RedPoint Global : How AI and Machine Learning are Humanizing the Customer Experience of the Future

RedPoint Global Video Transcripts: How AI and Machine Learning are Humanizing the Customer Experience of the Future

Transcription Below:
Hello, everyone, and thank you so much for joining us for today’s webinar, how A.I. and Machine Learning Are Humanizing the customer experience of the future. I am Sue Murray from Redpoint Global and I will be your moderator for today’s event. Before we get started, I’d like to announce a few housekeeping details. This presentation will be 30 minutes and is being recorded linked to today’s event will be sent to you via email within the next 48 hours. Please keep in mind that you may also ask a question at any time during the presentation by typing your question into the Q&A box.

In the interest of time, we will be sure to email you a response to your question after today’s webinar. I am pleased to introduce our guest speaker, George Cornetto, chief technology officer and co-founder at Redpoint Global. A former math professor and seasoned technology executive. George has more than two decades of business and technical experience. George is responsible for directing the development of the Redpoint customer engagement hub Redpoint leading enterprise customer engagement solution. At this time, I would like to turn the floor over to our guest speaker.

Welcome, George. So we’re going to talk a little bit about humanizing experience with Guy. Well, first of all, let me clarify, we’re not trying to humanize all experience with some human experiences aren’t so good, right? If you get a cold pizza folded in the box three hours late. We don’t want to repeat that. But the kind of humanizing we’re trying to do is the kind that you might think of. Right. It’s, you know, why, why and what kind of experience are we trying to humanize?

So let’s set the Wayback Machine, Mr. Peabody. And here we are sort of framed from an old movie. And you got Charlie there at the counter and you’ve got Joe bellying up to the counter, kind of talking to Charlie, letting him know what he needs for the family, that kind of thing. There’s a level of implied intimacy in this, in spite of the fact that Charlie looks like he’s about to take off the customer’s head. But but there’s you know, let’s just look beyond that and think of the implied intimacy there.

You know, maybe Charlie Grool saw this guy grow up from a kid and seen his kids grow up. But the point is, is that there’s a lot of knowledge, a lot of understanding, a lot of context. And what the intimacy is, is that there’s trust. Right.

There’s trust between these two on what to get, one to get.

And there’s not you know, he’s not feeling like Charlie’s trying to sell something that he doesn’t need. And sometimes Charlie may really even go out and find things that he didn’t even know he needed.

Right. So how do we get that really positive part of that experience realized? And and what role does A.I. play in that? Right.

So so today what we’ve got is a situation where you’ve got silos and those silos are doing the best they can to send you out emails, use a list, get something out. Maybe there’s some modeling behind it for the list. What’s who’s got the best response, et cetera. But it’s it’s not really contextual necessarily. It’s not really at the cadence necessarily that you would want. And it may not really provide the right intent behind it. Right.

So so there’s a lot of problems with this and it affects lots of pieces. Right. So the problem is, you know, what is it if we go back to Charlie, what’s the problem here? Why can’t we get that? Well, maybe CBS are friends at CBS, have stores at every corner. Maybe they can do it through training and getting there. People all kind of familiar with the customer, local customers.

But go, daddy, go is really interested in being personalized so they don’t have stores everywhere.

So what’s what’s the inherent restraint? Data’s one. Scale, right? How do you do this at scale? Do you get a hundred of these Charlies? How do you clone him? Right. How do you get all of that history distributed out to all these folks? And then the size of the stores make these look like little toy stores. Right. So it’s all about scale. And that’s where I really comes in and helps.

So. Today, when folks sit down and build models and you get you gather up all your skills, all your skilled people, your smart people, you get some modeling tools out there. Often what you get out of it is a set of clusters that are built around personas. And that’s really not bad. That’s actually a step in the right direction.

But to get to that real level of humanized intimacy, right.

That level that feels like, you know how you’re talking to somebody on the phone and as you’re talking to them, you can kind of tell when they’re not listening. Right. There’s that some nonverbal cues that, you know, this person is not listening.

Right. And and so that level of connection. Right.

Doesn’t come through unless you get it just right. Unless the timing is right. Unless the content is right. When you do this at a cluster level, you’re not going to get it really right for everybody. You might get it for this person right there in the middle. But it’s not going to be for those people around the edges. Right. So what you’ve got to be able to do is do this at scale and do it for each individual, set a plan, make a plan and execute that plan for each individual.

OK, so now what are the challenges of that?

