Meet Andi

 

On rare occasions, we’ll dedicate an episode to something new in our pricing platform, but it’s really got to be special to stray from our usual programming.

The background behind the creation of Andi fits the bill. In this podcast, Jim Young and Carl Ryden talk about how banks can unlock the potential of artificial intelligence and machine learning.

   

 

Helpful Links

Podcast Transcript

Jim Young: Hi and welcome to The Purposeful Banker, the podcast brought to you by PrecisionLender where we discuss the big topics on the minds of today’s best bankers. I’m your host Jim Young. I’m joined today by PrecisionLender CEO Carl Ryden. Before we dive into today’s topic, I want to give you a little bit of background on the thinking on this one.

On this podcast, The Purposeful Banker, as well as on our blog, we’ve made a conscious effort to talk less about ourselves and more about, as we bill in our opening, the topics that you care about. We build our content with a goal of providing value to the audience and we’re not here to try to sell you something. On rare occasions we’ll use these forums to tell you about something we’ve done with our pricing platform.

It’s got to be something pretty special to cause us to stray from our usual programming and today’s topic certainly meets that standard. We’re going to talk about Andi, who she is and frankly why you should care. Carl, let’s start there. Assume that some of our listeners of this podcast haven’t read our blog post on Andi or watched your discussion on the PrecisionLender University webinar, or didn’t receive our introductory letter in the email. Who is Andi?

Carl Ryden: We pride ourselves on providing unlimited consulting training and support to our clients and it’s a big part of what we do. We do it not just because it’s good for them, but really it’s good for us. We learn a lot from them, from the interactions we have with them, so we don’t want to ever put a barrier between us. We have quite a few of our clients who are really wonderful at saying, “You know it would be really valuable if you could do this. Well, it would be really valuable if you could do this.”

Then we gather those in and we do them and we deploy them back, and generally people are really really pleased when we do that because I think there’s not a lot of vendors out in the world who do that for banks, so the bar is fairly low and we jump way over it.

One of those people, and actually a couple of those people, by coincidence who would give us lots of great ideas, their names were Andy, was Andy. We have an Andy at one bank, an Andy at another bank and then Andy at a prospect who joined our client advisory board who now is a client of ours.

We would respond to things folks ask us to do, but we’d also come up with things on our own, look ahead and see where things are going and try to build those things. As we were thinking about what to build, we would find ourselves oftentimes referring to one of our clients, one of the Andy’s, and saying, “What would Andy do? What would Andy tell this lender to do?” Because we’re trying to lead that lender, that relationship manager, to a better solution and ultimately we asked ourselves that enough and we said, “Really shucks. What this lender needs, what this relationship manager needs is not software. What they need is their own Andy.”

We said it joking, but it really, some things you say are silly but they stimulate a lot of stuff if your head. You go, “Why can’t we do that? Why can’t we actually give every lender their own Andy where it’s constantly with them, coaching them, helping them get better, answering their questions, proactively offering suggestions?” We said a lot of what we have done with what we used to call the profitability, which was offering up suggestions on how to make the deal better, came from that same place, came from the same approach of, “How do we help that lender earn a better deal, earn a better return or build a better relationship while they’re doing it?”

We said, “Well, let’s think about it that way.” A couple things happened as we started thinking about it that way. It really focused our efforts and it allowed us to … It’s a funny thing about humans when you actually personify things and think of it that way. What would you build if you were building everybody their own Andy? As we head down that path, we found our ability to, we were making better and more valuable stuff. Eventually we said, “Why don’t we just surface it and call it that?” Because to be honest, the name “Profitability Wizard” was a pretty stupid name.

Jim Young: I was going to say, for those of you listening who may not have been aware ever that that was the name of it.

Carl Ryden: Yeah. We were horrible at marketing I would say. Most lenders didn’t call it the Profitability Wizard. They called it the magic green dots or those things and they all loved it and used it but they didn’t really understand it. Also with us communicating with new folks, you would say, “Well, we’ve got green dots,” and it didn’t really capture the full benefit of this.

I share this with folks because I think there’s a lot of things in banking that are that way. You name it based on what you do but it doesn’t really fully communicate the value in the way that humans can understand. I think we learn from that and when we started calling it Andy internally, we saw the value we got from it in terms of thinking about what we were building and what was possible. Then we decided to call it that externally.

Now we switched from Andy with a “Y” to Andi with an “I” at the suggestion of one of our developers, James Gieszelmann here, because it’s got the A and the I either in, so it kind of is nice. The other thing is, it is a female personification and the thought was a lot of the personification Siri and Cortana and Alexa and a lot of those things have a female persona to them and our thinking on that was somewhat simple, twofold.

