How AI is Disrupting Audit and Introduction to Ai Auditor
In this webinar you will learn:
- The history of AI and why it is emerging now.
- Challenges and shortcomings of the current audit process.
- How AI is disrupting audit.
- How the MindBridge Ai Auditor platform leverages AI and machine learning to help you audit better and faster than ever before.
Followed by a demonstration of the Ai Auditor platform.
A little bit about myself, I'm John Colthart, I'm the VP of Growth over at MindBridge. I've spent a lot of my career in finance or analytics relating to finance, and I've spent a lot of time in the last year and a half working with MindBridge to help share the knowledge of how AI can impact a profession.
Let's first start talking about AI and what AI is all about. I know it's this wonderful big broad term, but let's talk about what it really is and why it's so important to the world at this stage and with all the hype. Right now, when you look at everything that happens within a business, there's lots of data being piped through the various systems and subsystems.
And 68% of CEOs believe that data and analytics are going to generate the greatest return on stakeholder value. So when you let that sink in for a bit, you start to realize just the importance of having, whether you're a CPA working in an audit firm, working for clients, like CEOs and CFOs, or whether you're in the organization working for the CEO or CFO, you start realizing just how important it is to make sure that the data itself is correct and accurate.
Because if 68% of CEOs believe that data and analytics is the key to generating the highest return, then we need to make sure that the data is correct and accurate, and we need to give analytical tools that drive that type of ability to gain the insights that everyone is looking for. At the end of the day...there's a statement out there that says that we're at the end of what's considered tech companies.
There's a real blurring between what a tech company is versus a company. You can take the likes of the five largest companies today who are considered tech companies and you can notice that two of them aren't really tech companies at all, especially one that stands out being Amazon. Now they have an amazing plethora of technology that services their customers.
But really they started that with selling things like books and they've driven into bricks and mortar stores and they've managed to build an ecosystem of technology that is around the delivery to the customer. You take other ends of the landscape and you take a look at Apple who has always been considered one of the largest tech companies.
But look at how they're changing the user experience paradigm and have over the years especially in the last decade from the release of the iPhone through to the iPad and having all these competitors of theirs working towards mimicking them. So when people talk about the end of the tech company and only having a company survive, it's because technology is the enabler for everyone to be able to move forward.
And then this era of data, we have to make sure that we can gain control of it. So when we start dealing with artificial intelligence and why it's important, you start looking at the pieces that really matter. And one of the biggest pieces in this whole problem is the fact that data is expanding at an exorbitant rate. We have more data now generated in the last two years than 90% of the world's data that was there before.
So you end up with this big issue of all these organizations that you're working with that you're helping to drive to benefit their stakeholders and you realize the data really becomes a problem. The people that are the heart of this or at the crux of this are these little-known groups called data scientists.
Now most people have heard that term but that term is really very unique. And I'll talk a little bit about why it's unique. There's always been analysts, there's been analysts for the last 50, 60, 70 years in business. Analysts at, you know, category management, analysts within finance, analysts within supply chain, analysts in all different elements of your business.
And they used to do a great job at uncovering all sorts of different pieces of information to try to optimize that business's performance. But the reality is that we've had to grow even further even in the last decade to include a lot more types of information processing. No longer is it that an analyst is going to use a rows-and-column grid orientation to message their information.
And because the data is getting so big, you now have outgrown all the ability to even process it in a simple spreadsheet with all the data that you're looking for, with all the type of formulaic expression that will drive to that insight. So what's happening in the marketplace is, you've got this massive amount of data that's being generated every day and you've got a massive hole or a crater where we're not able to actually fulfill these types of roles like a data scientist.
And what companies are saying today is that 83% of the companies, that same argument around 68% of companies'CEOs are looking to analytics, they're saying they're struggling to find data scientists. And it's because of the diagram like this, on the right-hand side, because a true data scientist will naturally have elements from these three pieces of the Venn diagram.
