Let's ERPNext - KI gestützt Eingangsrechnungen einlesen

In this video, we see the beta version of our AI-powered processing of incoming invoices and other documents. The video demonstrates how a variety of different document types are imported, which the AI ​​then interprets and correctly categorizes. At phamos, we believe that properly automated business processes must be transparent and designed so that human expertise can be applied at any time to improve the results. We've incorporated this functionality into this version. Users can continuously refine the system's prompts to make them more precise and capable of handling an increasing number of scenarios. The prompt design isn't always straightforward and constantly challenges users to express themselves very precisely. It's fortunate that the German language is an excellent vehicle for this! Currently, the process only demonstrates the import of data required for the next step: forwarding the documents to our external accounting department. We send our documents from the system to DATEV via email, from where our tax advisor then takes over. In a future version, we plan to identify the articles, taxes, and accounting entries in order to create entries in ERPNext that are either from the "Accounting Record" or "Incoming Invoice" doctype. It's also conceivable that order confirmations could be imported this way to generate orders, to which the corresponding incoming invoices could then be automatically assigned and reconciled. Further possible future scenarios include importing business cards into CRM/Sales, importing sick notes into HR, or importing handwritten work reports, for example, from construction sites. In schools, homework and classwork could also be imported. Notes from project meetings, and much more, could also be imported.