WorkBuddy Keeps Getting Stuck Summarizing Expense Reports: Any Ideas Why?

Last night at 9:30 PM, the client dumped another zip file in WeChat. Inside were reimbursement PDFs, invoice screenshots, and a messy Excel. I've been using WorkBuddy for about a month, originally just wanting it to consolidate files into one CSV, but I got stuck again.

I created an isolated test folder, didn't connect WeCom, nor Feishu. I only just started using WeCom these days and dare not mess with permissions. Then I asked it to extract date, supplier, amount, tax, and export CSV. Identifiable ones were okay, but amounts in two scanned PDFs remained empty, and once the entire exported table was all nulls, with a prompt saying "Parsing result does not match target structure." I didn't understand this prompt.

I tried two approaches. First, I asked it to directly generate a summary table, but half the fields were missing.

Second, I asked it to convert to tables first then merge, but column names went haywire, mixing Chinese and English in one line, like encoding mismatch. I previously saw many Python, generative AI, and big data courses on ewant and almost signed up. But I had just finished overtime and really had no energy to learn a course from scratch; I just wanted to get this workflow running.

Should I have avoided letting it process PDFs and screenshots simultaneously? Has anyone encountered this? Do files in WorkBuddy need OCR first, or can field names not be in Chinese? Maybe I misunderstood; has anyone else faced this?