<img src="https://ws.zoominfo.com/pixel/BUJfPb8NrEnpjSiz8kRz" width="1" height="1" style="display: none;">
Company Culture Artificial Intelligence

The Biggest Challenges to Implementing AI and a Robo-Auditing System

on October 18, 2018

If you’re a CFO looking to improve your company’s processes and save money by implementing a robo-auditor, you might be wondering where to start.

For companies of any size, the key is to seize on something readily achievable, so you can build momentum and attain tangible results in the first month or two.

This allows everyone to see how the process can bear fruit before the robo-auditor takes on advanced tasks that might take longer to roll out and show results.

Working with artificial intelligence is a new competency for your company. You want to get used to using it, see what it’s like, enjoy early success, and build trust in it.

As the trust and comfort increases, you can expand the scope of the robot’s duties.

In this article, we’ll look at one area where it’s smart to unleash a robo-auditor at first and the biggest challenge you’re going to face while implementing this system.

Where should new users start with a robo-auditor?

A great place to start using a robo-auditor is in expense reports.

Duplicate expenses are a black-and-white problem that’s usually the result of human error, but sometimes you’ll find that people are gaming the system.

If you can flag these issues and correct behavior, you’ll get an immediate ROI.

Down the road, you may also look for inefficiency in meal and hotel choices when employees travel, but that issue is typically too big to tackle in the first month because it requires difficult conversations and changed behavior before you’ll see returns.

You want your team to engage with the robo-auditor, which means not overwhelming them by taking on too much out of the gate. Get them started, get them successful, and then you can iterate, improve, and build on what they learn.

Most accounting departments won’t immediately abandon practices like sample-based audits, choosing instead to run parallel processes until it is comfortable with the robot and confident enough to stop some of the things they were doing in the past.

What’s the biggest challenge when implementing a robo-auditing system?

Many AI systems produce stunning reports and dashboards. That part can be magic and automated, but analytics are just the beginning of the process.

Your biggest challenge will be focusing on what you want to achieve with the insights produced by those analytics. What job are you trying to accomplish?

Your goal might be to minimize the risk in travel and entertainment spending or minimize the risk and optimize your use of cash in accounts payable.

Whatever your goal, analytics play a role, but they’re not the result itself.

Your system needs to incorporate both the inspection side and the resolution. While the inspection and analytics are automated, the resolution is not.

You’ll need an interface that incorporates both inspection and resolution, commonly called a workbench. Think of it like a case management system.

When a robo-auditor surfaces an issue, a good system will explain why it surfaced that issue in very plain, straightforward language, then provide the needed context.

For example, say someone ordered a boxcar of supplies, but previously the most your company ever ordered was a box of that item from the vendor.

Once this issue has been flagged, the next step is for someone to deal with it.

Your workbench should allow you to track responsibility and the reasons why something happened, which will help drive future process improvement.

When evaluating a system, consider the case-management side of the work. How will you respond to the analytics your system delivers? Answering that question solves the biggest challenge because that’s where most users will spend their time.

What else should users be aware of during a robo-auditor rollout?

As part of the rollout phase, the system needs the opportunity to learn about your organization by reviewing lots of historical data, usually six to nine months of historic transactions. During that process, it will find problems from the past.

Many people may think they’ve got to run those problems down, but that’s not always the best idea when trying to implement a new robo-auditing system.

It’s tempting to go after old problems, especially if there’s real dollars involved.

However, if your team hasn’t learned how to use the system, and the AI system hasn’t learned as much as it should, chasing past problems can lead to frustration and team members feeling overwhelmed because neither party is fully prepared.

A better approach is to focus on what to do in the next month rather than correcting what happened in the last. You can’t improve everything right away. Get a couple of quick, easy wins and continue to tune and refine the new system.

You’ll become more capable with your robotic process automation (RPA) system, and you’ll get comfortable abandoning old methods. Build momentum and get better each month. As tempting as it is to chase down the past, it can be a quagmire and can thwart a project launch.

You’re accustomed to chasing the dollars found in problematic transactions, but what you need to do is refine future behaviors, because that’s where real savings are.


This is the ninth blog in an ongoing series based on the recently released book, Robo Auditing: Using Artificial Intelligence to Optimize Corporate Finance Processes by Patrick J.D. Taylor, Manish Singh and Nathanael L’Heureux.

Nathanael L’Heureux

Chief Client Officer