With LLMs getting smarter every day, the next generation might be writing specs like this instead of actual code.