Large Language Mathematicians: Public Records in Record Time

AI is great at a lot of things, but math and number crunching are not its strong suits. The American City Business Journals wanted a tool to surface stories from public records, but kept running into the same problem: the LLMs would dream up what it thought was the right numbers, instead of actually doing the math. In this session, we'll explain how we paired a series of LLM inputs with behind-the-scenes JavaScript to transform hundreds of public records into a useful output for reporters and editors. This same logic flow has been applied to anything that requires a precise output: word counts, data analysis, and more.