Hey everyone! I'm trying to find a solution to automate the processing of documents in our accounts team. We receive a lot of invoices through email, PDFs, and even some scanned copies. I'm hoping to find a free, self-hosted tool that can help us recognize these documents automatically and then store them for future filing. Ideally, it should identify key details like invoice numbers and lines so we can store that info in a database and integrate it with Sage. Any recommendations for a reliable solution? Thanks a bunch!
4 Answers
You might want to check out Paperless-ngx along with some OCR options that can be easily hosted using Docker. While it’s not a professional-grade solution, I’m using Paperless-ngx with paperless-ai on my homelab, and it works pretty well for personal use. Just keep in mind that it does require some tuning, especially with larger volumes, and ideally, you'd have a Postgres database rather than the default SQLite.
This sounds promising, but we don’t have any dedicated GPUs on our servers.
Paperless-ngx is definitely a good option! It’s free to self-host and can handle OCR for invoices, tagging them, and pushing metadata into a database. It's much easier than trying to develop a solution from scratch!
Considering that you don't have GPUs and will need to build the integration with Sage, have you thought about using Azure's invoice model? It’s about $10 for 1000 pages, which is cheaper than self-hosting if you don’t have a massive document volume. It generates a JSON with all the relevant invoice lines and supplier info, even with less-than-perfect invoices.
While I don't know of any free solutions, if you can manage to get the JSON of the invoice, you might still run into formatting issues when integrating with Sage. Just make sure there’s a review process to catch any mistakes before approving payments. It sounds like a project that needs solid management commitment to fully pull off.
Great setup! I’ve got a similar configuration, and it works wonders for handling my personal documents.