How AI will Help Customer Conversations

Get Your AI Strategy Sorted https://activatecx.arroyo360.com/ai

In this Episode we delve into the profound impact of contact center AI solutions on humanity and the evolving landscape of ai agent assist. Learn how Balto AI integrates with various systems and ensures compliance and security in the world of data handling. Subscribe to stay informed about the latest developments in AI technology!

Chapters
00.03 The Obstacle is the Way
03:28 Towing the Line
09:55 Becoming Indispensable
15:18 Staying On-Point
16:59 Can’t We All Get Along
19:18 The Files are In the Computer
23.32 Flying Cars and Silver Suits
27:07 X Marks the Spot
28.46 Getting Better

Chapters

• 0:03 The Obstacle is the Way
• 3:28 Towing the Line
• 9:55 Becoming Indispensable
• 15:18 Staying On-Point
• 16:59 Can’t We All Get Along
• 19:18 The Files are In the Computer
• 23:32 Flying Cars and Silver Suits
• 27:07 X Marks the Spot
• 28:46 Getting Better

Transcript

0:00

Hey Mark, welcome to the show. Thanks Frank. Great to be here. I'm a big believer in stoic philosophy just from the standpoint that It really shows how an individual can manage their own character for a better outcome for all the people that they are involved with and Ryan holiday has a great quote that says, never forget within every obstacle is an opportunity to improve our condition. And so it's a very. Positive look at what we face every day and how we overcome obstacles. And so for Balto AI, like what are the inherent obstacles and just human limitations that we're overcoming with the solution. Yeah. Frank, first off real funny. I actually when I do product demos, I usually read a quote from Ryan holidays, the daily stoic. And , if it takes five seconds or 10 seconds of processing time, I'll pull out a quote and read it and be like, look, by the time I just read you that quote, we just analyzed all these interactions. So the stoic philosophy, like absolutely love it. I think that the biggest thing that we're all going to need to figure out right now is like a new level of humility because we have, for as, as far as human history has been recorded, been the most intelligent beings on the planet. We are not the strongest, we are not the fastest, we don't have the best sensory perception but we have been the most intelligent. And that's something that I think we hold, we're very proud of. And I think what we have to be prepared for very soon is that our ability to think and process and problem solve will be matched by machines, by artificial intelligence very soon. And we're going to have to. Derive our value as people from other things from, simply the fact that we exist and that's just a wonderful, like, joy to, to be like the fact that, that each of us are here, that you can listen to this podcast, through speakers or through Bluetooth or through the internet, and that all of that exists, that you get to experience it is unbelievable and that there's like kind of this very. Special value that humans can create when they are by themselves. And also when they're creating with other humans, no matter what it is, no matter if it's a piece of art, that's not as good as something that AI can create the fact that is human created with your hands and the best reflection of you taking what's in your mind and how you feel and putting it on a page. That's beautiful. So I think that, like, our society is going to shift much more to valuing those sort of things and not placing so much of our self worth in, in how smart or intelligent we are, because I think that's going to be, we're going to need to have a very healthy dose of humility there. Yeah, there's no doubt. And, that humility is going to be something that either people choose or succumb to, it's this 800 pound gorilla. I really tell people all the time that AI is akin to fire. It is a paradigm shift in the history of mankind and nothing before it was the same that will be after it. So when it comes to some of the technical elements of the application there's a series of check boxes that people use to assess something that we can bring in and use and a lot of it has to do with compliance, a lot of it has to do with data security, PII. How are we handling sensitive information? Because your solution is not only looking into data that has been accumulated. Whether it be recording data, whether it be knowledge based articles, but also potentially, are you pulling data from a database? So, is there information that's coming to the table to assist the agent that may have sensitive information in it? Yeah, this is the world in 2024. Any software that you bring into your company needs to have access to some form of useful data in order to do something. And each one of those softwares has to build appropriate safeguards. For each of the types of data that they're touching. So, we're touching audio. Like actual, the audio conversation. We're producing transcripts. We're producing labels or events on those transcripts. Saying, here's all the things that happened in the transcripts. We are pulling in metadata from the CCAS and UCAS, the phone system. We're able to save notes and, you can, put notes in a CRM or, or your telephony or your quoting system. We're able to have real time APIs actually trigger either Balto trigger an event in another system, another system trigger an event in Balto. So there's a lot of data that Balto touches and there's a lot of connectivity between Balto and other systems. Standalone systems that don't have integrations are, Really not what buyers want right now. You, you want it to work with all the other stuff. So that just means that we have to assess every risk one at a time and figure out how we can be. Smart and compliant and safe and secure. So for example, PCI, PII, PHI, we scrub at all of it in memory. So actually what that means is that there is never a moment where it is actually ever even saved. It's not saved and then deleted. It's just never saved. We just don't save that that part of the audio recording. We don't save that part of the transcripts. It, we actually have almost like a a deleter or a snippet, if you will, that will snip out the audio part of the audio recording where