Many people underestimate the role of the accounts payable department, often in charge of large sums of money. In theory, the process is relatively simple: appoint third parties to carry out particular tasks, receive their invoice, check it and pay. Therefore, it may surprise you to learn 81% of companies globally have suffered as a consequence of payment fraud. The average loss is around 5% of annual turnover.
Mitigating financial losses
Before we look at how artificial intelligence (AI) can assist with accounts payable automation, it is crucial to consider the main areas of concern. The vast majority of financial losses are a result of:
- Duplicate payments
- Overpayment and overbilling
- Fake and fraudulent invoices
- Payments sent to an incorrect account
The natural assumption is that such financial losses occur due to fraudulent activity when this is not necessarily the case. Some of the largest companies in the world have registered significant financial losses as a consequence of human error. Thankfully, automated solutions have emerged in the shape of Digital Coworkers, smart bots that incorporate cognitive solutions.
Data Extraction And Data Entry
In the past, many data extraction and data entry departments were effectively transferred to data entry outsourcing companies to save money. These companies employed operators to extract and enter data manually. While using companies based in countries such as the Philippines/India may have been cost-effective, it did not alleviate human error. Data protection has also become a political topic in recent years, with huge fines handed out to those placing client data at risk.
We have seen great strides in data extraction tools over the last few years. The following technologies will become more commonplace in the short to medium-term:
- Automated Character Recognition (ACR)
- Optical Character Recognition (OCR)
- Intelligent Character Recognition (ICR)
We can create a Digital Coworker, referred to as an AI bot, based on your specific requirements here at Roots Automation. Effectively, we can build the perfect employee, carrying out an array of mundane tasks 24/7. Our Digital Coworkers have a 95% straight-through processing rate that perfectly illustrates the colossal improvement in accuracy – especially compared to the 10% to 30% human error rate.
So, within just six to eight weeks, you can have your own personalized Digital Coworker. The AI bot can scan invoices for the relevant information, extract them, and automatically enter them into the appropriate system. This is just one area where reduced human input has slashed entry errors and fraudulent activity.
See also: Data Extraction Software vs. Data Entry Outsourcing Companies
Data Organization
Data entry heavy organizations are now making full use of OCR to lift the required data from physical invoices. This, together with Natural Language Processing (NLP), means that Digital Coworkers can lift the relevant information and understand it from a human perspective. Consequently, all data is passed to the relevant departments and automatically transferred to the appropriate computer system.
Collating documentation and information is an integral part of any accounts payable department. Our Digital Coworkers can also recognize when data is missing, incomplete or incorrect and contact the appropriate parties for confirmation. These "monotonous" activities are a vital part of the data organization process. The use of automated solutions such as Digital Coworkers has slashed processing times and effectively eliminated human error.
There may be situations where human assistance is required, perhaps borderline decision-making, but these tend to be few and far between. Consequently, many companies can retrain and transfer a raft of employees to other departments.
Fraud detection
Those who think that fraudulent activity is under control might like to review the Google and Facebook case of 2019. This involved a Lithuanian man and co-conspirators who set up a fake company with a similar name to a Taiwan-based hardware manufacturer. Using forged invoices, contact letters and creating general confusion, they managed to relieve Google and Facebook of a combined $100 million. This was not a one-off payment; they presented fraudulent invoices between 2013 and 2015. So, what can an AI bot do to check for fraudulent activity?
Digital Coworkers will automatically compare and contrast invoices against contracts. As part of the ongoing learning process, they will also remember how each counterpart traditionally receives payment. If there are any anomalies in invoices and contracts, unexpected changes in payment methods, these will be flagged. It is essential to maintain a fluid payment process within the accounts payable department but with one eye on potentially fraudulent activity.
Submit payments
While Robotic Process Automation (RPA) has proved very useful, this tool only works within rigid parameters. It will simply extract information from invoices and arrange payment. The same cannot be said of Digital Coworkers; they are set up to cross-reference invoices submitted against contracts. If there are anomalies such as payment method, invoice value, or even discrepancies with the typeface, these will be flagged and payment suspended.
Until you see a Digital Coworker in action, it isn't easy to comprehend the depth of behind-the-scenes checks that are carried out. In simple terms, the speed at which these AI bots work is the equivalent of between four and eight employees, and they work 24 hours a day, seven days a week. Some of the potentially fraudulent payments may turn out to be perfectly valid, but any anomalies will be noted and passed on to administrators to carry out further checks.
Work without IT’s involvement
Unlike many of our competitors, here at Roots Automation, we undertake the in-house building and training of Digital Coworkers around your specific processes. There is no need for your IT department to get involved in the training process or once the Digital Coworker “joins the workforce.” Traditional employment interviews can be like searching for a needle in a haystack, looking for that perfect employee. When building a Digital Coworker, you tell us the processes you wish to carry out, the systems currently in place, and we do the rest.
We have a comprehensive library of different tasks which you can add to any Digital Coworker. In effect, you tell us the characteristics you require, and we create your new employee. The process takes 6 to 8 weeks from start to finish, but that is not the end of the story. The use of AI ensures that your Digital Coworkers learn from new experiences and adapt all the time. This means that the efficiency and cost savings are never-ending, allowing you to concentrate on decision-making and client-facing activities.
See also: How to Best Prepare Your Human Employees for Automation
Confidence is the key
While it has taken some time for businesses to embrace AI fully, there is no doubt that confidence has been the key. Historically, data extraction and data entry processes were inaccurate and unreliable. Unless the information was clear and in the correct place on the form, the automated process could not cope. Digital Coworkers use systems such as OCR to decipher even the most adventurous handwriting. So rather than waiting for information to be handed to them on a plate, your modern-day Digital Coworker will go looking for that information.
Even though RPA has proven helpful in the past, it only works within clearly defined processing structures. Any changes will require reprogramming of the automated system, which can be expensive and time-consuming. The AI element of a Digital Coworker allows them to adapt to any changes in the process without outside assistance.
The speed of a Digital Coworker is impressive, the accuracy second to none, but for many, it is the ability to keep learning and adapting which is the most remarkable characteristic.