Automate This Episode #03: Intelligent Fintech ft. Chris Happ
Automate This #03: Intelligent Fintech ft. Chris Happ
In this episode of Automate This, Ryan Nelson, Goby’s COO, is joined by his fellow co-founder and CEO, Chris Happ, for a discussion of the technologies that make automating accounts payable possible. They describe the core components that work together to make automation intelligent and use cases for utilizing AP automation to support and enhance the capabilities of finance departments.
Subscribe to Automate This on your favorite podcast provider & never miss an update!
Read the podcast transcript:
RYAN: Welcome to Automate This; the podcast for conversations about accounts payable and beyond. This is episode three where we’ll talk Tech. My name’s Ryan Nelson. I’m the Chief Operating Officer of Goby. Helee Lev, usually my cohort in this is out doing other productive things I’m sure. So we’ll be without her today, but see if we can’t have a good time anyhow. I think we will because we have a special guest today to help us through this topic. Twenty plus year old friend of mine, business partner in a couple ventures specifically our current one here at Goby, the Chief Executive Officer of Goby, Chris Happ. Welcome to the podcast.
CHRIS: Hey! Thanks for having me. My first one, I’m excited. I don’t suspect that I’ll be as able of a guest or a host as Helee, but I’ll do my best.
RYAN: Yeah. That’s a high bar, I’m sure.
RYAN: So interestingly I mean I think we tried to address this last time, but accounts payable and beyond, we’re still wondering what the beyond is. So if you get to that which is interesting because last night for something to do on a Sunday night, I watched Toy Story 4.
RYAN: Have you see this one yet?
CHRIS: I didn’t. I was still on two or three.
RYAN: Yeah. They’re on…
CHRIS: So, yeah.
RYAN: I think it’s four, but it’s that Buzz Lightyear. He’s like, “To infinity and beyond.”
RYAN: So I don’t know. AP and beyond. I don’t know. Maybe it’s today. Maybe the Tech is the beyond because it is a little bit beyond which is kind of maybe where we can start. So we’re going to talk Tech. We’re talking about applying Tech specifically to accounts payable. And one interesting topic is, let’s say I’m an AP practitioner, do I actually need to know this kind of level of detail?
CHRIS: Yeah. That’s a great question really. I think you do. If you’re an AP practitioner, I think
you think about, just like any person does, how do you spend your day. And maybe this is a little bit nerdy, but anything I do, I kind of think how do I make this more efficient whatever I’m doing. And so AP has historically been a lot of repetition of manual tasks. I think in your head, you start going, “Are there better ways to do this? Tech, can it help me here?” Maybe I’ve tried things in the past and didn’t quite pan out, but it’s been three, four, five years. Maybe you know a lot has happened in the last six months even. So is Tech something I should be looking at? How do I stay on top of that? That’s my personal belief on it.
RYAN: Yeah. No, I would agree with that. It’s important to understand this. I think what we’ll find today is some of the things that I want to have you walk us through, probably you don’t need to know if you’re an AP manager or you know even a CFO. We’re going to talk about RPA, machine learning and AI and like how this stuff works…
RYAN: A little bit which is interesting…
RYAN: But yeah, totally agree that it has changed. Some of the stuff’s fairly recent.
RYAN: So unfortunately, maybe for some people or providers out there and they’ve had a bad experience three, four, five years ago or maybe even two years ago. Or maybe built stuff that you know sadly isn’t the latest, greatest stuff already so that can be challenging, but the idea is to solve our business problems too.
RYAN: Of course. Right.
RYAN: That’s what…
CHRIS: Well, you hit a good one. We talk about this all the time here is technical debt. So you’ve invested in something. At what point do you stop throwing more money, good money after bad? It doesn’t mean it was bad. It was probably right at the time, but at what point do you just say you know what? I’m moving forward. I’m not paying off this technical debt or I’m going to just accept the fact that technology changes and kind of embrace newer, better technologies regardless of what I’ve done historically. And I think that’s a really valid point though that you bring up on how do you make that decision.