One is, these companies have probably spent far more money researching this than we ever will and they probably figured something out so we’ll do that too. Second thing is a little bit unique. Andi with an “I” is a little more memorable and the other piece of it was there was a little bit of sketchiness. Should it be a female? Are we making it into a typical assistant? You can Google this, there are articles on “Is it sexist that all the assistants are female?”

We said, “Well, we’re not building an assistant. We’re building the world’s best pricing analyst and if we finally build something that is not stereotypically female, and we choose to go away from that, we’re almost worse,” so we said, “We’ll go with Andi with an I,” and that’s what it became.

Now again, this has helped us tremendously in terms of how we think about things. The idea is Andi would take over what was formally the Profitability Wizard but then take it beyond that and use the full understanding of what’s going on at the bank. Our mission is, our statement to that is, “For this lender who’s pricing this deal at this bank with this borrowing pricing this opportunity to this relationship, what are the things I could do to make it better?”

Once we went down that route of that idea, make it better for the customer and better for the bank, and we used all the data that we see to train Andi on how to do that better. Some of these have been surfaced in the app. Some of these are coming and the release we just came out with Monday of this week actually was really to set the stage and lay the groundwork.

We started with a really simple flavor of that which is providing help to the lenders using it: how do I do this? How do I do that? How can I do this? How should I do this? All those sorts of questions resolve to helping that lender and we’re continuing to improve that one but there’s a whole series of, we call these skills that we start to deploy on this Andi platform. The ones we launched immediately were a few basic ones about helping your price conversion loans, a few basic ones on recognizing when you have duplicate deposits versus opportunity versus relationship.

I showed some of those in the PLU, but then there’s some fairly sophisticated ones like predicting the usage of a line of credit based on the full picture of the opportunity and the relationship and our entire history of seeing similar relationships. One that’ll be coming out fairly soon is recommendations of full on structures of other scenarios you might evaluate. If you’re doing a 62 40 fixed, it’ll say, “Try a 31 adjustable or a 55 adjustable or a 11 adjustable or a floating,” all these things.

You’ll start to see these things, we call it getting into the mode where we’ve laid down the railroad tracks now. We’ve sent the first easy pilot train down the tracks with the help skill and a few others, but you’re going to see train cars moving down this track at a rapid pace. We call our data insights team here the skill of the month club. They’re going to start rolling these things out pretty rapidly.

Jim Young: Yeah. That’s actually a point I really wanted to get across because I will admit on the communication side, I struggled a little bit when we were first discussing it and how to communicate it to people because some of the stuff, I knew it was a big deal but I couldn’t quite … It feels like, I guess if you can talk a little bit about the potential aspect of it, and I think about it in terms of the help aspect you mentioned which is, right now Andi doesn’t have all the answers. If you ask Andi a fair amount of questions, Andi will give you the, “I don’t know. Should I put you in touch with someone?” Can you talk about that, within that prism, about the development of Andi?

Carl Ryden: Yeah. This is something we see, and banks are really good at struggling with these sorts of things, where you let the perfect be the enemy of the better. We know, we have a vision of what the perfect is. Every lender gets their own Andi, their own, “What would Andi do,” back from our old days with our clients named Andy, is the vision of what it will fully be, but we don’t know … we know how to get off zero. We call it getting off zero, so it leads to getting something there that you can begin to snowball and build on, and that’s the help.

We also think about, “Okay, there’s 100 skills that we can add in.” We’ve got a list of what we think are some really good ones, particularly utilization, “Stop asking lenders to tell you what the draw schedule is with a construction loan because I’ve seen all the construction loans you’ve done and I know what the draw schedules are so I’ll just tell you what I think it’s going to be and you don’t have to enter that,” all the way up to predicting the full potential of a relationship where it says, “How much deposit should a relationship like this and an industry like this have,” where we actually look at that larger piece and everywhere in between.

What we didn’t know was what’s the right order, what’s most important to customers and having that conversational interface where you can ask questions, so every question you ask, everybody asks, if there are some that Andi doesn’t understand yet, we use a natural language understanding AI right now to parch that and to figure out what the intent is, every question you ask, we see that on our end and our support team does.

They go through, they come back and they train Andi how to understand that so you’ll find … the first day we launched Andi, several folks asked questions. We put in every one we could think of and of course we didn’t hit them all, but as folks ask questions, even when Andi says, “I don’t know that,” know that when that happens, somebody on our side’s going in and teaching that to Andi. What you’re doing is uncovering to us what we need to teach Andi.

Every interaction you have with Andi, she’s getting smarter and more capable and even over the course of the first day, you’d see the same person ask the question at the beginning of the day, another person would ask the same question at the end of the day and they’d get a wonderful answer. We’re adding in additional layers of things and we found just the questions folks ask us surprise us and lead us down a different path than if we had taken … We could’ve said, “We’re not going to release anything until we’ve got this wonderful experience,” but we’ve just been sucking our own exhaust trying to build what we think folks want, put it out there, let them interact with it, iterate on it quickly, and we have a system where we can actually improve it really quickly.

