I built an AI that turns phone photos into organized data (not a wall of text)

I was the person at my company re-typing invoices. ~60 a week, half arriving as photos at every angle imaginable. Every tool I tried read about 70% and handed me the other 30% as "editing work", which is the whole job again.

So I built shenshen: it looks at a photo the way a person would, figures out what matters, and hands it back already organized. Vendor, amount, date, line items, totals done. Low-confidence items get flagged, so you only glance at those, not every field. Photo in, spreadsheet row out.

It's on Product Hunt today. Would genuinely appreciate your support and honest feedback:
https://www.producthunt.com/posts/shenshen

Two things I'd love to hear:

1. What document breaks your current tool? I'm building against real failures.
2. If this saves you a re-keying session, tell me where. It decides what I build next.

Free to start, no card. Try it on your messiest document.