What is going to keep us from doing it and or what are the things we have to overcome to get it done? And those are three things. One is we’ve got to collaborate certainly across the organization, but we’ve got to collaborate with the technology. We’ve got to work with the A.I. And that may seem a little bit counterintuitive, but I’ll explain that in a minute.

We’ve got to get the data just right. And that’s really critical. And that’s something that you hear us consistently obsessing about, is data, data, data, fast data. It’s got to be all day. That’s got to be everywhere, et cetera. Data has to be there and it has to be just right to support the AI. And then the third piece is strategy. And, you know, I’m the CTO and, you know, I worry about technology all the time.

But all I see is if you can have the best technology in the world, the technology should really be invisible. Your strategy should be what is dominating everything. So think about that strategy and I’ll show you another slide in a minute. Why it’s really critical that you got that strategy down, right. So why? Well, let’s think about it. A.I. gives you that scale. It gives you scale. It gives you a tremendous level of power to derive insights.

And frankly, it does it better than humans. You know, this is all about the algorithms aren’t necessarily new. This is really all about machine power. Right. Machines are more and more powerful every day. They can just run through iterations faster, consume more data and really perform better than humans in lots of settings where decisions have to be made, where information has to be derived, etc.. The thing is, though, if we work together with the A.I., it gets even better.

Can you give any hints?

Those hints are in the form of constraints, you know, or their prior knowledge that you have to focus on one area of your customer base rather than another.

All those hints actually really helped to make A.I. even stronger than it would be on its own.

So an example of that is things like chat bots, chat bots and digital assistants. They are usually a combination where I can take it to a certain level. But then the chat bot is further trained by experts and that actually delivers a tremendous level of performance. For those of you that have applied chat bots to your call centers, you’ll know that they can be very successfully deployed and reduce a tremendous amount of the expense that would normally go into a call center.

And then the data. Well, you know, we obsess about data day and night. The data is all about that CDPR. And what I will tell you is that where we have been investing over the last year is in crushing the latency between the moment of ingestion and the moment of activation. Right.

If you look at the bottom, if you look at the bottom of this picture day or the side over on the left, you see that data come in. Well, that data comes in and there’s that golden record that we want to build. And what we do is we take the golden record and build it and you take that transactional tail and build that. The key is, is that there’s more data going to be. Coming in all the time through our batch and real time or streaming processes and what you’ve got to do is crush that time.

And what we’ve been able to accomplish in this last year is the ability to update the gold and record both sides of it to the identity graph in real time. And that transactional tale and all the aggregates that you use to measure and monitor and figure out the customer value and everything else, we can actually calculate those on demand in real time. So we’re taking that latency and crushing it down in such a way that nobody can match that level of performance.

And that gives the A.I. the best possible conditions to deliver magic at the point of engagement with the customer.

Now, why do I emphasize strategy?

Well, it’s because A.I. works, but A.I. and the way we use it with which is a combination of machine learning and optimization, requires that you give it a fitness function. What’s a fitness function? A fitness function, really, in its most simple iteration is just the metric, right?

You got to give me a metric. The metric you want to optimize that can be something as transactional or tactical as a clickthrough rate. And we can optimize it with the with machine learning and optimization. Or it can be something as global or strategic as your overall are or why any of those are fair game for our machine learning. And it does work. But what that means is you better get the right metrics. If you don’t, it’s going to go there anyway.

And all of a sudden you’re saying, oh, no, wait a minute, there’s the expected unexpected outcomes if you pointed to the wrong metrics. So it’s really critical, again, to have the right strategy. Strategy defines what you measure. Those measurements are the metrics that you optimize with A.I. or with our machine learning.

So really important part of it. So finally, how to embrace this, right.

How do you how do you take advantage of this capability and really make it your own right and make it part of the the.

Fabric of what you’re doing so well, the first thing, of course, is can work on the Red Point platform. And why is that? Well, the reason is, is that we’ve integrated these things exactly with this level of process in mind.

Right. The bottom line is, is that, first of all, the data we obsess about the data, we talk about that all the time, how to bring the data together in any way possible so that it’s unified across the organization. Because what’s going to happen? Think about it. What you’ve got is lots of touch points in the organization.

And those touch points are how that picture of Charlie and the guy leaning on the counter. That’s how it’s happened. Although most of the time now it’s mediated through technology, whether it’s your website or your mobile app or the call center or whatever it may be. Sometimes it involves a person, but not always. So that data has got to be brought all together so that in this world of multiple touch points across the top, what you’ve got is a single brain, a single kind of organism.

