AI
May 18, 2015

Using AI to automate expense report audits

Kunal Verma

Artificial Intelligence and Big Data is now being used by companies to combat expense report fraud and lower risk of expensive compliance litigation.

In today's world, there is a lot of pressure on CFOs and Finance Controllers to not only comply with internal spend control policies, but also ensure ethical practices and comply with a number of laws such as Foreign Corrupt Practices Act (FCPA).  Relying solely on human auditors to accomplish these tasks is very risky.

Human auditors are given the gargantuan task of manually auditing hundreds of expense reports per week and this leads to them looking at expense reports above arbitrary thresholds and miss many of the fraud and compliance issues.

AppZen has developed a unique solution that uses big data and artificial intelligence to automate expense report audit, and has deployed it at several of its large enterprise clients.

Understanding and interpreting expense data is key to automating the research and reasoning that human auditors do. The audit process involves looking at multiple sources of data, web search and even the history of previous expenses. AppZen leverages artificial intelligence and big data to automate expense report audit, augmenting human audit tasks by automatically integrating structured and unstructured expense data from multiple sources, performing semantic and statistical cal analyses, and then assigning a risk score for each expense.

AppZen learns from customers' data and can also be configured by adding additional business rules. Some examples of the kinds of issues detected by the AppZen audit engine can be found here.

1: Statistical Analysis

AppZen's Audit Engine leverages statistical analysis to build statistical models of client spend and expenses, merchant statistical signatures, etc. and leverages them for anomaly detection and statistical inference.

2: Semantic Analysis

AppZen's Audit Engine leverages in-house ontologies linked to Web Scale Data from sources such as such as DBPedia (structured representation of WikiPedia) to semantically intrepret and classify expense data. 

3: Reasoning

AppZen's Audit Engine leverages in-house built rule-based and probabilistic reasoning engines for reasoning on clients business and audit rules.

4: Unstructured Data Analysis

AppZen's Audit Engine leverages an in-house built unstructured data engine that allows seamless extracting and integrating of data from multiple sources including user's justifications for expenses, itineraries, receipts, calendars and business rules/policies.