By fact checking the AI through a new verification process for Large Language Models, it’s possible to reduce hallucinations from 5-10 percent to as low as 0.1 percent.