Not only can we use language models to answer question from a context we can also use structed extraction to provide character level citations that allow us to fact check each statement in the context, minimizing halucinations.