PCI PII or PHI occurred. So in fact, the, the, what we're working on now and the crest we get from our customers is, well, can we save some of it, but not all of it? Right. Maybe, we don't you know, maybe we actually want to save the PII, the identifying data, but we don't want to save the PCI. So we're actually starting to put more advanced redaction models in place to kind of loosen up some of the guardrails, specifically where customers want to, like, enrich their data in one place or another. But our philosophy has been started as strict as possible and then like loosen to meet specific customer demands rather than, than realize you're not strict enough and then go, Oh shoot, we need to beat this up. That makes perfect sense. From the standpoint of contact center platforms there are quite a few, most of them are in the cloud. There are still some that have some on prem, but nevertheless, there's a good number of them. And many of them are opening themselves to integration with different AI solutions. Some of them have AI. Implemented and integrated into their own technologies. Big investments in AI where do you fit in that ecosystem? How do you ensure that the product works? Whether somebody is picking up a Genesys or a Five9 or a ringcentral ringCX or a nice CXone. How do you fit and work? And how do you see yourself perceived in that ecosystem? Well, out of all the players you mentioned, we integrate with all of them. We integrate with over 50 different CCAS and UCAS and there's a environment of what folks will call coopetition right now between kind of the conversation intelligence players and the CCAS players. And the CCAS players on one hand really want this open ecosystem so they can bring a whole bunch of tools in and their customers have access to everything and they can plug in any of their own tools and everything works. On the other hand, especially around ai CCAS players want to build as much of their own AI as possible. And if there's a deal sometimes, we'll be competing directly against the CC. The CC says, why don't you try our I and we say, why don't you try our ai? And there's, competition there. Mm-Hmm, . So there really is a push and pull dynamic. And, historically CCaas has been a little bit more closed off where I think that they wanted to build more of it in the house and they, would close down their APIs or if they gave you a real time API, you have to pay for it. So we actually started with a dynamic of, we're going to grab the audio straight from the user's device. So whether you wanna give us the audio stream or not, we're getting it. And then we're gonna take, the metadata from your system, the campaign Id call, id, customer id, geography or area code. We'll take whatever the metadata is that you have that's just easily readily accessible, and we'll populate that in into Balto. So, whether you, you, you like it or not, like we're able to serve your customers. And then we were able to get a certain amount of traction where they, they all say, Oh, you have like dozens of our customers, that, that you're working with. I think it's probably better that, we're playing friendly rather than kind of shutting you out. So it's a, it's an interesting dynamic. But our philosophy is we're going to make sure we go grab the data. And then when folks are excited and the CCaaS players and you guys players are excited to kind of really partner closely with us, we'll look for really great ways to enhance your customer experience and make our mutual customer as happy as possible. By passing more data back and forth. And if we do that, well, that's really good stickiness for you. Cause the other CCaas player may not have that depth of integration. If we look at bringing this to life inside of an organization and they haven't had any type of agent assist before, there's obviously a learning curve and adoption. And when I think of your solution, It's almost analogous to a pilot having a HUD, a heads up display. You're trying to get people to have attention on the task that they're performing, but give them that guidance and that support in the process, of delivering on that mission, How do you ensure that as you interject this new way of having a conversation that people get up to speed with it quickly and that people adopt it and adhere to it. Yep. The getting up to speed part is easy because we made the application easy. So, basically there's only a few different types of guidance that you'll receive and people kind of intuitively just take that and mix that into their calls. So we actually don't really have a problem with people getting up to speed with it, especially if you're a new agent and you're brand new, the contact center, you're like, what do I do? What do I do? Look at the screen and just see, see what it recommends and try it. So we find that's the kind of entry point is not difficult. The harder point is. Kind of the overall long term adoption of these systems and one of the questions is, well, let's say that you're doing this real time prompting and you're doing again and again and again and again at some point, don't people learn it? Like, didn't you really drill it in? And the answer is absolutely yes, they do. The other answer is, and if you stop reinforcing it over time, it will drift, especially as new at your workforce turns over. And then we have to cater to the changing needs of different agents at different parts of their career. So if you're a brand new agent, what you need is different than if you're an agent who's been there for a year or two years, if you're an agent who's been there for a year or two years, you might just want speed. You might be like, Hey, give me the links. Give me the knowledge articles. Finish my notes for me. I'm using this for speed. I took all of the items. Like I customized them myself and I'm just on a flow and I'm good to go. I'm not going to use this to like, do more closing or do more objections because I'm a top agent. But if you're somebody who's new, you might say, I'm actually like hanging on every word of it. So the needs adjust and we have to make sure that we're providing value at all different levels of tenure. And those, those needs are different. One of