vendorrecon
VendorRecon is a lightweight, hyper-focused B2B SaaS.
Bookkeepers spend hours every month matching supplier statements against the AP ledger line by line — the literal yellow-highlighter job. I built a tool that does it automatically: drop in the statement + a ledger export, and it returns the matched lines plus the exceptions that matter — missing invoices, amount mismatches, unapplied credits — in about a minute.
Here's my problem: I built the whole engine and have only tested it on AI-generated statements. I have no idea how it holds up against the beautiful mess of real-world PDFs (every vendor formats them differently, half are basically scanned).
So I'm looking for two things:
Anyone who's done this manually — does the output above sound like it'd actually save you time, or am I missing the real pain?
If you've got a sample/anonymized supplier statement lying around, I'd love to run it and show you what comes back. No signup — just email me the file, I run it, I send the report. (I delete the file after.)
Brutal feedback very welcome. [link]
3 Comments
This is a strong problem to attack, but I’d be careful with how you validate it.
The risk is not only whether the parser survives messy PDFs. The bigger risk is testing on random statements and thinking that proves the workflow is useful.
For bookkeepers, the real value is probably in whether the exceptions are trustworthy enough to change how they review the month-end work.
I would not solve that loosely in the thread because the first real-world test needs to prove the workflow, not just the extraction.
If you’re open to it, share your email and I’ll put the tighter validation path together properly.
now i am searching for founding design partners to test my algorithm on real data. vendorreconorg@gmail
Sent you a note by email. Main thing is making sure design partners validate the bookkeeping workflow, not just the parser.