Moderated by Sherry, Digital Transformation Consultant at Hyperbots
Sherry: Hello, and welcome to all our viewers on CFO insights. I am Sherry, a financial technology consultant at Hyperbots, and I’m very excited to have Jon Naseath with me, he is an accomplished executive with expertise in AI machine learning and computer vision driving impactful technology solutions in education, healthcare, and business. Thank you for joining us today, Jon, to discuss straight-through processing of invoices. Now let’s dive right in. Why do you think companies implement constraints on straight-through processing of invoices, even in seemingly straightforward scenarios, like three-way or two-way matching?
Jon Naseath: Yeah, I think the basic answer comes down to trust. The idea that someone’s manually looked at it and verified just gives additional comfort when someone’s not trusting the source data that they’re looking at. I think the short answer is just trust.
Sherry: And in your experience, what are the most common value-based constraints that companies put in place for STP and VI?
Jon Naseath: Yeah, I think usually it comes down to certain dollar amounts. So if it’s a high-volume, low-dollar amount processing, they’ll let those payments go through. But then for bigger ones, they’ll do so like even banks have thresholds of, say, $10,000 for something international. They’ll want additional details about it. I think, though, that doesn’t necessarily mean that you’re safe. I think that there are often situations where someone is going to do something fraudulent, they’ll intentionally do it below that threshold. Quick example: I was talking to a friend of mine, who’s the CEO of a company, and over lunch, he told me about how one of his trusted employees over the last year had been buying products and shipping them to their home for different things that they wanted and using them for personal uses. And it was small amounts here and there, but when they added up, it was very impactful for the company and certainly nothing that was authorized. So I just say be careful of even the small amounts. You need to have good control and be careful around them too.
Sherry: And looking at the industry, how do vendor-specific constraints impact the straight-through processing process? And what factors might lead a company to require manual review for certain vendors?
Jon Naseath: Sure. Different vendors, you know, every vendor is going to run into trouble at some point. Every vendor is going to try to stretch a payment or just pay you late because they can. Even big companies will do that just out of policy. So having a way to review what is there and watching for controls around fraud is important. Even companies with a consistent record of accuracy and reliable payments may have their invoice processed automatically, and you can let those go through. But you need to monitor and kind of give a credit score or credit profile for the different vendors you’re working with.
Sherry: In talking about different obstacles, what are the challenges associated with STP for foreign versus domestic vendors? And how do companies typically address these challenges?
Jon Naseath: Well, one aspect is that foreign companies, may take longer to get paid, or it may take longer when you do pay them. You have to be able to pay via wire, where otherwise you might be able to send a direct payment. So just the processing is different. I wouldn’t say that there’s an increased risk, although some countries, sure, if they’re ones the US has limitations on, you’d want to be careful but your bank will usually not let you send money to those anyway. So as far as straight-through processing, I don’t know that the processes themselves would be very different, but it would just be different risk profiles of the companies individually, as was mentioned before.
Sherry: And can you discuss the differences between contract-based and non-contract-based invoice processing, and how these differences affect the decision to use STP?
Jon Naseath: Sure. With contract-based invoices, you have a contract that defines what you’d be able to then pay on the invoice. If it’s non-contract-based, there might not even be an invoice either, and so there can be situations where a payment needs to be done without a formal invoice or contract. While most would say it’s not possible, I’ve seen situations where this can happen. And there are a lot of scary things happening with AI, where someone mimicked my voice and told a CEO to send a payment. Luckily, we had controls in place to catch it. So, contract control, invoice control, or even keyword control there are bad actors, and companies need to stay ahead of these with good controls.
Sherry: That sounds like quite a challenge, Jon. From what you’ve seen in the industry, how do companies typically handle PO-only matching scenarios? And why might they opt for manual intervention in these cases?
Jon Naseath: With PO-specific invoices, it’s approved, but you still need to verify if the work has been provided. A lot of times POs are done for services, and you end up relying on the project manager to verify that the work has been completed and approved before payment. Hopefully, the program manager stays on top of the project and approves only the work that has been completed and deserves payment.
Sherry: What are the risks associated with processing non-PO and non-GRN invoices? And how can AI help mitigate these risks?
Jon Naseath: With non-GRN and non-PO invoices, the risk is that without proper documentation, it’s easier for fraud to slip through. AI is great at spotting patterns, discrepancies, or things that don’t tie out. I’ve seen situations, where people swap out letters or change small details to make fraudulent payments, look legitimate. AI is much better at catching these subtle fraud attempts than humans.
Sherry: And finally, since AI is a trending topic, how do you see the role of AI evolving in the context of STP, especially with the various constraints we’ve just discussed?
Jon Naseath: AI is great because it can act like someone watching your back, monitoring transactions, and raising red flags when something doesn’t align. AI can help with both preventative and detective controls, identifying things designed to trick humans. In the long term, AI will play a crucial role in safeguarding against inconsistencies and helping businesses process payments more efficiently and securely.
Sherry: Thank you so much for these insightful answers, Jon. It’s clear from this interview that AI offers a promising solution to navigate such complexities.
Jon Naseath: Pleasure.
