One consequence of the LLM-geist is that your clients/bosses might think that any kind of dusty pile of unstructured data is now an immediately actionable goldmine of insights. The reality is that a S3 bucket filled with pdfs, slides, and random user tracking data, needs a lot of creative finagling to become useful in any way. So what does it look like to turn slush into data without just making up stories. What does it mean to analyze data when you're basically making up "the data" as you go?

Bluesky | Website