When work is cheap, do more

Generative AI and automated workflows have made (some) engineering work much cheaper.

This creates an opportunity! But instead of using these productivity gains to produce more (arguably low-value) output, we should use them to raise the standard of the work itself.

What would our products look like if we did more of the flossing work; work that we know we should do more of but often skip because of the effort and time required? More documentation. More flake detection. More mutation testing. More observability. More performance benchmarking. More query plan inspection. More runbooks. More tests.

Used well, generative AI should help us do more of the work that improves quality, resilience, and understanding. It should help us to strengthen our engineering practice, and not just accelerate output.

Generative AI should be a tool for improving the quality of the work we do, not a tool for mass-producing low quality faster.