What CFOs Need To Know About The Robo-Auditing Revolution

The following in an edited excerpt from the book, Robo-Auditing: Using Artificial Intelligence to Optimize Corporate Finance Processes, by Patrick Taylor.

When you mention robots, most people think of Star Wars or those car-manufacturing assembly lines where giant, agile machines move heavy parts, make spot welds, or secure bolts on chassis.

Robo-auditing isn’t quite as dramatic to watch, but the results can be just as astonishing.

In the world of finance, robo-auditing combines automation and artificial intelligence to spot trends that humans often overlook and efficiently address any problems. It augments what humans do and enhances their effectiveness, speed, and efficiency. It’s similar to the revolution we’ve seen in manufacturing robots, because the combination of automation and artificial intelligence in accounting is faster and more reliable than humans can ever be.

In finance, we’ve automated things for a long time. There are general ledgers, ERP systems, and spreadsheet programs like Excel.

But robo-auditing allows you to do more than save time by automating these operations. It allows you to redesign business processes to increase efficiency and effectiveness. You don’t just work faster; you work smarter.

Auto manufacturing is an example of how this should work. In the 1980s, American cars were not good quality, and the Japanese dominated the auto industry with less-expensive and higher-quality vehicles. At the time, the Japanese used a lot of robots in their manufacturing process, and Japanese carmakers were happy to let competitors tour their factories and see how they had automated so many steps.

American automakers came back and invested billions in robots to build their cars. The problem was that they automated the same work they had done before. They could make cars faster with fewer employees, but the vehicles were the same poor quality. They were not using these automation tools to redesign how they manufactured the cars. American automakers needed to refine and improve their work, not just automate an old approach that resulted in vehicles that broke down sooner than their Japanese counterparts.

Today, there is a lot more parity in the auto world, because automakers have learned to use robots to build a better product.

The same thing is happening in finance. Although your company’s accounting procedures are more straightforward than the quality challenges Chrysler, Ford, and GM faced back in the day, companies today find that artificial intelligence and automation help their financial people get their work done faster and more dependably than they could before these tools became available.

Helping CFOs Get The Job Done

The idea of using artificial intelligence to augment accounting and auditing procedures can sound intimidating if you equate AI with some of the more extreme ways it’s employed today, such as in the development of self-driving cars. Fortunately, accounting doesn’t offer as many challenging variables as designing a car to navigate the crowded highways around SanFrancisco.

But some CFOs are cautious for other reasons.

The IT landscape is littered with shiny, new technologies that didn’t deliver the ROI they promised, and this has made some chief financial officers reluctant to adopt the latest technological innovations. Most CFOs quickly get to this question: Will my employees utilize this product and be willing to change their behaviors to take full advantage of it? Many products can address a company’s real pains, but the challenge is adopting those products. CFOs want to know whether a solution can be integrated into their particular operation and help them achieve higher-value work.

I’m empathetic. When someone pitches me a product that promises to make my sales force more effective, it doesn’t take me long to ask the same question: Is this a good fit for our operation? Can I make that product work here?

In my business of selling subscription-based Software as a Service (SaaS), we need customers to succeed. If they aren’t successful, then we aren’t successful.

We depend on our customers renewing annually, and they won’t renew if our product doesn’t help them. This has taught us over the years to focus on the jobs that people are trying to get done. To go back to the auto industry analogy, we don’t sell robots to help our customers weld their cars faster. We sell robots to help them build better cars faster, so they can be more successful. My robot is there to help you get done what you want and need to get done.

Artificial intelligence financial systems help CFOs do their jobs better, increasing their companies’ profits while decreasing risk. As an example, we’ve seen across hundreds of robo-auditing implementations that companies can save an average of 5 percent just by analyzing travel expenses and educating employees to make smarter decisions. But that’s only one place where artificial intelligence pays off.

AI Does More Than Catch Mistakes

When you explain how artificial intelligence improves corporate accounting, the first thing people think is that they can use a robot auditor to catch mistakes, for example, in expense reports. Maybe most of those mistakes are human errors, maybe some are fraud, and it’s true that a robo-auditor will catch them.

But where the real money comes from is when CFOs, controllers, and CIOs use those audits to educate employees, so they don’t make mistakes in the first place.

Robo-auditing teaches employees to make smarter choices that stretch their travel budgets farther. Employees don’t waste money as much as they used to, and that’s what drives that 5 percent savings. It’s less about, “Oh, I’m not reimbursing this particular item,” and more about helping employees see the patterns in and implications of their behaviors, and driving financial impact by using analysis to improve company policy.

