Raw LLM Responses

Vertical slice architecture where individual modules are self contained and have their own arch/promt files along with the high level ones. Running type safe programing - almost all my projects are now C#/Blazor .net. Keeps JS bullshit out of the way and having a build step before deploy means the AI automatically fixes stuff. Run clean up tasks to keep documentation in sync and simplify stuff (deduplicate and remove unused specification references). Pretty much 0% error rate in greenfield projects. I am at the point where we are in a classic Wait Calculation / Lightspeed Leapfrog scenario. The first spaceships we sent are going to get lapped by the newer and the speed of development is so insanely fast with AI that all projects should be re-built/re-architected to promot AI dev.