WorkBuddy Scheduled Task Extracts Scanned Docs but Date Column is Empty - Config Issue?

No orders taken today.

Client sent contract attachments last night—mixed PDFs and phone photos—to organize into a translation list. I've used WorkBuddy for about four weeks, mainly for task breakdown and workflow planning. Automatic field extraction is a first attempt. Saw a tutorial claiming file processing and automation can batch-extract info. Wanted to throw this pile in.

Created test folder with ~10 contracts, quotes, scans, plus scheduled task. Prompt: 'Identify client name, date, amount, language in each file, output as CSV.' First run showed completion. Exported table had only filenames. Fields barely landed; dates, amounts, languages jumbled. Second attempt changed prompt to 'extract' and instructed to leave blank if field not found. Output structure still messy, headers misaligned.

I suspect field mapping in scheduled tasks isn't configured properly. Field recognition might not explain everything. Tutorial mentioned binding templates or Excel headers first, but couldn't find where to set that. Previously thought WorkBuddy suited clean PDFs. Today's test suggests even photo reading stability is questionable.

Has anyone succeeded with similar tasks? Should prompts be shorter? Or is auto-summarization inherently unable to deliver ready-to-use tables in one shot?