They'll have mathematics and/or computer science and they will have to have an element of domain expertise. Now you can see the fringes where you've got statistical researchers and machine learning codifiers and you've got data processing analysts, but at the heart of all of it is data science. And so artificial intelligence is really about the fact that you need to take those big elements of data with the data scientist and other types of engineers and really create and craft a way for it to be expressed to the users.
Now you can see on your screens here kind of the definitions of machine learning and artificial intelligence, I tried to make it really simple. At the end of the day, artificial intelligence's whole goal is to make sure that it can do some of the things that people are normally doing faster, better, and typically with more levels of intelligence identification than what you can do based on the human brains and the human cognition capabilities.
So AI is out there, it's persistent. So the question becomes why now? Why is it all of a sudden AI is this big hot topic? Well, at the end of the day, we're in what's called an AI spring. We're in the AI spring because over a number of successions, two major ones, we went through AI winters where AI was just on the shelf, you know, dormant.
The first one started in the '60s after a 10-year run of trying to build artificial intelligence to translate English to Russian and back and forth through a conversation and through textual documents. And with a significant amount of investment by the U.S. Department of Defense, it took many, many years to realize that they couldn't actually do it.
They did not have the technology elements there yet. But we have the technology elements now. We have the technology elements that in just three years ago or four now that we're in 2018, we had an organization like DeepMind that was able to harness a traditional gaming system that some of us grew up on, the Atari 2600, and have a neural net programmatic software learn based on the pixelation of the game itself and control it and win more effectively than a human.
And not only did it do "Space Invaders,"as seen on the screen here, it did "Brick Breaker," did all of these things. And if you think back in history, even in the last decade, this was probably the second most important inflection point. The first one might have been this little thing called Watson that IBM was talking about at the start of 2011 and eventually culminated in a game of Jeopardy.
These two events really started driving almost like arms race in technology firms to really get closer and closer to artificial intelligence. There's been an amazing amount of news about AI because of this new coming, this new spring, and it's because we now are able to process more information more effectively and we've broken through some of the barriers in an algorithmic expression to deliver on what Andrew Ng talks about as the most massive transformation in the last 100 years.
And he's really talking about the fact that when we go back in history, when we go back in time from this point backwards, the most significant other player in terms of really changing the world was, in his opinion, electricity. And now where we are is just at the pinnacle of the starting point of really a massive transformation.
And that transformation is happening all over the place. People talk about robotic process automation, people talk about what's happening inside of, say, the auto industry over the last two decades with the amount of robots that are out there. But now this is coming into more of the traditional white collar jobs, whether it'd be claims assistants in financial institutions, whether it'd be, you know, teams within legal.
And it's great because at the end of the day, what that's enabling and empowering is it's empowering the people that were originally training for something very specific to have more insights available to them to make the right course corrections or the right decisions. So you end up with just a couple of smatterings of ideas here in terms of where AI is being used in the professions, but at the end of the day, if you just take the last one, healthcare, one of my most fascinating discussions was talking to the Watson Health team a few years back.
And for them to share with me things like it takes for a doctor today, an MD today, it can take up to 160 hours a week just to read all the literature they're getting just in the U.S. That doesn't take into account the rest of the world's publications. And the last time I checked, we don't have that many more hours in the week than 160.
So a doctor would have to literally be reading almost full-time for an entire week just to catch up. And so when you can apply it to these types of professions, when you can now streamline the collection and coordination of that data and bring in meaningful insights when a doctor either is asking a question or trying to find out something specifically on symptoms, is pretty remarkable. And I'd like to talk about the fact that these professions have always wanted this type of enabler.
I don't think, and if we start bringing it into the audit space and what I've started the conversation with about turbocharging your year-end audits. When you look at CPAs, when you look at their profession, I don't think there's any auditor that went through their CPA and said, "Yes, I want to be an Excel jockey for the rest of my life." I think all of them had a really strong desire to help organizations in a number of ways.
Make sure that their statements are coming out materially free of any errors or anomalies or issues, and actually help guide those businesses. But it hasn't been until the last few years where we've actually been able to make some differences. So there's this big new hope about the fact that AI will take us from here to there.
