In “Technology trends and automation in the financial sector”, it was examined that the financial industry is adopting automation throughout to increase operational excellence. Specifically, AI-based operations to tackle some of the more complex processes and combining this with simpler automation tools for administrative procedures.

Intelligent reporting

In a world of big data, large amounts of information, and complex organizations, structure and reporting has become an imperative to successful management. Strenuous and time-consuming reporting processes are accepted and necessary for management to obtain transparency and make the right decisions.

Reporting is a means to understanding how the organization behaves. Intelligent analysis allows to monitor progress and identify improvement areas. Very often, these reports are still created manually by retrieving data or creating large excel dashboards. A combination of RPA and AI helps to automate statistical analyses, lower error rates in reporting, and avoid possible confirmation biases that may arise from choosing preferred datasets. By using Excel datasets as input (can be unstructured), Artificial Intelligence can deliver distinct statistical inference. First, the AI can provide anomaly detection, e.g., outlier/ discontinuity detection, time series analysis, correlations, fraud detection, etc. On a deeper level, the AI can give predictive analytics, i.e., trend detection, scenario analysis, and inclusion of macro-economic and industry-specific data. Lastly, it is even capable of analyzing root cause problems from the data. The AI examines and compares anomalies such as outliers and patterns from different statistical inquiries and then draws customer-specific inference for problem solving. Result presentations varies from classical management stories, real-time alerts, or can be fully customized.

Intelligent reporting AI RPA

Exhibit 1 above provides a comparison between manual and automated reporting procedures. The numbers are approximated and evaluated from our experience from past reporting automation projects and will, therefore, be different, yet similar, in other instances and organizations. We observed a process throughput time of 35 minutes to create the report based on an existing Excel file. In addition, monthly number of manual errors amounted to around ten (Severity of error not taken into account). After the process had been automated, process throughput time decreased by 86% to 5 minutes while the error rate was lower by a factor of around three, or by 70%. This meant that employees could spent 30 minutes of productive work more per time unit.

Invoice processing

There are many different sample processes connected to invoice management. It is only one instance that is presented here, however, invoice processing is one of the favorites to be automated by companies. Previously, employees validated vendor invoices manually. This kept them busy with administrative burdens and away from productive activities.

In an automated invoice process, vendors send their invoice to a specifically configured email address. The automation solution exports the document (e.g., PDF) from the attachments and transfers them into the OCR software. The OCR will then extract the relevant information from the invoice and trigger the invoice approval process within the workflow management software. The software will recognize if the vendor does not exist in the ERP system and will further initiate the vendor creation process if required. In case of a purchase order, approval follows automatically. Following this, a digital posting document is created and sent to the ERP through an interface system. Workflow management is then informed about the successful invoice posting. With this automation in place, service center employees can re-focus on other value-creating tasks instead of manually checking invoices, updating the status, and verifying if the service of the vendor was already executed. Previously, this required storing large amounts of Excel files where it now is replaced by direct information transfers and status updates.

Invoice processing RPA AI

Miscellaneous automations

There are endless opportunities, and therefore use cases to present, for automations in the financial sector. Here is a collection of miscellaneous finance automation use cases:

Expense appraisal:

Previously, auditors had to open and validate expenses individually. This manual effort can be minimized by using a combination of RPA, AI, and OCR technology. Employees enter and upload their expense information into an expense tool together with the respective receipt. The implemented automation is extracting the data and downloads the attached receipts for every single expense position. By using a standardized interface (Rest API), the automation sends the data, including the downloaded documents, to the OCR software. It will then analyze the documents and extract additional information from the receipt. All data, including the enriched data from the OCR, will be transferred to an Artificial Intelligence (AI) for examining the data for expense policy applicability. The AI decides autonomously upon the status of the expense position and updates the status in the expense tool. The employee receives a notification email (other notification options possible) with information regarding further conduct or approval of expense position. If approved, payment will be triggered and executed automatically. For a four-year period, this process has demonstrated an expected annualized ROI of 71%.

Refund of customer credits:

Refunds are recorded daily in a customer management system. Employees will than process these entries by booking them into the SAP system and balance the credit account in the customer management system. An RPA robot can initiate bookings in SAP automatically and,afterwards, balance the credit account with peripheral systems. The bot will also document all receipts and bookings while guaranteeing error-free processing using built-in tests and escalation steps. The client had immensely reduced processing time and administrative work by automating the refund process for customer credits with validation checks including complete documentation.

The future is now.

This article has presented only an excerpt of use cases in automation using RPA and other technologies. Similar to Moore’s law, there will be new ways to combine different technologies automating processes only humans were believed to be able to execute. For example, combining RPA and AI has already paved the path for new automations and has unleashed more efficient processes. Opportunity costs of not following this trend will increase tremendously over the next few years. The future for companies will be to have a completely digital and automated end-to-end process landscape.