Close this search box.
Close this search box.
Resources   /   Blog

AI in Accounts Payable: Can a Computer Do My Job? 

January 11, 2024
net 30 podcast logo

If you’re worried that artificial intelligence (AI) may make your job obsolete, you’re not alone. A recent Forbes Advisor survey found that 77% of consumers are concerned that AI will cause human job loss in the near term.  

Experts have identified many use cases for AI in accounts payable (AP) and other finance and accounting functions, noting that the technology is particularly well-suited to these data-centric departments. But can it do the job of a human?  

AI Fundamentals

“Artificial intelligence is the science that uses data to help us make better decisions,” said David Tareen, senior director of product marketing at AvidXchange. Tareen has a robust background working with machine learning and data science. According to Tareen, when it comes to using AI for data analysis, the more data you have, the better AI’s recommendations will be.  

The math required to analyze large data sets demands advanced chips and microprocessors. The popularity of AI has surged in the last year because chip companies have developed technologies that are less expensive and run much faster, using less energy. The processing power necessary to run AI applications is no longer reserved for large universities and governments.  

AI in Accounts Payable

Since AP departments handle a high volume of financial transactions, they’re full of data that AI can leverage to inform decisions within a business. Additionally, AI can help AP staff make manual processes more efficient.  

AI is a broad term that encompasses many types of technologies. According to Tareen, the following are those you’re most likely to find within an AP department: 

  • Machine learning: Tareen calls machine learning the “cornerstone of AI technology.” It trains itself using data. In finance, it can detect and flag anomalous transactions, preventing fraud.  
  • Deep learning: Deep learning is similar to machine learning, but it involves larger data sets. It’s ideal for detecting objects in videos and can be used to detect fraud in high-volume transactions.  
  • Natural language processing (NLP): NLP is the basis of generative AI. It lets individuals access AI applications without knowledge of computer languages or coding.  
  • Computer vision: Computer vision is used to look at invoices and automatically detect information like vendor name and payment due date.  

In a recent episode of AvidXchange’s “Net 30” podcast, Tareen shared potential use cases for AI in accounts payable departments. “The possibilities are endless,” he said, but these are a few applications he expects to see adopted near term to make AP duties easier: 

  • Extract data from an invoice and automatically code it within your internal accounting system to ensure costs are allocated properly.  
  • Detect anomalies in invoices and highlight any that may be fraudulent.  
  • Match invoices to purchase orders and sales receipts, flagging inconsistencies.  
  • Analyze workflows and find the optimal path for efficient invoice approvals.  
  • Conduct general ledger code searches to allocate expenses appropriately.  
  • Rapidly process payments with payment APIs.   

Will AI Make Accounts Payable Jobs Obsolete?

Though there are many useful applications for AI in accounts payable, Tareen does not believe AI will take away AP jobs. “The approach [for implementing AI in accounts payable] has to be a partnership between people and algorithms,” he said.  

“People are good at some things, such as intuition, imagination, creativity and making very quick decisions. Machines are good at other things like handling really large data sets and finding patterns within data. Those skills are not common. They have lots to give each other. It’s through that partnership where you really start to see the benefit.”

On our “Net 30” podcast, Tareen shared an example of a successful partnership between AI and employees at AvidXchange. Our invoice indexing process utilizes both AI and professional indexers.  

“When an invoice comes in, the algorithms take a look at it. That speeds up the process. But then a person comes in and also takes a look at it. [That person] is familiar with the account and knows what’s going on, so that [our customers] are getting the best user experience,” he said.  

AP professionals shouldn’t be scared to embrace AI, but Tareen understands why some are. “AP professionals live in a world of a lot of regulation and compliance. They’ve invested a lot in the skills and the infrastructure that they have today to be successful,” he said. “I think there’s hesitation because it’s unknown how the algorithms are going to perform.” 

To hear more of Tareen’s thoughts on AI in accounts payable, stream this podcast episode, “AI 101: The Fundamentals for Accounts Payables Professionals,” on your favorite platform, including Spotify, Apple, iHeart and Pandora, or click below to listen now.  