The data is the lifeblood of that organism. So you got to bring it together. And then what you do is you launch the A.I. against it. Right. And we call this in line analytics for a reason. You know, a lot of companies that do analytics, a lot of the consulting companies are do analytics. What they do is they take data. They take kind of some of your plans, your metrics, whatever it is that they’re trying to trying to work with you on.

And they’ll spend a bunch of time with your data, pull the data, play with it and come up with a whole bunch of models. Right. But the problem is, is that by the time you get those models in the field, you’ve already gotten this latency from when you ingested the data and all the time they spent building the models and everything else.

So so how do you how do you fix that?

Well, the way we think about it is that A.I. is best used in the flow of the process.

Right it right in the middle of the process where data is coming in, the A.I. is running and unattended manner. Right. It’s just automated. It’s ingesting data. It’s figuring out whether that data changes any of the outcomes of the models that are under management. And then it may or may retrain, may not all those decisions can be made by the system in the operational flow. And that’s how you get the best or that’s how you get the best.

Minimized latency is by putting it right in the middle of the day to day operations, data and models churned. If there’s better models, they get pumped up. If they don’t, they just get forgotten and then they move on and then they start to drive the orchestration.

And the orchestration, of course, is what glues all those channels together. Right. And gluing all those channels together. Now you’ve got a single point of operational control and a single point of data control. In some sense, you’ve got a virtual Charlie, right?

Because now it’s one brain. It’s not a bunch of different brains. It’s not a bunch of different fragments of brains across all the silos of your channels.

It’s one brain, one set of data, one set of analytics, and the fact that we can connect it to all the channels. I was gratified to hear any good that he was talking about getting five channels currently fully integrated into the platform and they’ve got another five lined up to get fully integrated into the platform. That’ll be ten channels that completely know what the other channels are doing.

Imagine the level of coordination of precision you can execute there if you’ve got the right messages and the right strategy and then the API supporting that.

Right. So there’s some real magic there. And and again, I want to just go back to this idea of how, you know, someone’s not listening on a phone. Right. It’s really hard to figure out what it is, those little nonverbal cues. You know, someone’s not listening. But you also know when someone is listening and there’s that click, right? There’s that connection that happens. And that is very hard to achieve with technology.

But we can and it happens when we’ve got the one single brain sitting behind all this, determining things like cadence. What’s the kaiden? Charlie’s talking too slow, you might think he’s not listening. Charlie’s talking too fast, you may start to wonder what he put in his coffee that morning. Right. So, so the thing is, there’s that natural cadence that everybody has a different version of. Right. The machines here will figure that out and then execute on that.

So. Now, what does that really look like, so. OK, that’s the art of the possible. Now, how do you go and actually get this done inside the application? And this is a real important aspect of it. Well, there’s this little nugget in there called message hlas, because message lists are a secret weapon in generating and letting A.I. integrate with the application, the message list will actually define what is that path to purchase for each individual.

Remember, we talked about individuals as opposed to clusters. What is that path to purchase? And then what the message lists allow you to do is then capture that path for each individual.

And then what happens is that every single person has a message list and has a set of messages that are very strategically defined by the A.I..

What is I guess, really unique about the message list is that they do not coupled the channel with the message. Those messages stand alone and deliver through any channel that the person may choose to visit you through.

And that is a very unique concept. The decoupling of message and channel is something that is I don’t know that anybody else does that, but it’s really critical because that’s part of the siloing in the chaos that we live in today is often that, well, I’ve got this email with this offer and it’s just sort of as one kind of monolithic thing, the email offer. Right. As opposed to the offer that Steve needs and the email or website is just incidental, just happens to be the way he gets that message.

Right. That’s what message Lascelle will do for you.

They’ll allow you to go through model every single individual in your organization, select the right next messages for that person and then hold them until that person shows up to give it to them through the website, through personalized interaction on the website. If the person instead shows up at the call center the real time in general. No, wait, that Steve give message number one at the call center. And if it’s an outbound email that I haven’t heard from Steve for a while, it’ll be message number one.

Now, if Steve happens to come in to the website five minutes before I’m going to send you message number one, the machine knows the stop message number one from going out on the email and send a message to if you can.

And if you can’t, you just wait a little bit and then send a message to. Right. But the whole idea is that now it enables you as the marketer to market, to engage, to have that type of relationship on mass at scale with your customers.

Right. So this is one little nugget.