Jim Young: That’s actually a really good point. It’s been fascinating for me to watch it to see, “Oh, that’s a lot of the questions.” It’s a way for us to kind of get a glimpse into the minds of a lot of the relationship managers minds. That’s the questions they’re asking a lot and in terms of where her level is, you mentioned James Gieszelmann, he put it in a way that was good for me to understand which was, “Would you hire an analyst?” When you hire an analyst at your bank, you don’t expect them to know immediately how to do everything at your bank and how each relationship works at your bank. You hire them because you think that person is going to be able to learn it and get it down cold very soon, and that’s basically what you’re getting-

Carl Ryden: Andi, it’s not … If you interact with Andi, it’s not just you. It’s every lender using PrecisionLender’s interacting with Andi and every interaction they have with her, she can either answer them right then or we go, “That was a good one,” or, “Oh God, she missed that one. Let’s figure out how to train her how to answer that one, and then she’s got this,” so every day she’s getting a little bit smarter. Then that’s just the incremental stuff.

Then we add on the skills that we’re plugging in and I think you’ll see her be able to do much higher level stuff pretty quickly. I think stay tuned. We said in the launch that she starts Monday. It’s her first day on the job but she’s a really quick study, and I think you’ll start to see that. Now one other kind of meta thing we do with hire, take it up away from us, we have a lot of banks who AI is going to be a big deal in banking across the board. AI machine learning, advanced analytics, predictive analytics, all those, and we’re using almost every bit of that in what we’re doing, a big part of us doing this is doing it in a way that we can prove out a pathway for how to do this, then share with our clients.

A lot of banks now are working on building AI’s and predictive models and those things, and I think finding a way as early as possible to get connected to your customer, and you might be wrong and you might not know things, the first answers might not be good, one of the philosophies we said is, “Make sure Andi knows what she knows but knows what she doesn’t know.” That’s why she’ll say, “I don’t know that yet.”

We don’t want her … We try to, the big fail is when somebody asks a question and we think we know it and we say, “Here’s a support article on …” and that’s not what they asked. They didn’t ask for a support article.

Jim Young: Right. That’s a good point. Nothing irritates me more about Siri when she gets it wrong. I’d rather her just tell me, “I don’t know.”

Carl Ryden: “I don’t know that yet,” and particularly if you know when she says, “I don’t know that,” that … With Andi, when she says, “I don’t know that,” she not only says it to you, she goes to our support team and says, “Hey guys, I didn’t know this. Tell me this,” just as you would want. An Andi who was an analyst who started at the bank to do, “I’m sorry. I don’t know that yet. I’ll go check with the other team of analysts and figure out how to do this,” so she’s doing that every day now. It’s kind of fun to watch.

Jim Young: You mentioned about banks and that AI is going to be huge and certainly if you’re looking at all the publications out there, the bank directors, American bankers, all the … to the big business, Wall Street Journal, there’s lots about it out there. A lot of it, at least from my perspective, has been more on the retail side and a lot of that interaction. I’m curious about, when you started noodling on this or maybe thinking out loud about this to some clients and prospects from our angle on the commercial side, what was the reception? I think you said originally you thought, “You might think I’m crazy but …” but what was that reception like?

Carl Ryden: This happened a little bit less than a year ago, so March or April I think of last year, I finally put together a slide deck that we used internally and shared with some partners and share with our client advisory board of where we were going. The goal is for us to build the most important parts of the brain of the bank and to really accumulate, assimilate the intelligence. We get it through interactions with questions but we also get it through interactions through the application and through the data we gather to really assimilate that intelligence and to be able to deliver it back in a consumable and valuable way.

Well, in April of last year I put together a slide deck where I talked about how we were going to do that and how we were going to use AI and machine learning to do that. I had a slide in there that said, “You might think I’m crazy but the 5 top technology companies in the world are literally spending almost a billion dollars a year on this and it’s going extremely fast.” Microsoft, Google, Facebook, Amazon, IBM, on down the road, it is a huge huge thing. I put that, “You might think I’m crazy,” because it was kind of new then.

Six months in I had to take the “You might think I’m crazy” slide out because I’ve been in the technology commercialization business for some time, I used to do this consulting technology strategy commercialization for Bane over in London and this is the fastest I’ve seen something go through the 3 phases of adoption. My 3 phases are, for those who don’t know, the first one is any new idea goes through 3 phases and if you pay close attention, you can ring a bell as it moves through those phases in the organization.