RYAN: Yeah. I imagine that’s a buying decision that a lot of prospects of ours and people in the S and B or midmarket are having right now…
RYAN: When they’re looking at this kind of stuff. So let’s get into the meat of it a little bit. I think we could talk about this for a long time. We’re going to try and keep it to reasonable podcast timing, but I did want to talk about some of the stuff that I know we think is fun and it comes up a lot. What is RPA? Are robots doing this? What is AI? What is machine learning? Are all of these things that are happening, do they mean different things? Do I have to weight one over the other? You know how do I approach this? Let’s say I’m even the CTO or the CIO helping a CFO make a decision. And I’m the CFO and I’m saying, “Hey! There’s new tech. They got RPA, but they don’t have AI, but they have machine learning and predictive analytics.” What do these terms I guess specifically mean?
CHRIS: Yeah. We’re going to go deep here on this one. And I like your point before of do you actually care as an AP practitioner. It’s kind of the beauty of the cloud now, right? Remember when we started Goby, we used to be in here you know setting up the routers and the printers. And as you get more and more into the cloud, a lot of that’s done for you. I think Apple has helped a lot in our thinking of that of just do it for me. And so I think it’s good we’ll discuss all these topics, but then ideally you’re looking at software platforms. If you don’t have a large IT staff, these software platforms have figured this out and put all these together for you so you don’t have to necessarily understand it at that level of detail or invest in it. Some of these aren’t investable at a smaller scale. They scale better in the cloud.
RYAN: Agreed. This is a fun to know, not a need to know…
CHRIS: Okay, good.
RYAN: In our opinion, right?
RYAN: If you’re trying to improve your AP process. I mean to be savvy about it’s good, but yeah I agree it’s not a need to know.
CHRIS: Yeah, definitely. At the dinner party, you want to be able to you know drop these nuggets on people.
RYAN: Yeah, definitely.
CHRIS: I think that’s it.
RYAN: It’s impressive at a dinner party…
RYAN: To be able to articulate blockchain…
RYAN: Just one level layer deeper than the person next to you.
CHRIS: That is good.
RYAN: But no, you’re right. I mean we literally did you know have some downtime 18 years ago because we tripped over the server…
RYAN: And knocked the cable out, right?
RYAN: It was in the closet at one of our apartments and it was a true thing to have to worry about. Someone you know pulled the plug…
RYAN: Kind of accidentally. So it’s fun not to have to worry about that. Well, yeah. I want to start with blockchain. What is blockchain?
CHRIS: Yeah. To kind of distill it down, it’s a trusted network, if you will where the trust is built on underlying or basically securing a dollar with an underlying trust algorithm that’s built as opposed to a physical piece of gold or what it used to be where there was some physical item behind it. Now, we actually have a trusted network that is basically reserving that amount on this. We’ll call it unbreakable chain because of the algorithms put around it. So it’s a modern way of basically assigning value or storing value indefinitely for some bit of work that was accomplished.
One of the interesting things we had a group come in and pitch us, and I forget the name of the
underlying company, but their thesis was one of the problems with accounting for carbon is that there is a disincentive to make major carbon reductions now. Because if you believe that some sort of carbon tax will get enacted in the future, you’re more incented to doing it then when you can get a bigger delta. Well, blockchain could solve that where I could make a carbon reduction now and reserve the fact that that actually happened in the blockchain being underwritten by someone like an EPA or something like that. And then later I have that actual value stored for history just like a piece of gold or something like that in a physical monetary sense.
RYAN: So is this currently fitting into accounts payable?
CHRIS: I think it’s pretty early days. I’d love to get your perspective as you ultimately oversee all of that side of our house, but I feel like there is a fair amount of a little bit of a leap that the general sort of industry needs to take to be able to accept this. You and me would do it, but is someone else willing to get paid by Bitcoin which is basically just the representation of the blockchain results underneath it? We would do it all day long. Remember over at the Merchandise Mart, there used to be a Bitcoin ATM?
RYAN: Sure did.
CHRIS: I don’t think it made it.
RYAN: No, I don’t think it did.
CHRIS: They probably decided people aren’t quite ready for this.
RYAN: Yeah. I was going to ask you a personal question. How many millions of Bitcoins do you own?
CHRIS: I don’t think I have any.
RYAN: Yeah. I know I missed the boat too.
CHRIS: I would love to think I had some.
RYAN: I did miss the boat, but yeah I haven’t seen blockchain work its way into accounts
CHRIS: If you were to pay someone with a Bitcoin, would they even take it?
CHRIS: You know Subway, someone was going to take it?
RYAN: I don’t see it as a practical thing.
