The propensity of LLMs to "hallucinate" limits their practical application. However, their ability to analyze input based on prompt supplied parameters generally does not suffer this problem. This makes LLMs an adequate resource for classifying, and, extracting intelligence from, publicly available news sources at scale. This talk will showcase an example solution of this useless in which an application ingests a body of HTML for analysis, converts that HTML into a markdown format, assesses the text within the generated markdown for relevance based on criteria outlined within an LLM prompt, and finally, produces output text consisting of an LLM produced summary and extracted indicator set if found appropriate.