At least validation seems faster than without, but you get what you pay for when it comes to llm intelligence