For example, say you have an account executive who travels to New York City. It’s December — the most expensive time of the year to visit New York — and your executive spends $500 a night for her hotel room. That’s what you’d expect to pay for a hotel around Christmas, and your robo-auditor doesn’t flag that trip as an anomaly.

But say your executive travels to New York a few months later, when hotel rates are typically lower, and again pays $500 a night. Your robo-auditor — armed with data from thousands of expense reports not only from your company but from many other companies that also use the service — knows that is a bit extravagant for a trip to New York in March.

It flags the expense and lets your executive know that the going rate for a quality hotel in Manhattan should be lower. The result is that in the future, the executive makes a better decision about where to stay and how much to spend. In this way, artificial intelligence steps in with the guidance an account executive needs to do her job more efficiently while saving her company money.

So, Why Are CFOs Cautious?

Most CFOs are comfortable with automated financial processes. They’ve used general-ledger packages, and they know how to use them to get the most out of their businesses.

They can utilize artificial intelligence the same way. Fortunately, many business accounting practices are the same from one company and one industry to another, which means that CFOs who want to adopt robo-auditing don’t need to build custom systems. They can buy packaged robo-auditors, just like they utilize packaged general-ledger systems, and then focus on smart ways to deploy them. They don’t need to invent anything.

The applications of artificial intelligence and automation go far beyond examining travel expenses. For instance, CFOs can userobo-auditing to optimize their accounts payable and accounts receivable processes.

Are you making erroneous payments to vendors? Are you paying them at the right time? You don’t want to pay too early and tie up your working capital. But sometimes, paying early means you can earn a discount, so it may be worth doing. Your robo-auditor can guide your operations to pay at the perfect time.

How a corporation collects from customers influences both cash flow and customer satisfaction. A CFO’s company may already do that well, but robo-auditing augments such expertise so that the company collects as expeditiously as possible while maintaining good customer relations. Robo-auditing protects your company’s reputation by preventing flawed cases and quickly spotting malfeasance.

Robo-auditing can also smooth over the rough spots created when new employees come on board. New staff members can be prone to mistakes as they learn about your operations, but the robo-auditor notices when someone is making a mistake and helps new employees identify a better decision.

The expertise built into the software means you don’t have to worry about lost knowledge when you lose an experienced staff member. The robo-auditor helps you maintain consistent performance.

Robo-auditing is not a moonshot for your company. Someone else has already taken the moonshot and figured out how to do it.

What’s available now to CFOs is a packaged process that’s practical and pragmatic. AI is improving, thanks to increased computing power that wasn’t available ten years ago. Robo-auditing using AI and automation has moved beyond “Do I need to take this risk?” and is at the point where the risk is in not adopting it because your competitors already have (or soon will).

Getting Comfortable With AI Systems

This book will describe best practices for putting robo-auditing to work at your company. Again, CFOs don’t need to invent anything; they just need to find the best way to apply robo-auditing to their operation and profit from it.

The best way for CFOs to get familiar with artificial intelligence is to start with the well-proven approaches that others have already automated successfully. Start with bite-sized chunks, employing robo-auditing on straightforward business processes like travel expenses or accounts payable. As you gain more experience, you can use it on more complex processes. This approach allows you to learn the system and identify the best ways to apply it in your operation.

Start with the common decisions and judgment calls that your people make. The payoff is higher when you automate processes that occur frequently rather than occasionally. You want to start where you can get a lot of repetitions quickly and get comfortable with the approach.

Accounts receivable are a good example.

Since accounts receivable are handled the same way across all business units, many companies unite those functions and address them through a shared-services structure.

A robot is a great way to make that centralized organization work more efficiently. You can buy prepackaged applications that have already been vetted by peer organizations who have developed best practices for applying the technology.

As a CFO, you can leverage that expertise and focus your efforts on how to roll the system out successfully.

What You’ll Gain From This Book

Investors know that they often have to take risks to earn a high reward. That is not the case with robo-auditing. As this book will show, the risk of robo-auditing not working at your company is low, and the returns are very high.

I also want this book to be a practical guide for how CFOs can make robo-auditing systems work at their companies.

I’ll show you how you can walk, then run, with these capabilities. We’ve learned a lot about how to use the technology — that’s what you buy our product for — but we’ve also learned a lot about the human and change-management aspects of adopting a capability like this. This book details the best practices that will help CFOs realize the system’s potential.

How I Got Here

I got my undergraduate degree in mechanical engineering from the Georgia Institute of Technology in 1986. It was a traditional engineering degree, and one of the lasting things I learned had to do with closed-loop control systems.