And I'll leave you with one last comment on AI and just how interesting it is. The picture of the cat is there on purpose. Deep learning was enabling Google to find the cat detector and it finds cats in videos almost entirely unsupervised. It just throws it at its YouTube channel and it can identify and catalog every single cat, every single type of cat, what's the cat is doing and catalog it and attribute it.
Now if you're not a cat lover, then I apologize for going in that sidetrack. But what's great about that is those are things that a person manually would have to identify, catalog, and you can do this now with technology. So now you can type into Google looking for something specific and the cat detector will respond in kind and find you the right videos based on the criteria you want.
That's the type of power that we're going to enable professions with. So let's talk about a real-world example and how MindBridge is using AI, artificial intelligence techniques, to uncover errors in financial data. So when we go on this pathway of looking at financials, everyone's immediate reaction is, "Well, but we've been doing good enough, haven't we?"
And the reality is if you talk to most public accounting firms from top to bottom, there is a desire to change. They recognize that the current practices and processes that they're using are mostly ineffective at actually detecting the errors.
They're really creating a sample to find whether or not there might be an error or not. And that's partly because of the big data topic that I started with. It's partly because of the limited human capacity, and it's not that humans aren't brilliant individuals at different levels, it's that you can't go through tens of thousands of transactions necessarily and pick out the one that might cause an issue.
Also when you start looking at it from a subjective perspective and you start taking away from the data speaking for itself, you end up losing elements of the line of sight. So at the end of the day, this collision of outdated tools and the practices, sampling, and things like that, are really coming down to the fact that it's really hard to detect errors.
And therefore we felt that this was something that we would want to solve. And to really go to talk about how fine-pointed this issue is, I'll use the big F word, fraud, just in this little piece. Fraud is harming the world. I think everyone can agree to that.
In the last 10 years, this is based on a report done by the Association of Certified Fraud Examiners, in the last 10 years alone, in their analysis, they're stating that 40% of all fraud is caught by mistakes or tip lines. So when you talk about what happened with British Telecom at the beginning of 2017 where they issued a profits warning saying that there was an irregularity to the tune of £500 million over the course of 2012 to 2016 in their Italian subsidiary, it wasn't caught because the analytics caught it, it wasn't caught because their public accounting firm found it in the audit, it wasn't caught because the CFO was doing a review, it was caught because someone made a phone call anonymously.
Not only in the last 10 years has the amount of fraud continued to be found mostly by mistakes or tips and very limitedly by analytics itself. It's actually taking longer and longer. It's actually almost doubled in the last 10 years the amount of months it takes to detect fraud going on in your business.
Now for any company, big or small, it doesn't matter if you're doing a million in turnover or 500 million or 5 billion, I do believe that everyone would be concerned if it takes them almost three years to find out that they're being defrauded. And the stakeholders today are becoming that much more opinionative on the issue itself.
What we know today is that almost 200 billion is found, again, 40% of that based on tip lines and very easy mistakes to see. But what the ACFE, the Association for Certified Fraud Examiners also said in that same Report to Nations report from 2016 is that they estimate three trillion goes missing annually.
That's a big number. And what's happening is that the stakeholders, as I mentioned just a second ago, are starting to become very upset. And the hardest part is that they're getting upset with both sides of the problem. They're getting upset with CEOs, like the British Telecom note I was making.
So you can see here on the top-left. This was a statement that came after the fact and it was a 21% stock drop in 24 hours, massive. But you can see the public accounting firms are also being held liable for their issues in not being able to find these types of issues and errors.
So what do we do about that? Well, we spend a lot of our time working through making it easy for auditors to do what they do best, which is assess the information that's presented to them. So we give them the ability to actually look at every single bit of data, if they so choose, but we do the analysis for them on 100% of the dataset in minutes versus waiting for a junior or a staff accountant to process a bunch of reports and then go on site and ask a client whole bunch of information and then build the opinion.
We can actually give you context where you need to hunt without any problems of scripting or loading data multiple times. And at the end of the day, making sure that the entire audit team has access to the same information in the same way. So no longer are we worried about whether, you know, we ask Sally or Bill or some other individual who has 15 years of experience for their input, we can now actually have everyone seeing the exact same type of analytical value and output right away.
