Hyper-automation success story: supplier invoicing

The hyper-automation ecosystem is taking over the layer that makes sense of all the others: Process mining and advanced analytics to process the magma of data and decant the knowledge that allows a macro perspective, integrated with the rest of the company's processes.

This involves understanding one's own behaviour patterns and those of suppliers and customers, long-term stock needs, comparing services and supplies from history, anticipating incidents in order to carry out preventive maintenance or forecasting liquidity against foreseeable expenses so as not to improvise costly financing, among many other keys that help people to make the right management decisions. It is a matter of understanding the present through the past in order to plan for the future.

Hyper-automated invoicing works on levels that interrelate the parts of the process:

  • At first, RPA (Robotic Process Automation) software robots automate e-mail communications with thousands of suppliers to send and receive invoices.


  • The software extracts raw data using different OCR (Optical Character Recognition) engines to convert data to text. Subsequently, the robots analyse the content through Deep and machine learning, which extracts the information organised into "attributes" or basic data, such as date, order number, VAT, amount, tax ID number of the supplier, tax ID number of the Prosegur company making the payment, etc.


  • Machine learning technology can detect possible errors (e.g. missing a digit in a Tax ID number). AI engines assign each extracted data an associated confidence level which, when it falls below the 90% threshold, is notified to the responsible team for revision or completion. If it is not sufficient because it is incomplete, try to fix it or refer the alert to the Prosegur Administrative Centre (CAAP) so that the supplier can be notified, corrected and the payment can continue. In other words, the client, Prosegur, provides the proactivity that allows the supplier to be paid immediately instead of waiting weeks after the absence of income is detected, understands the cause and remedies the situation.


  • The next layer, these documents and the huge amount of data (in Spain and Portugal, more than 250,000 invoices are processed annually) are transferred to our own digital content repository SGD (Document Management System), which safeguards, organises and simplifies access. It already accumulates more than 40 million documents in digital format, grows steadily, feeds the machine learning layer and expands its capabilities with proprietary tools such as digital signature (Segursign). Prosegur saves money by handling this administrative management in-house, which on such a scale is usually outsourced. 


  • Another key technology that in itself introduces automatisms: ERP (enterprise resource planning), a sort of centralised process control. It standardises your communication with the DMS and therefore makes it seamless, confirms invoice validation, authorises payment and completes the entire invoice registration and posting. It also feeds back the machine learning and deep learning layer with the incidents resolved in the control tower and it does not matter if Prosegur changes the commercial version of ERP, as it standardises communication through the use of APIs.


Step by step, layer by layer


What now?

Hyper-automation is as simple to understand as it is complex to implement. In a single comprehensive model, it combines two or more technologies that add value when automating business processes. Prosegur's proprietary formula now includes not two, but multiple technologies, some even developed in-house. There is no ceiling to its development in the age of accelerated digitisation, with new use cases that respond to the changing needs of the company, its customers and the market.

Few global organisations have developed hyper-automation to the level of Prosegur, which already applies it in processes such as supplier invoicing in Spain and Portugal. Its objectives: serve as a pilot for other areas, streamline operations, boost efficiency, minimise errors, save time and money, meet more demanding service quality agreements, improve customer relations, free people from tedious tasks such as transcribing millions of pieces of data to devote them to other more valuable tasks.