Mixing PDFs and screenshots in WorkBuddy ruins exported tables—anyone know why?

Bros, the warehouse exploded again today. This morning, the team leader sent photos of delivery notes, screenshots of customer order changes, and several PDF scans to the group chat. I originally wanted to use WorkBuddy to summarize them into an exception tracking table using a fixed template, but the export looked like a mess.

I've been using WorkBuddy for about a month. I just started touching WMS (Warehouse Management System) these last two days and wanted to get the exception tracking table running first. I had sporadically used some AI tools before, also for about a month. I saw an article about information gathering in 2026 which said non-technical teams could turn info from web pages, PDFs, and images into structured tables, so I wanted to try it.

This time I didn't let it freestyle. I fixed the template first, with fields: Supplier, Arrival Date, Quantity, Exception Reason, Person in Charge, Deadline. The idea was straightforward: warehouses deal with the same crap every day, so automating it would be best. But after Excel came out, some dates turned into strings of numbers, some quantities were treated as dates, and supplier names had phone numbers mixed in. The 'Person in Charge' and 'Deadline' columns were basically empty. The template clearly had fields, but it seemed like it didn't recognize them.

I also tried uploading only PDFs once. It ran halfway and got stuck, finally giving me just a header row. I don't know if the field mapping wasn't set correctly, or maybe my description was too vague. Before, I thought just stating requirements clearly was enough; now I'm changing my mind a bit.

Has anyone used WorkBuddy for this kind of fixed-template export? Do you need to build field mappings first, or feed it step-by-step? I hate dragging things out; I just want to find a workflow that works.