The first one is dismiss this foolishness and often ridicule. “That’s a silly idea. Who’s going to share every detail of their life on the internet for all people to see? That’s just a dumb idea. Let that go.” “Stay at Harvard,” if I’m talking to Mark Zuckerberg. The second phase is opposed sometimes violently. “Look, you’re getting ready to make a big mistake here and it’s going to hurt. You’re going to misannounce your opportunities, whatever, maybe your parents told Mark Zuckerberg, ‘That’s a really dumb idea,’ but now you’re going to do damage and I’m going to oppose you from doing it.”

The third phase is accept it as having been self-evident all along. Of course we all knew Facebook was … I had that idea too. I knew it was going to be a big hit and everybody knew that. Everything moves through those 3 phases. I have yet to see anything in my life more through those 3 phases as fast as artificial intelligence machine learning, particularly as it applies to banking. It’s been particularly astonishing in, the reputation of banking is slow moving.

Jim Young: Exactly.

Carl Ryden: It is anything but. I think what I see among some banks is they have for the past 7 years, they’ve been paring things back and cutting expenses and cutting overhead because rates were really low and they were being squeezed a lot and they shed a lot people, a lot of … if there’s any fat there, it’s gone. Then it was kind of all they could do to hang on. Well now rates are going up and the wind’s turned around, it’s a little bit at their backs. The regulatory stuff has been lifted or at least seems to be heading in the right direction for banks.

Now I think a lot of them are like, “How do I now let … I went through the pain of leaning down, how do I not fatten back up?” So you’re seeing folks make a lot of investments in technology. I think one of those avenues is artificial intelligence. Now the last thing I’ll say about this on this area is I think it’s really important to start at your customer and work back and build it around the customer experience. Of course, we believe this.

We wrote a book called Earn It, so you want to start there and have that attitude and philosophy that, “We’re going to build systems that help us create more value for our customers.” The place that you can go that we’ll call short-term, that could call short-term benefit but long-term detriment is I could build an AI and a machine learning algorithm, predictive analytics algorithm to see where can we slip a couple basis points by a customer and they not notice? Where could we get away with it?

On the one end you got “earn it”. On the other end you got “get away with it”. You can actually choose to go either directions. This is why we have a conference called BankOnPurpose. I think AI will amplify whoever you are.

Jim Young: Interesting.

Carl Ryden: It is not an end. It is a lever and it amplifies the purpose you’re trying to serve. It amplifies the culture you have. It amplifies how you treat your customers and if you treat them with the, “Slip it by. What can I get away with,” it’s going to amplify that but it also makes it a lot more detectable. If you’re not coming at it from an honest point of view, you’re going to get exposed. Just know that this thing is a big magnifying glass and you have to be thoughtful about how you deploy it.

Jim Young: Yeah. It is interesting. We talked before about that almost defensive shell that banks were in, and understandably so from a survival mentality, but how quickly they have shifted now to sort of the, “I just want to survive,” to “I don’t want to get left behind.” It’s turned in that direction.

Carl Ryden: The 31st, we have Justin LaFayette from Georgian Partners is doing a BankOnPurpose webinar series.
Jim Young: That’s right, so this podcast is coming out or by the time you’re listening to it, it will be Monday the 30th, so this would be if you’re listening on Monday the 30th, it’s Tuesday the 31st.

Carl Ryden: Right, so we’ll put in the link to where you can sign up and see, there’s a BankOnPurpose webinar series done by Justin LaFayette. I’ll be on there with him. He’s presenting a framework for how to think about AI in banking that I think is absolutely tremendous. I saw this presentation at their conference and the reason Justin is an investor in PrecisionLender is because A) this is how we do it and it’s also how he believes it should be done, and he’s informed our thought process and expanded it, and crystallized it. I think it’s a really valuable presentation. He’ll be covering it live. I’ll be there with him. We’ll be asking questions and having a go back and forth. Then his presentation deck will be attached to that webinar afterwards, so I would highly recommend that.

Jim Young: Yeah. You may, for those of you who are regular listeners, Justin was actually on a podcast with Carl earlier so if you enjoyed that one, I can guarantee you, you will get a lot out of that webinar. You can go to our website and to our resources page and go to our upcoming webinars. Register for it there and if you happen to listen to this podcast a couple days after the 31st, we’ll have that recording up there. You can check it out there.

All right. Well, that will wrap it up for this episode. Carl, thanks for coming into the studio to chat. Remember, you can always find more information about today’s episode at precisionlender.com/podcast. If you like what you’ve been hearing, make sure to subscribe to the feed in iTunes, SoundCloud or Stitcher. We love to get ratings and feedback on any of those platforms. Thanks for listening. Until next time, this has been Jim Young and Carl Ryden and you’ve been listening to The Purposeful Banker.

Previous Article
Love Your Data [Podcast]
Love Your Data [Podcast]

Next Article
Designing for Humans, Not Averages [Podcast]
Designing for Humans, Not Averages [Podcast]

It’s common practice in the banking industry to design for the average customer. In this podcast, Carl Ryde...