High-end yoga apparel purchases and private helicopter rides are two charges one might expect executives to make on their personal credit cards, but a recent report from software company AppZen found that employees have actually tried to expense each to their companies. Such expenditures share something aside from dubious professional legitimacy: They were uncovered by artificial intelligence (AI). AppZen’s enterprise clients shared anonymized data on thousands of expense reports, and the firm used its AI-powered platform to uncover expenses that deviated from company policies.
AI and machine learning (ML) minimize fraud and lost funds, highlighting expenses that slip past manual reporting services. The report found that enterprise companies were able to review just 2 percent to 10 percent of the reports they collected without the help of AI, yet the software flagged nearly 9 percent of all employee-submitted expenses, drawing financial managers’ attention to potential abuses. AI also allowed employees to more quickly evaluate these reports, saving time and energy. Manual processing can cost companies approximately $7.50 per individual invoice, but automation can reduce this to $1.25.
The information-heavy nature of expense reports can also deter workers from requesting reimbursements, however. Many employees will simply absorb the costs of a $20 lunch rather than rifle through stacks of paperwork for the necessary documents. This can be detrimental to their finances and workplace productivity, and can also stoke needless frustrations with employers.
Automating this process can drastically reduce the hours employees must spend to complete their expense reports — a benefit for all involved parties. One company even claims that automation has decreased its workers’ time on these tasks by 90 percent since 2011. AI- and ML-powered reporting solutions ferret out fraudulent or noncompliant expenses for employers, in addition to drastically cutting employees’ time and effort in submitting reports.
High-risk or potentially fraudulent expenses represent money lost for companies regardless of employees’ intentions, and the cost of manually reviewing and processing these documents and invoices — and covering employees’ and contractors’ bills when they fail to make cost-effective purchasing decisions — remains high for those who have not upgraded their software to include such tools.
AI Expense Reporting Goes Mobile
Automated solutions both reduce costs and teach employees which expenses are acceptable. Private helicopter trips are an extreme example, but they reflect the complexity of expense guidelines and reporting, something companies and their workers have struggled with for decades.
Expense budgets can be messy. Construction crews might need to hastily purchase tarps and other supplies to secure sites ahead of approaching storms, for example, or event sales teams may need to furnish extra seating for last-minute guest list additions. More serious emergencies like natural disasters may occur as well, leaving less documentation as to where the money went. All such scenarios can challenge workers who are tasked with ensuring that expensed items fall within their companies’ internal guidelines.
Such employees must wade through giant binders documenting their firms’ processes to understand what can be expensed — an often confusing and frustrating experience for both them and their financial managers. AI- and ML-based solutions seem tailored to solve this problem, neatly categorizing receipts and creating equally simple reports for financial managers to read.
Automation can also save these firms money: One study claims it can save approximately $23 per report. These large savings especially impact smaller firms that lack the budgets for large financial management teams.
AI and ML can also be applied to tools like mobile apps, which can support workers in adding necessary, in-the-moment information to their reports.
Employees can use specialized smartphone apps to take pictures of their receipts and extract information such as location and time for expense reporting. One software-as-a-service (SaaS) solutions provider is even incorporating AI-based technology capable of reading handwritten receipts, exceeding the expected technological limits where automated algorithms simply sort through documents. The program can read through and categorize captured receipts without manual input, previewing how AI could work for all expenses in the future. The technology could save human employees approximately 10,000 hours of work, according to one analysis, providing significant — though yet unquantified — financial savings. Just hastening the process of loading paperwork into firms’ databases can significantly impact costs.
However, while they are already proving to be useful as expense tools, AI- and ML-based software still needs to be trained. Both must be taught to recognize important data on receipts or financial documents and ignore useless information.
AI and Its SMB Benefits
Automated technologies can especially benefit SMBs, which tend to rely on human employees’ manual efforts to sort through trips and financial documents. Humans cannot find patterns as quickly as AI, leading to incorrect expense reimbursement, among other problems. Such errors account for 14 percent of corporate fraud, for example, motivating companies to tap automated tools with superior pattern recognition.
Firms are releasing AI-based tools to help these entities: U.S. Bank unveiled such an app in May of 2019 that, when used in conjunction with its virtual card offerings, enables users to document expenses with their phones. An AI-enabled assistant asks questions to guide workers through the report-filing process, confirming lunch meeting attendance and flagging purchases that exceed approved limits or violate company policies.
And AI- and ML-powered spend management innovations can do more than improve efficiency and compliance — they can profoundly build worker satisfaction by making expense reporting more cost-effective, personalized and satisfying. This includes simplifying experiences before expenses are even reported. Firms could use AI to time airline ticket purchases by learning frequent travelers’ preferences and making recommendations for the lowest possible prices, for example. Such solutions could save finance managers up to 530 hours annually, according to one report on the U.K.’s expense industry, with yearly savings reaching approximately $31,653 just from reducing the time spent working.
Automation means business owners and accountants can allocate less time to reviewing and processing expense reports while knowing that AI-powered systems will flag potentially problematic transactions. The technology can also remove looming paperwork from the picture, helping workers focus on their assignments in the field.