Understanding Artificial Intelligence in Accounts Payable

Imagine if your Accounts Payable (AP) department could break free from the manual, repetitive tasks that bog its staff down, and have all the information at your fingertips to make informed decisions.

That’s the promise of artificial intelligence (AI) – technology that improves efficiency, accuracy, and decision-making across the invoice-to-pay process, transforming the AP function into a strategic driver. 

AP leaders are excited about the possibilities with AI.    

60% of AP departments plan to deploy AI by year’s end, Capgemini reports. What’s more, the global market for AI in AP is expected to reach $1.5 billion by 2025, MarketsandMarkets reports.

By eliminating manual tasks and improving decision-making with AI, AP departments can overcome their biggest challenges and reduce costs, improve accuracy, accelerate cycle times, and mitigate risk.

This article explores how AI works and the benefits it provides in AP.

What is artificial intelligence?

Few technologies have generated more excitement and concern than AI. 

AI uses machines to recognize patterns, make decisions, solve problems, and perform other tasks that typically require human intelligence. AI accomplishes this by processing and analyzing large amounts of data, drawing conclusions, and adapting its behavior based on the information it receives.      

Importantly, AI systems can learn from examples to deliver better results over time. Once trained, AI systems can capture and classify data, make decisions, and perform other tasks fast and accurately.

AI works through a combination of evolving techniques and technologies, including:

• Machine learning: a subset of AI that involves training sophisticated algorithms to learn patterns and make predictions or decisions based on large sets of historical data.  

• Neural networks: a type of machine learning algorithm that uses interconnected nodes that processes information and then passes the results on for further processing. Also called “deep learning,” this approach is effective for tasks such as image and speed recognition.     

• Natural language processing: a branch of AI that focuses on empowering machines to understand, interpret, and generate human language for applications such as chatbots.

• Cognitive computing: AI technology that can understand context and reason and make decisions (such as how to route documents), even when data is ambiguous or incomplete. 

AI should not be confused with robotic process automation (RPA), which focuses on automating rules-based, repetitive tasks such as uploading invoices into a system of record. AI also encompasses a much broader range of capabilities than optical character recognition (OCR), which specifically focuses on text recognition and extraction. However, AI, RPA and OCR can be used together strategically to provide a comprehensive solution for improving efficiency and decision-making.

AI’s ability to simulate human intelligence and decision-making is a game-changer in AP. 

Biggest AP challenges

The way that most AP departments process invoices and pay suppliers leaves a lot to be desired.

• Cost. The average cost of manually processing an invoice is $15, according to Aberdeen Group.  These high costs are directly attributable to the manual handling required to process, approve, and post invoices submitted by suppliers. According to the Institute of Finance and Management (IOFM), the typical AP professional spends 84% of their workday on manual, repetitive tasks such as keying invoice data, shuffling paper and emails, chasing down information, fixing errors and mistakes, and responding to calls and emails from suppliers about the status of invoices and payments. For their part, AP managers spend more time on transaction processing than on the managerial tasks they were hired to perform.

• Errors. Nearly 40% of the invoices that AP departments receive contain errors, IOFM reports.  Resolving a single invoice error can take days or weeks of back-and-forth emails and phone calls between internal stakeholders and suppliers, jeopardizing on-time payment. And a single type or transposed number from manually keyed invoice data can result in an incorrect payment to a supplier and additional rework and expense to resolve it.

• Delays. The average time to process a single invoice is 14.6 days, Aberdeen Group finds. Slow invoice approvals result in late payment penalties, missed opportunities to capture early payment discounts, more emails and phone calls from suppliers, difficulty forecasting cash, and potential supply chain issues. Suppliers are also less likely to provide optimal pricing or Service Level Agreements (SLAs) to buyers who frequently pay their invoices late.

• Risk. 25% of AP leaders say that the higher risk of fraud and compliance issues is the biggest challenge their department faces in today’s remote work environment, according to IOFM. It’s been difficult for many AP departments to adapt manual and semi-automated processes for remote working. Many departments rely heavily on unsecure emails to on board suppliers and approve invoices. Email doesn’t enforce segregation of duties, doesn’t provide chain of custody assurance, doesn’t track actions taken on an invoice, and cannot stop someone from deleting an invoice ahead of the organization’s retention schedule. And bad actors are using Business Email Compromise (BEC) attacks and phishing schemes to trick AP departments into sending electronic payments to a fraudulent bank account that they control.     

These are some of the reasons that more AP departments are deploying AI-powered systems.

Benefits of AI in AP

AI has the potential to significantly improve the efficiency and effectiveness of AP processes. 

AI can help AP departments reduce their costs by 30%, according to Aberdeen Group. And Forrester says AI can speed up the time it takes to approve an invoice by up to 80%.

Here are some of the ways that AP departments are using AI-powered systems:

• Invoice data capture. The combination of AI and OCR technologies improves the accuracy of header and line-item data extracted from invoices. Through contextual analysis and other techniques, typos and transposed numbers are less likely to slip through the cracks. In fact, Garter says AI can help AP departments improve their accuracy by up to 99%. AI also can learn how invoices are coded in the general ledger to minimize the need for keying.  

• Invoice matching. AI algorithms can accurately match invoices to POs and proof-of-delivery documents – even in cases where invoice line items are split across multiple POs. Analyzing historical data helps AI spot line-item errors and reduce the risk of overpayments.

• Invoice approval routing. AI systems learn an organization’s business rules and past actions taken on an invoice to accurately predict how specific invoices should be routed for approval.  

• Payment decisioning. AI systems can analyze historical payment data and databases of suppliers that accept digital payments to determine the optimal method for paying a supplier. 

• Fraud detection. AI systems can analyze historical payment data and inspect invoices for anomalies to identify fake invoices, duplicate payment, or unusual vendor behavior.

• Predictive analytics. AI systems can analyze historical invoice data to predict cash flow and corporate spending trends, to aid in optimizing payment schedules and managing liquidity. AI-powered systems also can monitor operations to alert managers of potential issues.

With a clear implementation plan, proper employee training, and continuous monitoring and refinement of the AI system, AI can deliver benefits across the invoice-to-pay lifecycle. 

Conclusion

Incorporating AI into your AP processes is more than a technology upgrade – it’s a way to elevate the AP function. By harnessing AI’s ability to automate tasks, deliver insights, and speed decision-making, AP leaders can overcome their biggest operational challenges and drive strategic objectives.