So what is it that the solution actually does and what is that we feel is really important? Well, it's an extensible AI platform. And when we talk about the audits, this can be for the public accounting firms that are doing the audits on behalf of and creating an opinion, or it can be based on chief audit execs in their team, an internal audit, inside the enterprise.
But the extensibility is one of the most important parts because as we look at the data itself, we're able to extend out past the typical blocking and tackling. We detect all sorts of different errors and anomalies and issues within the data. We find out when there are people who have maybe a higher level of risk based on the vendors they're paying, based on the amount of monetary flow going through their transactions.
We can assess whether someone makes an error, just a simple error when I'm using a number pad of hitting the decimal place or a three before hitting Enter inadvertently. But the biggest thing is we get a lot more insight to the right individuals and we do that at an amazing speed that allows you to get going with the audits that much more quickly which drives down the overall cost on all sides and ensures that the assurance for that and the high degree of risk is minimized, high assurance, lower risk.
So the way this works is very simple. We assemble the series of technology elements that work in the background as soon as you give us data in order to understand what's happening within that business's general ledger and subsidiary ledgers. We score every single transaction leveraging a number of control point insights as we call them or algorithms or tests that you may call them.
And we do that across all the major categories that makes sense. The first two, business rules and statistical models and methods are the things that every computer-assisted audit tool is able to do today, but we do it on steroids because we do it across every single dataset all the time and amalgamate the scores. We also inject machine learning algorithms.
And as I started off, AI, really big word, really big amount of confusion about what it is, what we do specifically is we allow unsupervised learning techniques identify issues and irregularities in the data itself, transactional flow, monetary flow, different elements like that.
We have supervised learning techniques that we've deployed where we've actually created an entire expert score, leveraging professionals like yourselves as our guinea pigs to understand how they work and train the system. So there is no training needed for yourselves, it's already been done. And then reinforced learning, which is one biggest parts of machine learning and AI and where the biggest defects will ultimately be is that over time every single audit you're working on will get better and better and better in terms of our ability to understand where the issues are.
And, of course, going across known knowns of fraudulent or erroneous patterns based on the database we've been building for the last few years and going out to other types of data like ensuring that we can understand other elements, your controls, your user lists, your vendor master file, things like that.
And make sure that there isn't anything going on where a payment isn't manually being created to an entity that doesn't even exist. We do all of that in minutes and seconds. One of the biggest challenges with artificial intelligence in most people's viewpoint is that it's this really big black box and it just gives you an answer. So what we do is we give you the ability to actually go in and look at the data and drill in and find out exactly why we believe a transaction has a certain risk.
We will tell you all the things that triggered. We will tell you to what degree they triggered. How odd or infrequent or suspicious that information is. And we do that in an effort to help the auditor, to help the professional be able to make the right judgment.
So we augment your team, we augment the individuals with more knowledge and more insight of the data, and we make it as explainable as possible. So you can go to your clients or if you're at the organization itself, you can go to your leadership and say, "We think we had an issue here.Here's what we're seeing. We're going go and investigate it." And, again, through that feedback loop of users leveraging the solution, following up on things, marking things as normal the system will get more and more refined for the dataset you're working on or the organization you're working on across time, but also it allows for, should everyone opt in and decide to do this, it allows for that learning to go across businesses.
So you have the power of hundreds of auditors sitting right besides you shoulder to shoulder analyzing the information. So we look at the Ai Auditor as two simple and easy steps and we really do want to turbocharge your audits. So we make it simple. Load your data, review the results. During those two steps, a lot of things are happening, a lot of things are moving around, right?
If you are on the CPA firm side, you're getting input from your clients, sometimes in text files, sometimes in Excel files, or maybe they granted you access to their cloud ERP like the ones listed here, QuickBooks, NetSuite, Intacct, and you're able to just log in and grab the data you need for the period you need. Our output to you is some graphics that let you see the data itself.
And what I'm going to do is I'm going to flip over and actually show you what this looks like. So I've pre-logged into the system. You can see I've got really crazy names for all my different organizations I'm working with. But let's start from the beginning.