the things if you're a tenured person that you love is being recognized, is being called out, is being a leader. So we'll often, use the tenured folks performance as like the background for the analytics or the baseline for analytics of this is what great looks like. So you can actually tag your top people and you're able to then see two samples, this is what top folks do versus this is what everyone else does. So you got to be very thoughtful about how you change the value depending on where the agent's at in their life cycle. I like that adaptive perspective. Is there gamification with inside the solution? There is. How does that work? You can set up leaderboards and challenges and that's about it. Okay. So, we've done a little bit on the gamification side. We have a gamification module, but I think the contact center has been pretty loud and clear that that's not the big lever mover. So, I think leaderboards are helpful a, because they give people kind of feedback on where you stand. Just literally in comparison to your peers and B, you always want to climb up the ranks. So it is a little nudge, but I think the second you start going into gamification world of like, you have a flower and every time you handle objection, the flower grows more. You're on, you're a level three flower grower. I think that like, folks are like, man, aren't there like bigger fish we can fry with AI. So we, we, we have the foundation of gamification. But that's not like where our like energy is pointed right now. I think gamification can be problematic and, and it matters really what behaviors you're trying to tease out because ultimately, there's that individualistic behavior. I want to be the leader, in this, it's almost a gamer kind of mentality. And then there's the other side of the coin where maybe you're just trying to create. Rising tides lift all ships and we want to see that the overall performance of the center is getting better and and perhaps, you've got data that you can use to tease out and and reward or lift up the performance of particular players, but you don't want to have that be like a daily gaming type of presence. So I appreciate your response to that Totally, and I think, I think we're kind of all trying to figure out what our relationship is with like games and screens right now. I think people, there's a lot of attention on like, you know, dopamine addiction, and how things like, social media and screen time, are affecting kids and teens and attention span. And. I'd like to be on the right side of history there. So I think we're going to try to like really respect the user's brain as much as we can. That sounds like some good discernment. From the standpoint of investment in your product, what does R& D look like for Balto? Well, the big thing that we're focused on right now are two big things. One is we're continuing to invest in QA co pilot and that is our QA product that scores more calls than a human can score at about half of the cost that you would pay a human to score the call. And the big kicker is actually more accurately than a human grader. Where, across a sample of 20, 000 calls, we actually manually scored them against our QA co pilot and QA co pilot was able to score those calls with a 96. 7 percent accuracy. Our humans scored them at a 93 percent accuracy. The average kind of contact center looks for two graders to be within five points of each other. So 95, right. Where they would differ by about 5%. It helps eliminate like potential bias. Absolutely, totally. And in fact, that's one of the things I think is particularly important is that the standards are consistent from call to call agent to agent. And that everybody knows that the exact same standard was applied uniformly. So I think that actually like eliminates a bunch of concerns that agents might have around,, this greater versus that greater, or, you know, this is, how I do it. It's kind of a gray area. And it's like, well, everyone's getting treated to the same gray area in the exact same way. So it kind of takes some of those like almost annoying culture friction points and then pulls those out. We talked a little bit about that pulling of the thread and what comes with that thread is you're trying to solve problems and just natively having that process as part of how you innovate, I would imagine that your platform, just because of its. of the thread that it's pulling, that it has to encourage and maybe facilitate cross team collaboration that isn't occurring natively with inside the contact center. Our, one of our founders and our architect that was actually his favorite part about the vision is when, he and I have, have visited hundreds of contact centers together, like in person, sitting with agents and supervisors. And. One of the things that he noted most was, man, what the agents are thinking and doing and what they think, what they expect and they think is expected of them is just always different than what supervisors think they're doing and should be doing and what they expect of them. And there is just a, a weird gap in communication between, agents and supervisors about what really is expected and what the, the most effective ways are to, execute your calls with customers. So, a big part of the vision that he had, that he's brought to Balto, is how do we get agents and supervisors. Communicating better together and, we've given supervisors real time alerting as well. So we try to create a mirror experience where if you're going to give an agent real time alerting, let's give the supervisor the ability to have real time alerting. So if the agent, gets a complaint and a customer says, this is ridiculous. You guys have been so unfair. The agent might get guidance on how to handle that. And the supervisor gets an alert, says customer complaint, and they can click a button and live listen right there. So we want to try to like really bridge these two parties. And I think that actually just the way that, kind of work and culture is headed is I think that managers and the people that they're managing are becoming more like collegial. I think it's a little bit less kind of command and control and a little bit more of like, how do I empower you? And