Reading documents automatically instead of typing them out

The greater part of the document load in medium-sized companies does not consist of difficult decisions but of transfer: what is written on a piece of paper has to go into a system. For that, people sit at screens and type out what a program could read. What is technically possible today has shifted considerably in the last two years. Even poorly scanned, inconsistent and handwritten-annotated documents can be processed — something that previously failed on rigid templates. Situations we encounter regularly. In no particular order — each case stands on its own. Invoices arrive by email, by post and as a PDF in the customer portal. Someone reads out the amount, tax rate, supplier and order reference and enters them. The documents are read out automatically, line items are matched to orders and a posting proposal is generated. Approval remains with the person. The proposal must be traceable. Anyone who cannot see where a value comes from will, in case of doubt, check everything twice — and then nothing has been gained. Delivery notes are filed and only looked out again when there is a complaint. Discrepancies between order and delivery are noticed late. Delivery notes are captured on goods receipt, line items are checked against the order and discrepancies are reported immediately. Capture where the goods arrive — not two days later in the office. Otherwise the process is bypassed. In regulated areas large volumes of records, certificates and evidence are produced. In audits or complaints a particular document must be findable within a short time. Documents are read out on arrival, indexed with batch, order and article references and stored in a searchable form. Traceability. In regulated environments it must be possible to demonstrate why a document was assigned as it was — and the process has to fit the existing quality management documentation.