To learn more about how AP departments are leveraging AI to create efficiencies and optimize processes, download our e-book, “The Ultimate Guide to AI in Finance.”  

Complete Transcript

Please note: The “Net 30” podcast is designed for audio consumption. Transcripts are generated using speech recognition software and may contain errors. Please check the corresponding audio before quoting in print. 


Hi, my name is Chris Elmore. I am AvidXchange’s Chief Evangelist and I’m the host of the Net 30 podcast. On the Net 30 podcast, we meet with industry leaders to unpack problems and solutions and talk about innovations that are impacting financial professionals. The best part about this, we’re going to do it all within 30 minutes. So let’s get into it. 


Welcome to the show, David Tareen. Thanks for coming on. I’m excited about what you’re ready to talk about. So you’re the Senior Director of Product Marketing. So tell folks a little bit about who you are and what’s brought you to this point in your life.  


Really excited to be here. I’ve been with AvidXchange for a little over a year. It’s honestly one of the most fun things I’ve ever done and I’ll tell you the reason why is because the work that we do here makes such a difference in people’s lives because accounts payable is hard and just keeping track of everything that you have to do within all the regulations and the compliance is just hard and our ability to just simplify all of that and automate all of that, that’s just game changing for people. So it’s exciting to be a part of that journey.  


Now, David, this is a three-parter. We’re going to debunk the misconceptions of artifical intelligence and we’re going to base it on AvidXchange’s 2024 trends survey that we conducted in September of 2023 where we asked 500 finance executives at middle market companies about their use of AI. And the way that it works is that we’re going to scratch the surface on part one today, we’re going to go a little bit deeper. And then lastly, we’re going to go super, super deep. We’re going to get really personal in episode number three. So you’re going to be along for all three of these episodes.  

But to start today’s off, every time we turn around, we’re talking AI, artificial intelligence. People are still trying to figure out how to use this stuff and how it can impact human capabilities and make better decisions, better reporting, improve outcomes, deliver impacts to organizations.  

So today we’re going to get into kind of the fundamentals. Three things that we’re going to talk about. We’re going to talk about real basic definition of artificial intelligence. And then we’re going to talk about the different use cases that finance and accounts payable teams can actually use. We’re going to talk a little bit about people’s hesitations and factors to consider. Okay, that’s the big build up. Why in the world are you the guest on an artificial intelligence podcast? What makes you the expert?  


I have a background in machine learning AI. Just from the number of places, my experiences where I’ve worked very closely with really senior data science teams. And at the end of the day, data science is really the magic. 


We’ve had a lot of conversations and I can’t wait to dispense some of the information that we’ve had one on one conversations about. So how about we get into it? So the first topic is just kind of the basic definition. So from your perspective, what do you think that the overall response has been to AI tools and either for everyday use and finance departments or just in general?  


This is a technology that’s captured such a broad swath of the population’s imagination, because usually you get, “Oh, it’s cloud. Oh, it’s mobile. Oh, it’s social and things like that.” And there’s really crisp definitions for it, but this is so broad that everyone defines it a different way. There’s no right or wrong way to define it. Honestly to just get the conversation started. But if you ask for my definition, I’ll call artificial intelligence is sort of the science that uses data to help us make better decisions at the end of the day. And Chris, you said it, you said it exactly right. There’s so many things this technology can do, but when we talk to a lot of our customers, we’ve got a lot of ideas on what they wanna do. We always try to ground them back in, “Okay, what is the decision that you’re wanting to make?” And how can we make that decision better with using these artificial intelligence techniques, machine learning, deep learning, computer vision, all of that stuff? But it really comes down to, like you said, making better decisions.  


Using science to make better decisions. I like it by the way, I think it’s applicable. From your perspective, why is this thing fired up? Like I said in the intro, it seems that every time we turn around, we’re talking about this. 


Why is it, why now?  


When you think about the math that’s required to make, to analyze a lot of data and the more data, the better decision you’re going to get. What that requires is really advanced chips and microprocessors. And the reason this is sort of kicked off now in the last, I would say in the last four or five years, really, is you think about what companies like. Nvidia are doing. Companies like Intel are doing. 