RYAN: In today’s accounts payable, when we talk about some of these other things that are very practical are part of Goby’s solution for sure as well as some of our competitions we know, but no one is touting like how blockchain has made their AP solution significantly better. Or CFOs are out going, “I need a blockchain AP solution.” I’m not…
RYAN: I don’t think any of us are hearing that. But RPA, I’ll go ahead and say I believe it’s Robotic Process Automation. RPA, useful, right? Part of the solution. Maybe tell us how that might be practical to AP.
CHRIS: Yeah. I think if you look at any repeatable task and what it costs a person anywhere in the world to do it and the AIR rate associated with that and then you apply a machine that can do it so a robot that can do it. And it doesn’t have to be a physical robot. I think that could the misconception in that there’s, what was that movie? Number Five is Alive?
RYAN: Johnny Five Arrives?
CHRIS: Johnny Five, yeah.
CHRIS: It doesn’t require that person…
CHRIS: And they’re moving things around. The computers are doing it with software and
CHRIS: We can mimic…
RYAN: It could be probably a laptop or something.
CHRIS: Yeah, that’s right.
RYAN: Not a cloud-based platform, but just…
CHRIS: Yeah. So it’s not truly a robot in what you think of or historically what you thought of it. So if you can create a process where you can replicate it with a machine or a set of logic really then I don’t think you would ever do a cost benefit where humans could do it cheaper in scale. But you know this better than I do just from all the process work that we’ve done over the years. The harder part is how do you design the process that a computer can follow?
RYAN: Right. Right. Yeah. Once you’ve got it down to the proper logic which you could design not thinking about the fact that a robot’s going to do it. You design the process and then you actually do put the time in for the air handling or exception handling which again is just a human process thing. Well, what is this happens? Oh well then you should do that. Well, what if this happens and that happens? Well then you should do that. So if you can actually have that built then a robot, a faceless robot, RPA can do it.
CHRIS: And you probably nailed it right there. Everyone says, “Oh RPA, I don’t have to touch it.” RPA, what do you think? Best case you don’t have to touch 80% of them…
CHRIS: Ninety percent of them. The air handling is the most important part. So I’ll let the computer handle the fast balls down the middle as our Head of Product said because he was a pitcher in baseball in college. That other part’s the most important part because those are the ones that are truly you want someone looking at and not a bot trying to make a valued decision on a non-binary set of circumstances.
RYAN: Good. So we’ve got blockchain, our wild, crazy auto trail money you know substitute for money that maybe is not a big day-to-day thing just yet. We’ve got RPA. Maybe it’s identifying that an invoice line item says something specific so I know every single time I’m now going to add this data point like a general ledger or an alert to this record, right? Just every time you’ve got this process that has just been automated. It’s not really doing any thinking for you or whatever, right? What about AI then? Is that where we start thinking with the robot or what do you think artificial intelligence is as opposed to maybe something like RPA?
CHRIS: Yeah. Isn’t it kind of funny that we call it artificial? Because if you think about the fundamental of it is that you create an algorithm or a bit of code or a process whereby a computer or something can teach itself, but you taught it how to teach itself. So is it truly artificial intelligence? I mean the limitations are what the human or whoever is programming it can think of. So by nature it’s not artificial, it’s limited only by our creativity. We could spend forever on that and that’s probably more of a debate that’s held in the pub than it is…
CHRIS: Here in the recording studio.
RYAN: A lot of soul in these things.
CHRIS: Yeah, exactly. At some point, it becomes a person. But yeah, I think that’s it. Is how do you trust that you’re going to teach a process to something. And the key again is how do you teach it and then let it learn based on the guide post that you’ve set up. And then what checks do you put in place to make sure it’s not continuing to learn a bad behavior. Much like a person, you’ve got to kind of take a checkpoint every now and then.
There was the great, it was just, it was probably about a year ago, but Facebook was making its AI layer public. And I’m going to probably mess up some of the details, but the net of it was that they had two bots trying to teach each other and they were going to negotiate. And the input was you value a banana at x, I value an apple at y, you have this many bananas, this many apples and let them trade. And they would communicate and just build their own language. So by the end, instead of saying, “I value an apple, I’ll trade you this,” they just said “I, I, I, you, you, you.” So they had built their own language that was hyper-efficient to trade these apples, but everyone freaked out because they started communicating you know in ways we didn’t understand, but yet they ended with the desirable outcome. So Facebook had to shut it down because certainly people were fearful of how they communicate.