Let's show you just how simple it is for you to get started with AI in the audit profession. I'm going to create a new organization, My Wicked Coffee Company. You might say that I'm an addict of coffee, that is. I'm going to go ahead and maybe I'll be doing an interim or an end of year.
I might have some other settings that I need, companies that are doing, you know, 630s. You know, we'll set that to July 1st as a fiscal start. And the first thing the system says is, "All right.What data do you want to load?" So, again, we can go and grab data from pretty much anywhere from any type of subledger system, from any system, and we can load it into the solution.
Now I'm going to open that Excel file that I just loaded into the Ai Auditor. But before I do that, I just want to show on that screen what the system's already done. It took seconds to load that data in. It's identified what type of ledger system it came from, Sage 50, and now it's going through the process of running our analytics pipeline to see the data.
So I'm going to go ahead and look at the Excel file and determine what kind of data is in here. So as you can imagine, I've got lots of transaction data by account. Really messy. This may not be the easiest thing for me to look at. Can you imagine trying to understand in this MindBridge Plumbing Limited dataset how am I going to find an error?
Well, most of you that have done this for a little while, you already know the answer that you're used to. That might be just, take it from this Excel model, build it into a pivot table, and start doing some analysis. For others, you might have taken a visual analytics tool and dumped it in there. But at the end of the day, you're taking a lot of time to strip out all the noise from within this data. And when you do that, the minute you start altering the file that your client gave you is the minute you have to start second-guessing, "Did I get it all?Is it complete? Are all the data elements being pulled in?"
With the Ai Auditor, you don't have to worry about such a thing. As I go back to it, we're going to start exploring what it's already done and what it's already learned. So I'm going to go and show you that without it ever seeing that data, you can see that bank account first line, next line is my accounts receivable. It's automatically been built into an account grouping.
So I can see exactly where all of my data is. So the minute the Ai Auditor gets your data, it's going to build an entire grouping for you of all the data and show you hierarchically where that belongs. If it doesn't find or it can't assess it, it will put it in an unspecified bucket that is close enough.
So your job now instead of it taking you maybe an hour or two hours, three hours to build the account grouping, you can actually do it in minutes. You can access it, review it, make changes, and process the data. If it's a client you've already worked on, you might just upload the account list directly. But as we were talking there, the thing that happened was it notified me on the top-right that, "I'm ready to go. My analysis is complete."
So when I look at the data itself for the high level, I can see exactly how the information is broken down. Now if you recall, I talked about these concepts called control point insights. These are the algorithms and the rules and the tests that we run on your behalf and we weigh them. A manual entry might have one fraction of weighing, a weekend post might have another fraction.
We get into a combined risk score. And based on a combined risk score, when it tips over 50 it goes into high, 30 to 50 it goes into medium, and below 30 it goes into low. And I can see very quickly, very visually where I may need to assess my information. Almost $2 million in high risk, almost $12 million in medium risk. That's just over a quarter of a percent of the total dataset sitting in high and medium risk.
Now that might be interesting, but what's more interesting is the question that you're naturally asking is, where? So I can click on any of those tiles and go look at the transactions themselves. But I may be curious to find out where across time those issues are. Oh, cutting over between years, right, end of year. I may also want to know a little bit more than just the time series.
I want to know where, what accounts, what groupings are most effective. So as I look at my diagram here, I can hover over and see exactly where the monetary flow is. Highest amount of monetary flow in this business is headed through assets, then through current assets. Then it seems to be a tie between AR and another.
But because of this dark green, it's telling me that, "You know what? These are lower-risk." Not saying that there may not be an issue in here, but the likelihood of it being an issue is really, really small. But I do see in my current assets a little sliver here that's trending towards yellow, and that is where I may want to spend some time. So I can click into its equipment and see all the transactions that are making up that more yellowish risk profile.
But, of course, the most concerning ones, if you've seen that visual, is that there's some orange and red. And if you look at the ledger on the top left there, you'll notice that red is pretty much bad. So I can see the director's fees, the management salaries, and bad debts are all sitting in that higher-than-we-want risk. So you might ask yourself, "So how will this help me?How will this turbocharge my audits?"
