Don't trust the NIH with your health. They're the reason the mass majority of Americans are deficient in vitamin d, potassium, magnesium and a bunch of other vital vitamins/minerals, primarily due to faulty research done 50 years ago and never updated.
Humans spent thousands of years out in the sun on a daily basis growing crops and hunting. They never got a vitamin D overdose and were in fact much healthier than you and I.
It's about the ability to combine ideas in novel ways, without breaking the rules in relevant frameworks. Sometimes the idea may even be to contradict existing theories where they are weak.
I wrote something similar where you specify the intent in Markdown at the file level. That can also be done by an AI agent. Each intent file compiles to a source file.
It works, but needs improvement. Any feedback is welcome!
I wrote IntentCode to let people write high-level specs in natural language, or directly in pseudo-code (Markdown with guidelines I call the IntentCode).
This is an AI compiler, the IntentCode files get compiled to source 1:1. The examples so far are quite basic (calc CLI app) but prove that the concept is sound.
You can also write extensions, which are mainly skills files for now.
I intend to keep progressing the compiler, so that larger and more complex examples can be implemented.
I'm actually building this and believe I've overcome the most difficult aspects mentioned here. It will be released as Open Source next week. https://intentcode.dev
reply