So you get in this kind of interesting thing about well artificial intelligence by nature, you’re letting them kind of go learn and do their thing. So what guide rails do you have to come back to and say, “I don’t want them doing this, but I don’t want to do all this work either.” So at some point, do they become smarter than we are through this or not? You know maybe they know the GL code to put an invoice to better than you know Bob in accounting did because of you know all these factors that it has learned.
RYAN: Sure. Bob’s saying, “Well, we should do it this way” and the machine is going to go, “You would think that, wouldn’t you Bob?”
RYAN: “But actually, that’s going to be a problem…”
CHRIS: “I made that mistake before, Bob.”
RYAN: “Downstream that you can’t. I just ran 10,000 scenarios real quickly, Bob and you didn’t do that. You ran the one that would get you to lunch quicker.”
CHRIS: Yeah, totally.
RYAN: And all we’re trying to do though is give Bob better things to work on than make that decision.
CHRIS: The 10%, yes. And then ideally, he goes back and trains that in with the algorithm. The AI teaches the bots at some point.
RYAN: I mean that sounds like, what you’re saying there I think I’ve been in negotiations that sound like that, “I, I, I, I. No, you, you, you.” So I can see how it went that way. And actually this is also relevant, but it was a Silicon Valley episode recently where the one who was getting bugged too often with questions so he just wrote a bot that was doing the messaging you know?
CHRIS: Yeah. Yeah.
RYAN: And then finally, a guy caught him. One guy was over getting water in the kitchen, but he was communicating at the same time…
CHRIS: Oh, I see. I see.
RYAN: So finally the light bulb went off. He’s like, “Wait a second, this is a bot. This isn’t Gilfoyle.”
CHRIS: Yeah. There was that movie, it was called Multiplicity and I don’t even remember
finishing it. I don’t know if it was that good, but the concept was Michael Keaton’s character kept replicating himself, but each time he copied him, he got a little bit less, a little distorted. So yeah, these bots are way more efficient. You’re grabbing a water meanwhile it’s handling your communication. You know it’s not just the way you would’ve done it per se, but is it good enough? And that sort of thing. You know one of the questions here is, “Is 99% good, 98, 97?” You know at some point, it’s just not good enough. The copy is not good enough, but if it’s highly accurate then…
RYAN: So many incredible, it’s one of my favorite types of movie genres to watch. You know time travel…
RYAN: And just multiplicity and this kind of stuff. There’s a good Black Mirror episode of it just recently which is a great show.
CHRIS: I’ve got to get into that.
CHRIS: People have told me it is, yeah.
RYAN: I almost want to like require everyone here to watch it…
CHRIS: That good?
RYAN: On a regular basis.
CHRIS: All right.
RYAN: It’s not good, it’s just that relevant. It’s literally that you know we always ask the questions, “Oh because of this tech, we’re going to turn this way someday.”
RYAN: And they just play that out every single episode you know. It’s like well what if you
could just look at someone and you could see their quality in a number above their head?
RYAN: You go, “I don’t know. Let’s play that out.”
RYAN: And they play it out. It’s the black mirror. Like it’s almost always you know…
CHRIS: Yeah, yeah.
RYAN: Just kind of…
RYAN: Like it plays out. So it’s…
CHRIS: That does sound like something I’d like.
RYAN: Very, very creative. All right. So now we’re into AI and machine learning where you kind of blend that there a little bit. I think we did and it kind of does.
RYAN: You know a machine can learn, it can run a bunch of scenarios, it can run future scenarios, scenarios that you would never have the volume to run and therefore it can do better. How much can it do to actually it’s maybe thinking and being more like a human? Which is the way we think it’s like that’s the ultimate success, I guess.
RYAN: It could be more like a human. The Turing test I was reading about recently, I think Turing or something like that. That’s where a machine or you ask questions to a bot essentially. Well, you don’t know if it’s a bot or a human and you’re asking questions to figure that out or whatever. So it’s basically can you put something behind a screen that’s a bot and a human will ask it a bunch of questions and the human will go, “That’s definitely a human back there.”
RYAN: You know that sort of thing. Like how far…
RYAN: Can you get? How well can that bot answer questions to trick you into thinking it’s a human. And I think someone tried it. I don’t know if you saw this or maybe you did. Someone tried to do that to Goby. They went on our website, pulled up one of our chat windows…
RYAN: It may be a first. We’re human and helpful. We’re very quick to put a human on there…
RYAN: because we wanted to make sure that we…
CHRIS: Our nature, yeah.