Well, first and foremost, all your engagement tools, all the different types of tools that are giving you guidance are going to use you subjective input to define where something needs to be and where you need to spend your time. You can very quickly here validate that, yes, in fact, that might be an issue or maybe not. You can also very quickly, if you're still doing random samples, go ahead and create your own sample very quickly.
Click of a button, I'm going to create the entire plan. If your guidance tool said, "Based on this population you need to do 60, 70, 80 samples," no problem, just type it in, the system will do the rest. Everything that you would expect the system to do, it's already done and it's going to then weigh that based on the risk profile. So every transaction has the opportunity of being selected but we refine that, just as you would when you're doing material transaction threshold, we do it across the board.
You want to add in materiality, no problem. The system is built for you to get to the best outcomes possible for your audit. Now I'm going to walk through actually some of the data itself, and I'm going to do that in a number of factors. First, I'm going to go to my reports and talk about the fact that everything that we do in that matter of one or two minutes as we were loading and analyzing the data, can be checked out.
We can deliver income statements, leverage ratio reporting, profitability reporting, a completeness check, if you're loading payables or another subsidiary ledger, a different set of reports come out here. We do that all in minutes. So think about the time it would take you to take any data that your client's given you and produce their income statement with confidence. We do that in seconds.
So let's look at the actual intact details of what's happening here. So we're just going to go back in and let's look at the transactions themselves. It's identified that I'm looking at the ledger, great. So I see all the journals that are hidden here whether they're coming from a batch process and a subsidiary ledger or not, everything is in here. All fully analyzed, all with a risk score. And because we look at all the data across multiple facets, we can slice and dice it any which way we want.
So we have the ability to go in and say, "You know what?I only want to look at my revenue lines and see where the risk is there.And within that, maybe I want to only look at certain monetary values.Maybe the threshold is 100,000, you type that in, away it goes.Maybe I want to look at it based on, you know, my traditional audit standards, like I need to see anything that was posted on the weekend and take a sample there."
And we generate the list for you. Now the interesting thing you'll note on this is I picked revenue, I picked weekend post, it tilted the list. But if you look at that overall risk score, you'll notice in this business weekend post really isn't the riskiest thing, it's sitting at 23%, remember that's in our low category. If I reset everything and take us back to the full list, I can see there's a lot more risk in a lot of other transaction.
So I'm going pick one at random here and it will give me the details. Now any auditor that's scanning through data might be looking for suspicious keywords. Guess what? It's built in. Part of the risk score in here is that it actually looked at a memo field and it was looking for anything that had a variance of cancel.
So that added to the risk score. If I was doing this manually, just looking at Excel, I would have been doing, you know, Find, and it would find all those ones. But remember that file I had would make me go to five different independent roads and I wouldn't be able to see it all in one. I'd have to massage the data somehow. Here it is, all done for you. But what's interesting about this data is that you get all the detail right there ready for you.
We can tell you exactly where we see the issues and to what degree we see those issues. So there's at least one line in here that's in the top 2%. There's two in fact. This line here, it's 156 on Accounts Receivable and this line here for 42400 in cement. Those two rows are my top 2%. Would I have picked this in a random sample?
Maybe, maybe not. But what I can see is all the other things that happened. The entire transaction was reversed out based on another transaction, 2949. So we give you all sorts of insights with every bit of testing we do. We can tell you when we think something has been adjusted and where we think those transactions are from. Because not every time when you reverse out a transaction do you reverse out the whole thing or in the unlikely event or unfortunate event that you might have a fraudster in your business, they will never do it out of the same exact transaction.
They will always try to find a way to split that transaction out. We will tell you where we think that issue is. We also have, as we talked about, this concept of machine learning AI. And here in the third row down, you end up starting to see the impact of that. We see things, the control points there are named rare flows or expert score or flow analysis.
These machine learning algorithms that are running have lots of different testing that they're built into and it's telling you whether something is potentially an issue. The expert score system has identified that these two lines, taxes on sales and accounts receivable, looks to be a high risk. And it tells me with what degree of confidence it feels that risk is.