They’re really enabling these chips to run so much faster with so much less energy that the cost of running this type of math, it’s no longer restricted to big governments or big universities or research places. Now, suddenly, if you’re a small medium enterprise, you can go rent space on Amazon and yeah, you’ll pay for that cloud space, but you can run some really advanced algorithms on just cloud capacity, so you don’t have to set it up yourself. 

And I think that cost coming down has just sort of been a big cause of the explosion in algorithms and everything that’s happened around artificial intelligence.  


Well so you’ve already touched on your definition, but then let’s talk about types. Now you gave kind of a brief explanation of particular types. We have types like natural language processing, computer vision, deep learning, natural language understanding, forecasting, optimization. 

Can you riff on a couple of those or are there others you can think of? 


Absolutely. I would encourage anyone who’s sort of interested to take the conversation beyond just the moniker AI and get into, “Okay, what, what are we really talking about? So, I mean, the most popular ones machine learning is a cornerstone, absolute cornerstone. 

In my mind, the way machine learning is different from programming is when you’re programming, you’re going to give a set of instructions and the computer is just going to follow it, right? If I’m programming a chess set, it’s only going to be as good as my chess skills. Machine learning is different. 


The way it’s different is that you just give the computer data off let’s say a million chess games or a billion chess games. However, many right? And it’s going to program itself. It’s going to learn all the patterns by itself. So that’s how machine learning is different from just regular programming. 

And that is the absolute cornerstone of any AI technology. Machine learning is absolutely first. The way AP professionals are using it, a lot of our customers are using it, is because it’s so good in being able to find patterns within data is that it’s good at finding fraud. So you’ve got 100 transactions going every day. 

You’ve got maybe a million going every day. Now you can start to see which one of those transactions is anomalous. And machine learning is absolutely such a great tool to find these anomalous transactions through a methodology called anomaly detection. There’s a lot of algorithms that do that so I think that’s the cornerstone technology as we get into AI. 


I had never thought about machine learning programming being only as smart as the programmer, but now machine learning is a huge pivot because it’s just looking at data. I think that that to me is really eye opening. But let’s talk about use cases within finance or accounts payable departments. 


Yeah, absolutely. I think to go back to one of the earlier things that you said is machine learning is just one technology. There’s deep learning computer vision, natural language, and I think all of these have use cases within AP departments. In my mind, deep learning is just like machine learning since you’re dealing with so much more data. So think videos. Deep learning is really good for things like detecting objects in videos, detecting quality control for manufacturing. Natural language processing, that’s where the excitement is right now in the market. So natural language processing, or NLP, is what generative AI is based off of. 


And I know everyone’s talking about it, right? But that’s really an NLP capability within the artificial intelligence umbrella. The last couple of ones are computer vision. Really, really great when it comes to AP. Use cases – just the ability to look at an invoice and being able to detect logos and things like that where dates are placed because every invoice looks different. Computer vision is great. And then forecasting and optimization. We’ll talk about those coming up here as well.  

But you talked a little bit about what are the different use cases in AP when I think about an AP process, there’s an invoice that’s coming in, there’s possibly you want to match that invoice with what was ordered and what you actually received. 


You want to do some kind of workflow just to make sure that invoice is getting approved the right way. A big part of the process is you want to code it within your internal accounting systems to make sure that you’re allocating those costs the right way, and then finally generating that payment. 

So, man, when I think about that workflow and all the artificial intelligence technologies that we talked about – the possibilities are endless. You can use text analytics and computer vision to look at invoices and analyze invoices. You can use anomaly detection using machine learning to see if any of these invoices are trending towards fraud. 

You can do the workflows by looking at all the other workflows that you had and finding that optimal path and making those recommendations and predictions those decisions. And then finally you can do searches when you’re thinking about GL codes and making sure that it’s allocated to the right cost. And then just payments or APIs – a little bit out from an artificial intelligence standpoint, but absolutely very applicable. 


You’re probably scaring people at this point because the idea that we could probably see in our lifetime, the accounts payable process as you defined it being 100 percent AI.  