And so for, of course, you start talking to people and that level excuse me, of personalization and what you get is an enormous lift in terms of engagement and in terms of revenue and everything else, because people really feel like there’s that closeness and that trust. Right.

That closeness and trust so much about timing, about message that it’s really delicate.

And this technology, the message Lois and the A.I. brought together, lets you deliver that very delicate message, that very specialized message at the right time.

All of that because the A.I. even figures out the cadence for you. Right. How frequently should we be talking to somebody, etc..

So so there’s a lot there to unpack. And I do recommend that we will we’ll figure out how to communicate a lot about the message lists and some training about that so that you can take advantage of it. And it really, like I said, when you add the analytics component to it, it really starts to look a lot, maybe even a better looking version of Charlie. I don’t know, because you get to put the face to it. Right.

So now I just want to touch base a little bit on our A.I. studio because, you know, this is the core of it. Right. OK, this is great. I want to come up with all these next best messages. So how am I going to do it?

And listen, I know that the term A.I. is hyped beyond all recognition. And frankly, you know, it’s a bit of a misnomer. We really focus on machine learning and optimization. That is what. We do, and we do it exceptionally well, and so what we what we have here is the ability to manage those models, because if you’re going to go do that level of modeling at that level of precision, at that level of individuality, well, for God’s sake, it’s going to take more than one model, I can guarantee you that.

And so this allows you to generate these models and manage them easily. Right. As easily as you as you can. Right. It gives you the ability to look at the model, see your inventory of models. How are they working?

What’s the fitness of the model? How many times has it been called through the API? Right. How many solutions were there with that model? Should we try a different solution, et cetera?

It’s a very dynamic environment where you can be very interactive with the modeling environment, with the models, and get that to be a really refined cadence of modeling generating models, automating the process of updating the models. Because one of the things that happens is you get new data. Well, do we have to go and drag out the data scientists and all this kind of stuff? Well, actually, no, you can actually automate that. You can say when new model comes, new data comes in that affects this model.

Just trigger a refresh trigger or retrain. Right. So a lot of automation available here to take the load off in terms of the modeling, it’s very, very powerful. And just to show you how you know, how interactive it is and how easy it is to get to that optimization and work through that, you’ll see here, you know, these are the kinds of parameters you see on the left. You’ve got a wizard that can walk you through those steps.

You’ve got your run parameters. So run settings automate, you know, and do we automate on time? Do we refresh models every day, every two days, every hour, every minute. I know it could. Last I heard, there were refreshing some of the models every hour that works. And then you can have triggers affect the refresh of the models. And then there’s all sorts of other parameters down here. And what I can what I really want to encourage you to do is if you do decide to work in a studio, just give it a shot test, test some models out there.

It has been an incredible inspiration to me personally to hear about the modeling work that Karig has done with our tools. They just pick them up and said, let’s give it a shot. Let’s try some of this. And we still get calls asking us what can you do this kind of thing and can we turn it on its ear and do this kind of thing?

And it’s just been great, right. But it’s a way of just starting and learning. If you have a whole data science team, well, that’s great. We can peel this front end right off of this. And now what you’ve got is a service interface that will let you embed this modeling and all of the various features of these models into any application.

So we’ve done some of that with RPI. We’re doing some more of that with data management. And the fact is that is where the future lies, right? The future is A.I. everywhere. The future is embedded, machine learning everywhere. Every little decision, every little nuance of what might be happening in a system is going to be somehow driven by some level of intelligence. So you can use this on either end, you can use it as an interface, or you can peel the interface off and get right down to the guts of it and embed it wherever you want across your enterprise.

So lots of power here, lots of capabilities here. But let’s think about what we just talked about. Number one, we want to get to that level of intimacy, of engagement and do it at scale right scale. But the thing that everybody here has to really bring to the table or bring to the party is strategy and a well-defined set of metrics to connect these two.

Right.

You do that, you’re going to have phenomenal success. I promise you, you will have phenomenal success. But just try get started someplace and get the ball rolling. And I think you’ll find it a very gratifying process of just starting something, rolling with it and see where you get to. And a lot of valuable. Driven that way, so anyway, so that’s it for today. Thank you very much. Again, I really appreciate all of you being here.

It’s been a wonderful experience for me and I hope it has been for you as well.

Thank you.

Learn more about RedPoint Global and Checkout Their Articles on:

data enrichment services

identity resolution

i connect data

data quality issues

mdm data

omnichannel personalization

digital acquisition

customer data management

behavioral marketing

customer profile system

personalized banking