In this case, it feels the risk is sitting around 67%. Again, drawing your attention to how you can make your audits that much more effective would also tell you all the other transactions that have the exact same type of marking. You just click on the purple button and away you go to see that list.
So within our machine learning algorithms, we're looking at all sorts of different things. As I said, we've trained this system. That's what the expert score is all about. It will identify patterns in your data where there's potentially a concern or potentially a risk based on peers in the field. These are based on chartered public accountants, certified public accountants, depending on where you are in the world and what lingo you speak.
And it was also guided by the ACFE members, Certified Fraud Examiners and Certified Financial Crime Specialists. So three different teams looking at the data in different lenses helped us build this expert score. Why? Because when we're looking for errors we want to know the different types and reasons why something might be erroneous. So we did that.
We built the expert score for you. It will run on any of the data. The flow analysis and the rare flows, these are two that are really critical to making sure that you're uncovering anything that potentially is an issue. Because these two will look for any type of anomaly where an account communication looks to be off. So it's looking to see, in the context of this dataset, in the context of this organization, My Wicked Coffee Company, where exactly are the transactions that don't normally appear.
So it's telling you that these three accounts are normally not used in a single transaction. So I can really start unpacking and understanding exactly where that potential risk is. As usual, you always have the ability to build your judgment-based sample list. Again, I'm looking at the ledger, so I'm giving you lots of context.
But let's just go and, as any good cooking show would have, let's go and look at something I prepared for you as it relates to accounts payable, just to give you another flavor of how your engagements can go differently. So, again, my colleague, Mike, has loaded up some data. Some fictitious data for the current year, which is kind of funny. Probably should have had him use last year but away we go.
It's telling me where the risk potentially is. It says there's 85 transactions with almost $1 billion of flow going out to payables. It's got another $1.1 billion and representing 990 transactions in the medium. Now this is a lot of work. This could be a lot of work for an individual to go through. So how do you filter it? How do you look at where things are?
Well, again, you have that same ability to look at it by month or by week or by day. We can show you exactly where we think the risky points happens. We can equally tell you which account groupings, where you're paying things from, where some of those risk items are. So wages and salaries, data processing, $29 million. A little bit of a yellow flag there.
It's potentially something we want to look at. But typically, when you're doing your payables review, you also want to look at it other ways. You want to see where you're spending money with a particular vendor. So we can see here very quickly all the different vendors that my organization is communicating with and the size of the box is the amount of money and the color of the box is back to that risk.
So I can tell you that based on the system's viewpoint, Anthony PLC, Pruitt, Cummings, Hodges, Goodwin, and Peterson. They're all relatively low-risk. They have decent amount of monetary flow, $60, $70 million each of them. Combined, that's about $210 or so million.
But look at this one big one, Shaw, Brown, and Gonzalez, $713 million with the flow being paid out to them and representing roughly 14%, a fairly high risk score in terms of total aggregate. Even high on Kelly-Cook. So we can go in and look at what's happening to that vendor.
All the various transactions. And you'll know that we don't just have the vendor, we have users as well. You can look at this from the user perspective. I can see one here that has 25 different entries. Let's go figure out what's actually happened in this transaction. I see lots of different monetary flows in and out of these accounts and it's telling me what I need to know about. There are 12 transactions that are part of the high monetary value.
There's a few transactions, let's see which ones. One, two…two of them which are outliers, meaning we haven't seen a data element like this in that payable to this type of sub-account. Right? Now the good news is, it is in the duplicates. The purchase order didn't actually happen after the invoice was posted in.
The vendor looks to be fairly solid in terms of this particular vendor, but there was a lot of risk, if you remember, for Kelly-Cook. So which ones have risk? And we can show you all of those in detail. We can show you exactly, again, which control points were having an issue and which users.
So this whole dataset is broken down with you having the ability to see absolutely everything that's going on within your business. So let's go and take a look at our controls. And we can see very quickly that Catherine90, pretty good, Ncollins, pretty good, Rachel05, pretty good, Mitchelltracy, great, in terms of the depth of green.
