bottleneck.