how do I help you? And we want to be a tool that kind of fits right there in that dynamic. Yeah. More coach leader versus manager. Totally. So what feedback are customers giving you right now? That leads you to, understand more importantly, what business outcomes you're driving inside of those organizations? I think, the feedback we're hearing, of course, on the sales side is improving conversion rates on the customer experience side is reducing handle time. And I think the feedback that folks are most interested in and excited about going forward. Is I want to learn from my data. I want to learn from my data and I don't want to have to dig. I don't want to have to come in with a hypothesis and say, well, here's what I think is happening. Let me dig through and explore. I want to like open up my workstation. And I actually saw a product at CCW that I thought was really cool. And CCW contact center week, the conference that was in Vegas last week. And what the product did is actually at the bottom of the screen, kind of like a, like a news breaking news kind of thing, like a little news roll. It, it had insights about things happen to contact center. It's like handle time up 7%, a customer NPS up 4%. And it was just kind of going like scrolling through like a live scroll. And I think that that's actually. Really touching on something important, which is people want to be presented with insights and be able to latch on to the things they like most. So we're really focused on making it easy and giving businesses the ability to learn from our data. We've been able to over the past seven years really crush the metrics. So we have, systems be able to drive our sales, be able to reduce your, your handle time, and we can do that consistently. So now kind of the next step for us is. How do we give you insights without you having to dig for it? I love that we, our company is very similar to Arroyo 360 where fundamentally we have three critical points that we're trying to impact with our solutions and our services. And one is drive better conversion. Kind of makes sense across the board. But conversion is not only in sales, there's, there's conversion in the customer support side as well. And across really the entire customer journey, what are the touch points and what are the opportunities through a call to action to have a conversion? And then the second thing is amplifying engagement or interactions. And that's across whatever channel that, that you're having that interaction on and whether it's elevated or not. And then thirdly is. Drive better data insights. So ultimately at the end of all of this, we should have better data to make better decisions or to drive automation, whatever that is. So I like the alignment for, for your company is multi language becoming a thing. Is that something that is part and parcel of your solution? Today it's English and Spanish. So we have, we have two. And we're thinking about our language roadmap because previously the way people would do it is they would add on one at a time and you'd say, congratulations, there's now French. Or there's now, Arabic. But I think that with the really new, powerful LLMs and the, these real time translation engines, we might actually just pursue an approach where we add all the languages at once. And that's something that's now becoming doable. Because you can prompt, these LLMs and say, respond in whichever language is , relevant to the conversation. So we're looking at those sort of approaches and we want to make sure we make those decisions wisely. But today we do have both English and Spanish and we're thinking about what the next thing is. Yeah. I think that context is, huge with that, right? Because when you have language, you not only have the academic view of language, but then there's also the cultural context you're working with. And so, things can, can move from being translated from English into another language that could result in a very, very bad experience or recommendation. So I would imagine that that is a little bit of a tricky chore to, to navigate. But it also for right now, I would imagine that the majority of your customer base is North American or at least English speaking. Is that true? Totally. North America, South America. What industry shifts outside of maybe Contact Center are influencing you now? Is there anything that has to do with things like VR or any other technology emergence that you're looking at and saying, Hey, this is something that might make it into our bag? Well, I love watching what's happening in other industries. And kind of being inspired from them and learning from them. But in this world right now, if you want to win, you have to stay focused. It is so easy to go way too broad, get spread too thin, build too broad of a solution, go into too many markets, and end up forgetting who your customer is and not building like the right solutions that you need at the right pace. So we are really focused contact center, contact center, contact center, sales and customer service teams. That being said, the technological releases that really have inspired me recently, that the biggest one is GPT four Oh, and. That's basically the, for those who aren't familiar, you know GPT 4. 0 is this back and forth you can talk to it just kind of like, Alexa or Siri, and you can ask it questions and you can actually have like a full conversation with it. And the big kicker is that it's multi modal, so you can, have a voice conversation. You can also pick up your phone and say, Hey, what's that? And take a picture or something that says, or show it a live video, it says, ah, that's a leaf, it's able to do that. That image recognition piece, I actually think is the more important aspect of it. I think we've seen a bunch of these kind of semi conversational, abilities to communicate with AI and we've seen that in, in, in voice and it's close, but it's not there yet, it's still interrupts. It's still babbles. It doesn't quite match your cadence or your speech or , just the kind of natural mirroring people do. It, it was good, but it wasn't, like the next coming of speech. But I do think. That what I saw in the vision recognition, they're really trying