RYAN: Capture and you know don’t miss anything. And so at first we say you know, “Hey. What’s your name? How can I help you?” Very quickly, this person said, “Do you poop?” So…
RYAN: They were just getting right after that you know.
CHRIS: That’s a good…
RYAN: Are you a bot or not? Do you poop? If you will.
RYAN: But that’s a very easy one…
RYAN: To lie about. So it’s not a good Turing test question I wanted to tell this…
RYAN: This gentleman.
CHRIS: Right. You know comical, yeah. Right?
RYAN: Yes, right. Good. Well, anything else you wanted to add or kind of summarize as AP practitioners and team members and CFOs as they start to approach automation. That’s kind of what we’re talking about. They approach automation and there’s certainly a layer of tech in there. Any gotchas or any words of wisdom as they take that journey?
CHRIS: I’ve gotten into this notion of agile which is I think gets lost in this like programming sense where you develop, you know think of things in a yearlong and build a huge project plan for a year or you kind of use smaller, more manageable chunks. I think sometimes you can look at this stuff and get overwhelmed or try to understand it all. And then make zero decision because you want to understand every bit of it. Whereas we buy out from the concept that it costs me 20 dollars to process an invoice just with people. And that by mixing people with technology, I can do it let’s say at you know half or even less than that. So then why wouldn’t I just start with something and prove that thesis out; that sort of being the agile methodology. Why don’t I take some very manual invoices or take some that are tough for people and just try it out and see what happens and then prove that thesis. Particularly now with the cloud where you don’t have the investment and infrastructure that you would historically have so I’ll end my sales pitch there. Notice I’m not saying Goby is the only one that can do that. We certainly believe we’re the best at it out there, but there are a lot of ways you can accomplish this and it’s independent even of the vendors like us. It’s really like starting to go down that path.
RYAN: Yeah. I think we’re in a world where being agile is easy and usually a good answer. You can start easily like you said.
RYAN: So why not?
RYAN: Maybe that wasn’t always the case, but now it’s easy to start without huge commitments and fear of creating more technical debt in a process that even it’s…
RYAN: Good. Good. Well, that brings us to the usual hilarious concept we have here. First time for you. It’s now called, it was called like “Craft Brew” or “Craft Coffee” but our editor threw out the name “Beans or Beer.” So now it’s called “Beans or Beer.”
CHRIS: That’s a good idea. I like that.
RYAN: Yeah. Yeah. I do too. So I’m going to name somebody who is passionate about making either of those things and you’re just going to tell me with logic or pure guess…
CHRIS: Got it.
RYAN: Which one you think it is if it’s beans or beer.
CHRIS: Got it.
RYAN: Bosacki’s or maybe Bosacki’s.
CHRIS: That is definitely beans.
RYAN: That is beer.
RYAN: So I got you on that one. I thought it was interesting kind of talking about you know our friendship and time together. I connected this to Mundelein. It’s in Mundelein.
CHRIS: Where I was born and raised and lived?
CHRIS: You know twenty something…
CHRIS: Years of my life. Bosacki’s?
RYAN: You don’t know Greg and Bridget, I think it is. Bosacki?
CHRIS: Greg and Bridget, sure.
RYAN: Well anyway…
CHRIS: When did they move there?
RYAN: Anyway it’s their story here looks nice. I don’t know. Let’s see back in 2000, yeah.
CHRIS: Oh man.
RYAN: You were gone. You were gone.
CHRIS: I probably just left, yeah.
RYAN: Anyway Mundelein I’m sure is still a nice town. And from their story here, this sounds like a nice local beer brewery. So maybe enjoy it next time you’re visiting your folks or whatnot.
CHRIS: Yeah. Interesting. You know I kind of had it with Mundelein when they sort of were going after building a downtown and then didn’t make it pedestrian-friendly. So maybe this is post and so there you know was a reason to not have walked and found Bosacki’s…
CHRIS: Was, yeah.
RYAN: Well, if nothing else hopefully Mundelein will benefit from our guest today.
RYAN: Thank you, Chris, Goby’s CEO for joining us. This is the end of episode three of Automate This; the podcast for conversations about accounts payable and beyond. I’m Ryan Nelson. Thanks again.