Yeah. Well, no, I think you’re on such a good point and I’ll tell you, this is an active conversation when we think about how AvidXchange should use artificial intelligence and how we do use artificial intelligence. 

So, yeah the possibilities are endless, but the approach has to be a partnership between people and the algorithms. I mean, I’ll give you sort of my thinking on this. People are good at some things, such as intuition, imagination, creativity. Making very quick decisions, right? Machines are good at other things. 


They’re good at things like handling really large data sets, finding patterns within data. So those, these two skills are not common. They have lots to give each other. So I think it’s through that partnership where you really start to see the benefit. One of the areas that AvidXchange uses artificial intelligence in partnership with real people is in the indexing process for invoices. 


When an invoice comes in, you want to have the algorithms take a look at it. Absolutely. That speeds up the process, but it takes that person to come in and also take a look at it. Who’s familiar with the account, who knows what’s going on so that you’re getting that best user experience. So I think the first thing is it might be scary, but we’re not giving the keys to it yet. It’s a partnership right now.  


If someone was kind of new to this whole thought of artificial intelligence and all the different variations, if they don’t use AI and then they use it successfully, what would kind of be like a couple of benefits?  


Yeah, absolutely. So I’ll give you three in my mind. The first one, especially for AP professionals is you’ve got a lot of processes going on and you always want to make sure that those processes are following compliance and rules are optimized with automation and artificial intelligence in the process. You can have repeatability and scale. So you’re, once you define the best practice, that algorithm is always going to follow that best practice and it’s going to continue to learn. 


So you’re optimizing and you’re getting the best process every time. The second benefit is user experience. They’re getting a consistent experience no matter if they want something at noon or at midnight or whatever time it is they’re getting that consistent experience. 

And then the third thing is a huge cost savings. As you think about scaling operations, it’s algorithms are really easy to scale. There’s some challenges, absolutely from a capacity standpoint, but to me, those are the three benefits.  


Last topic at a high level. Why are people hesitating?  


Yeah, no, I think there’s some real risks here, right? So for AP Professionals who live in a world of a lot of regulation and compliance, they’ve invested a lot in terms of the skills and the infrastructure that they have today to be successful. I think there’s hesitation because it’s unknown on what, how the algorithms are going to perform. 

There’s absolutely tech ways and methodologies that, and I think we’ll, we’ll get into that in the next session on how you can start to approach it. And it doesn’t have to be a big risk. You can start small. There’s a couple of other things that we’ll talk about. So stay tuned, but I think just a fear of the unknown is a big element of why people are hesitating quite frankly. 


What was your definition of AI science? What was it? Was it again? Science  


I define it as it’s the science of using data and analytics to make, to help us make better decisions.  


Yeah, I think that’s fascinating because it doesn’t start with your typical scientific method where you have a hypothesis and then you go improve your hypothesis. You just have information and then something comes out of that.  


And full disclosure this is a definition from some really, really smart data scientists that I used to work in the past. So not something that David Tareen just created.  


All right, more to come on this, David. Thanks for being on this introductory episode of Artificial Intelligence. I appreciate it. I got a lot out of this myself, especially around the notion of machine learning, just seeing what the data tells us. That’s fascinating. So more to come. Thanks, David, for joining us.  


I appreciate it, Chris. Thank you. 


Thanks for listening to the Net 30 podcast presented by AvidXchange. I desperately need to know what you think about this. Leaving a five star review would be fantastic. You can subscribe to the channel too. And, oh, by the way, while you’re waiting for the next episode, head over to to learn more.  

One more thing. If this conversation has somehow piqued your interest, which I really hope it has, about accounts payable automation from AvidXchange, there is an additional link to get a demonstration of our solution. 

I always say it shows way better than it tells, so click on that link, fill out the information, and someone will be in contact with you. So thanks again for listening to the Net 30 podcast, and we’ll see you really soon. 

A female and two male co-workers looking at a tablet.

Featured Download

Addressing 6 Common AP Pain Points with Data

Leveraging data analytics to drive accounts payable efficiency

We use cookies to improve your experience, personalize content and analyze our website’s performance. For more information on how we collect and use this information, please review our privacy policy.