to bring computer vision to the masses. And, now you can write a prompt, , over an image and say, tell me all the things that are in this image, and if you're able to do that and you can pull from their APIs to do that, that's where we go to what I was talking about phase two, intelligent screens, where every screen is going to be basically intelligent, where you can say, what is on the screen? You know, list out, all the different applications you see, list out the people, list out every five seconds what their emotion looks like, chart how the emotion is changing over time, list out what they're wearing, like everything. And so basically our screens are going to be like serving up this sort of intelligence. And in the context center, you're going to see. That go to agents in call and after call work. So are they putting in the right data? Did the data match what the customer said? Well, instead of just looking at that, maybe you can actually enter the data automatically based on what the customer said. So there's a whole bunch of opportunities to do that type of work. And I, I think that GPT 4. 0 is like the most inspired I've been that, Balto's vision of the intelligent agent desktop, basically where agents talking to customers, they don't need to use their hands at all. Like everything's getting filled out for them. Their information is getting searched and prompted for them automatically. They're just having a conversation with the customer. It gave me, it really inspired me more than anything I've seen recently that like, we are going to be able to build toward that in the near future. So is the intelligent desktop your longterm vision, that anything that falls within the context of that phrasing is what you're about? Yep. If, you are. An agent having a conversation with the customer your, everything on your screen should be intelligent. It should be intelligent in the data it takes and looks at in order to create analytics for the organization and give the organization insights on what's happening and then it should be intelligent and everything it gives you. So it's filling out your notes, it's doing your, your in call work, it's pulling up knowledge articles for you and you're not having to go and search. It just is recognizing this is the right knowledge article for what they're talking about. That I think is like. Absolutely the future. And you can think about how much more effective, and efficient agents be able to be if they're not doing anything besides having a conversation with the customer. If a client has a knowledge base already, that's been built out in whatever form it could be inside of a SharePoint repository, something along those lines, is that something that you're able to dip into right away where you can say, what are your sources of truth inside of your organization? Yeah, one of the things that we'll do all the time is when a certain topic comes up, be like, Hey, here's the knowledge article, like, click this button. There it is. That's where you go. And we make it super easy to get anywhere you need with the click. We're going to start almost generating the knowledge articles ourselves. And even pulling out like little summaries or excerpts of the most important parts of those articles. And that will be coming in the next few months. That's outstanding. I can see that your organization is essentially very fluid in terms of how it's adapting to change. Would you say that that's a cultural dynamic inside of your business that you've intentionally created for there to be this viewpoint on constant and continuous improvement? Absolutely. Part intentional, part unintentional, part of the, the fluid aspect comes from that, we started out with three 24 year old founders. In a 75 square foot room where we just bootstrapped the thing ourselves. So there really is this go, go, go, hustle, hustle, hustle culture. Which is really wonderful. It also can make things messy, right? Like, when you're focused on going fast and you're focused on just getting in there and doing things yourself, it can make delegation more difficult. It can mean that folks go too far in a project before they pick their heads up and say, hold on, are we pursuing the right strategy, the right objective here? So there's like absolutely blessings and curses. Some of it has absolutely been intentional, but we also have to moderate for it. On my whiteboard, the full first year, year and a half, we were starting the company. I had a big thing written up there. It stayed the entire year, year and a half. And it said, what is the most important thing I can be doing? And that I think is something that you have to keep stepping back and keep asking yourself in order to make sure you're not going fast in the wrong direction. So blessings and curses overall. I'm super glad we have it. And we've like figured out how to channel in the right way, but it certainly is a beast that needs to be channeled. One thing you can possibly add is a great Marine phrase that says slow is smooth and smooth is fast. And it really just speaks to the fact that. When you fall over your own shoes, you create a level of calamity that is padamout to technical debt, right? Like you have to refactor yourself all the time because you didn't take the proper order of operation. So it isn't the fact that you need to go slow. It's just that slow is smooth and smooth is fast and you moderate that pace to create the quality that you need. That is like such a good analogy. And actually I was just the other day, I was watching, a video of one of these elite units training and they were kind of busting down doors. And I was thinking to myself, like, it's interesting how they're not sprinting at full speed., they're not sitting there sprinting at full speed. They're kind of all just like methodically going at like almost a trot. And I think that's actually a really good example of what you're saying. Which is like having that control is the most important thing. And as long as you don't keep falling into ditches. You're able to ultimately it's the tortoise in the hair For sure. Brilliant conversation, Mark. Thanks for being on the show. I appreciate you! Frank. You too. This was awesome.