But Otucker, Dorothy77, George01, all seem to have a higher-than-we-would-like risk profile. So I can look at and see by vendor, by user, all the different information that you would want. George is a Senior Accountant. That's probably not so great if he's also in red. Dorothy is also a Senior Accountant. We've got quite a few of those here.
But in all these cases, we can see exactly what's there. And then, again, to help facilitate a better process for you and a speedier time through the audit, you actually have the ability to go in and see exactly which transactions might be at risk. Every time you work through your dataset and you assign a follow-up to any one of these transactions, it goes into the audit plan.
So I'm going to just quickly assign a couple. I'll take my friend Stefan and I'll give him a week to go and look at this one. And maybe I'll do one more just for visual. And this one I will assign over to Robin and give him a little bit less time.
And now I can see in my audit plan what people are working on. So if I do a bit of a recap, and I'll go back in and show you a couple of quick summary slides, we have the ability for you to see exactly where the system thinks the risk is based on over 50 algorithms. You can go through the process of manually filtering and sorting or you can create a random sample using our intelligent sampler.
Every time you create a task for follow-up, it goes into the audit plan. So you immediately have real-time context out to your user, then your team, with any feedback they're getting to come back in. And then finally, obviously, as we had mentioned, you actually get lots of different reports right out of the box the minute you load data.
So, again, just going back and talking a little bit about what we do and why we do it. We look at this as very key to your success in the fields of being a CPA and an auditor, whether you're working for a CPA firm or whether you're working for a corporation. At the end of the day, our whole focus is on finding anything that looks to be suspicious or an issue and score it for you so that you can very quickly without any need for coding and scripting or anything like that, you can get to the insights that might help you through that audit as fast as possible.
We have the ability to go through massive amounts of data and we're being recognized for that. We spend all of our time working with you, the practice. At the bottom you can see a few press releases that came out from the behemoths like Thomson Reuters and BDO and some very large significant players in Kingston Smith and KNAV.
But at the end of the day, it's all about getting to some very interesting thing, helping you get through your datasets as effectively as possible so that when your users, if they're your clients, or if they're your team members internally to an organization, you can understand exactly where you stand on your financials. You can now enable the entire team.
You don't have to have a specialist who knows this one esoteric little tool over in the corner and feed them a whole bunch of, you know, pizza throughout busy season, right? You can let everyone share in looking and understanding where your client's challenge is. We focus on helping you do a much better job and we're doing that through powering a system with AI and machine learning.
So just as I start to get ready to close out, there's a question and answer panel. Please go ahead and start pushing some Q&A in there. I will take time to review it very quickly answer some of them. But just a couple of quick testimonials around what we're doing.
If you're doing a forensic view of things, take a look at BDO's quote, right? "Investigations that took months combing through millions of transactions are now focused and can be concentrated on in just hours," right? Becky Smith over at Kingston Smith, "Being able to explain to the regulators exactly why they picked particular samples instead of just saying, 'Oh, one of my trainees picked 10 randomly.'"And Goliaths in the market space like Thomson Reuters Tax & Accounting.
"We are unique in the audit field based on what we do and how we help the profession." So our goal clearly is to make sure that we can do the best we can for your organization. There's a YouTube link here. Just search on Becky Smith.
You can hear exactly from her, her words on why they picked and selected and are very confident in using the Ai Auditor service.
Demystifying Artificial Intelligence and the Impact on Auditing
AI is helping of a number of professions, including auditors, deal with a changing landscape. Learn how the Ai Auditor platform leverages AI to provide faster, more highly assured audits than ever before.
MindBridge Ai Technology, Mission and Product Overview
Learn about our MindBridge Ai technology, why we created it and how we take a hybrid approach cross-correlating business rules, statistical methods and machine learning to deliver better financial analysis.
Interview with Becky Shield of Kingston Smith
Kingston Smith, a top 20 UK chartered accountancy and audit firm is leveraging MindBridge’s Ai Auditor. Becky Shield talks about the value that Ai Auditor brings to the firm.