That’s the idea behind designing a system to monitor and control a process by measuring its output against your objectives and making adjustments to the inputs to keep the system on track.

This process allows you to compare the actual result to the desired outcome in a way that enables you to reduce errors. If the system is disturbed, you can use the feedback to bring the output back to original levels. If you measure your output and compare it with what you wanted it to be, you can make adjustments to eliminate the discrepancies.

A traffic light is an example of this. One objective is to minimize unnecessary wait times. Nothing is more annoying than sitting at a red light when there is no cross traffic.

An open-loop-control traffic signal is set on a timer: it’s green for two minutes, then briefly yellow, then red for two minutes. Thankfully, you don’t see those much anymore, because most lights today are on a closed-loop-control system. They have sensors that tell them when a car pulls up. The signal will be set primarily on green for the busy road until a car pulls up on the less-busy street, and then the light changes.

More sophisticated sensors can tell when cars line up on one avenue, or when there is a line of cars waiting to turn left. All these are examples of taking feedback from the environment and adapting the system to achieve the objective of keeping traffic moving.

Two years after graduating from Georgia Tech, I got accepted into the MBA program at Harvard Business School, where I took a course by James Cash called Management Information Systems. Cash taught us that the real computer revolution would come when we used computers to change how we worked as opposed to automating the work we were already doing. I wanted to be part of that revolution, so I embarked on a career in technology.

At the time, Oracle CEO Larry Ellison decided to hire more strategic sales people and ordered his managers to hire fifty newly minted MBAs with experience in the information-technology business. (I didn’t qualify on the second count, but after weeks of persistent follow-up, they gave me a shot, anyway.) I worked with relational databases, which at the time were about doing analysis to make smarter decisions, and then went to work for Symantec. In 1996, I moved from Silicon Valley to Atlanta to work for a small company called Internet Security Systems.

Our software measured how well the security functions of a network of computers worked. We would know how a system should work, and what activity we should see, and compare that to what was happening. This process allowed you to spot any activity that wasn’t supposed to happen and alert you to the presence of hackers that may have gotten into the system. We applied closed-loop controls to information security. I joined the company as employee Number Twelve and left about five years later when we were fourteen hundred people, and the company had gone public. IBM bought the company for $1.6 billion in 2006.

I co-founded Oversight Systems, Inc. in 2003 with the idea of applying closed-loop control to the world of finance. Although companies lose a lot of money to hackers, they lose even more through internal fraud and waste. We were fortunate to hook up with Dan Kuokka, who has a PhD in Artificial Intelligence from Carnegie-Mellon, and develop what has become Oversight Insights On Demand, which currently provides solutions for Travel and Expense, Purchase Cards, and Procure-to-Pay, and will be expanding to additional business processes in the future.

At the most basic level, these systems find problematic transactions. But more importantly, they allow managers to zero in on the people who are causing the problems. There is usually a small number of actors who commit most of a company’s errors, and when you identify those behaviors, you can set up process improvements to minimize the damage. Instead of chasing after and extinguishing a series of small fires, you can make process improvements and people improvements to keep those fires from starting in the first place.

Our clients include many Fortune 500 technology companies. These are companies that build their own artificial intelligence systems for product development and research, but they use our turnkey financial AI systems because this makes more sense from a time, effort, and effectiveness standpoint. We relentlessly work on problems in finance and accounting because those are the most essential things in the world to us. Our clients didn’t invent their own financial AI systems, because we already had a tested and proven one. This allows them to put their researchers on more strategic projects. There was no reason for them to reinvent the wheel when it came to automating financial processes.

As we matured as a software-as-a-service business, we focused on how to make these tools work best for individual companies. We’ve deployed more robot auditors than anyone on the planet, and, as a result, we’ve learned a lot about how to best use AI in finance processes. Often in the technology field, you tell potential customers, “Hey, we’ve got this great set of capabilities that can do a million things; what would you like it to do?” But we’ve learned that the best approach is simply to tell people, “This is how you should do it.”

That’s not meant to sound arrogant. It’s just that we’ve developed the best practices already.

The purpose of this book is to share these best practices with you. On these pages, you’ll see how to make robo-auditing, with its combination of automation and artificial intelligence, a success in your organization.

Are you ready to make that happen?

For more on why and how to implement artificial intelligence systems into your financial operations, pick up Robo-Auditing: Using Artificial Intelligence to Optimize Corporate Finance Processes, by Patrick Taylor.