Invofox Self Serve

Id love to see some real world benchmarks comparing extraction accuracy across invoices from different countries and formats.

The validation and feedback side caught my attention. Getting reliable structured data is usually harder than simply reading the document.

The API first approach makes a lot of sense here. Document parsing sounds simple until you have to deal with different layouts and inconsistent fields. Nice solution.

The concept of consolidating the entire document extraction pipeline into a single API is excellent.

Thanks for opening this up to small teams, haha... Booking a sales call just to try something is annoying, so 500 free pages with no card is an easy way to start.

Putting a guaranteed SLA and a no pay on mistakes rule behind document extraction is a confident move, and it is the pricing model that made me look twice. Learning from feedback on the specific document types a business actually handles should help accuracy compound over time. Does the model get per customer fine tuned or does it stay shared?

Great idea :) Can I use it locally for sensitive information?

Great idea! Am I allowed to create my own schema for different types of documents or select predefined schemas such as invoices, receipts, and bank statements?Additionally, Congratulations @nachogabaldon , @alberto_gimeno & team Invofox Self Serve ✌️🚀

@alberto_gimeno How well does Invofox handle Messy or low quality scanned documents, that's usually where document parsing gets really difficult.

Hey Product Hunt 👋We started Invofox because we kept watching good engineering teams lose entire quarters to document processing. The weekend prototype always works. Then come the scanned faxes, the multi-document PDFs, the tables that span three pages, the tax IDs that only appear in the footer, and suddenly two engineers are maintaining an OCR pipeline instead of building the product.So we built the pipeline once, properly: intake, dual-pass OCR, splitting, classification, extraction, cross-field validation, confidence scoring, provenance, and a feedback loop that keeps improving on your specific documents. You POST a file to one endpoint and get validated JSON back.For the last few years that only existed behind an enterprise contract. We were heads-down serving some of the largest tech companies and enterprises in Europe and the US, and honestly, we never built a way for a small team to just try it.That always felt wrong.Today that changes. Invofox Self-Serve is the same pipeline and the same accuracy bar, available with a signup form instead of a sales call. Two things we care about most: accuracy is part of our SLAs, and any document where we make a mistake is completely free. Pricing is per page, it goes down with volume, and there are no credit systems to decode.For Product Hunt: the free tier is 500 pages with no credit card, and we're adding 1,000 extra pages on top for anyone who signs up from here.Bring us your worst document. The one that broke your last parser. We'd genuinely like to see it, and we'll be in the comments all day.