Launching today
untxt.
The shoebox of receipts, auto-sorted into clean data.
3 followers
The shoebox of receipts, auto-sorted into clean data.
3 followers
untxt. auto-turns messy receipts, invoices, bills, scans, phone photos, and PDFs into reviewable bookkeeping data: the cleanup before the books. Zero setup. Zero rules. Zero training.
Hey PH, Sebastian here, one half of untxt (Ben's the other).
I started this because I was asked to digitize a bookkeeping firm, and I quickly realized their productivity issues were relatively small compared to the real bottleneck: hand-typing receipts, invoices and every other document, simply because the error rate on existing automation tools was high while trust was low.
So we built the boring thing nobody wants to: the cleanup before the books. You throw in the messy receipts, invoices, bills, phone photos, PDFs, and untxt turns it into clean, sorted, reviewable data. No setup, no rules to configure, no model to train. We built it in reverse: instead of asking you what to extract, it shows you what's in there. It just works on the pile. Rather relentlessly ;)
What it's not: it's not trying to replace the tools you already use. It sits upstream and feeds them clean data. And it doesn't invent numbers. Every field traces back to the source document you uploaded. We went out of our way to make it un-magical on purpose: you can always see where a number came from. From dropping a document to getting clean, ready-to-use data, it's literally drag, click OK, click to review.
We'd love the brutal version of your feedback, especially: what would you throw at it to try and break it? Drop your ugliest receipt scenario below.