In this episode of GTM Innovators, we sit down with Scott Newman, Chief Strategy Officer at OneSource, to explore how AI is fundamentally transforming the BPO industry. Scott shares how OneSource is leveraging its platform, CalibrAIte, to move beyond traditional contact center operations—empowering frontline agents, automating coaching and QA processes, and generating the structured data companies need to truly adopt AI. From real-time feedback loops to global scalability, Scott offers actionable insights and hard-won lessons on what it really takes to innovate in one of the most human-centric corners of the business world.
OneSource Report: Analyst Insight: The Hidden AI Adoption Barrier – Why 80% of Companies Aren’t Ready
Subscript to GMT Innovators Series on the following platforms:
- SoundCloud: https://soundcloud.com/research-859405782
- YouTube: https://www.youtube.com/playlist?list=PLsoV6fwX4cpGR2Hg98rc1e1k2-cOqCbhb
- Spotify: https://open.spotify.com/show/1gvDzcl0jxpPIfu9WYu6U4
- iHeartReadio: https://iheart.com/podcast/258127960/
- Pandora: https://www.pandora.com/podcast/gtm-innovators/PC:1001097038
- iTunes: https://podcasts.apple.com/us/podcast/gtm-innovators/id1790738579
- Amazon: https://music.amazon.com/podcasts/1b000615-31cc-49dd-a5d8-f80d5098bf2d/gtm-innovators
Transcript:
Kyle James 00:00
welcome everybody to another episode of GTM Innovators by 3Sixty Insights. I’m your host, Kyle James, and today we’re having a conversation with Scott Newman, the Chief Strategy Officer. OneSource, Scott, how you doing today?
Scott Newman 00:21
Hi all. Thanks for having me. I’m doing great. I actually got my start on the other side of the contact center industry. I was the user of outsourcing services, so I at Edison worldwide. I was the director of operations, and we were a direct response company, and we had all of our sales service, retention and collections outsourced to various providers around the world. Had contact center seats in Pakistan, India, Philippines, Costa Rica, Dominican Republic, just to name a few. And after doing that for a few years, I became frustrated with the overall process of what I was experiencing. I almost felt like we were opening and closing partners more than anybody should ever do in a lifetime, and I was almost more focused on firefighting and just trying to keep centers up and running, rather than actually working on the customer experience and doing something meaningful. So in in mid 2009 I said, Hey, I, you know, and I was, I was 27 years old at the time, right? So, a little bit of naivety, and, yeah, right, some entrepreneurial spirit. And I say, I’m going to quit my job and do this, because I I see a void in the market, and I’m going to start the contact center outsourcer that I wish that I find found to work with. So I at the end of 2009 I actually moved to the country of Belize. And there’s more to the story about how I arrived there. But I moved to the country of Belize at 27 years old, and started transparent, BPO with my partners at the time. And we started very small, bootstrapped the business. I hired the first 30 people and trained them myself, and what we were trying to do really started to catch on. And over the next 15 years, we started building Transparent BPO and a bigger and bigger brand, and we became the largest private employer in the country of Belize, something we were really, really proud of. We had about, you know, at the time, about 1500 employees in the contact center business. We’re the only, the first real outsourcing company to ever open up in the country. So we really helped sort of pioneer the market and define what that is. And then we moved into the Philippines in 2019 and then into Jamaica in 2022 and it was just a really awesome journey. And, you know, as we continue to grow and wanted to expand into a bigger footprint and offer more services and technology, you know, that’s when Transparent BPO became part of the one source family, and I took the job as Chief Strategy Officer there to help sort of look at technology, emerging technology solutions, and where we could really differentiate in the market. I’ve always had, I’m not a technical person, but I’ve always had a real passion for just creating things. And with my former partners, we built the first one of the first contact center systems in the market, and I sold my interest in that in probably 2015 and so I’ve always had sort of a knack for technology products and wanting to build stuff, and I think that’s a big part of what I bring to the table today, and where I spend a lot of my time, that’s
Kyle James 04:37
really cool. I want to go back and, like, double click, dive a little bit deeper into something like you mentioned, like the frustration and kind of, you know, what you’ve built now with kind of call centers, like, what, what was that frustration like? Have you kind of identified? It sounds like you have kind of identified as, like, best practices, blueprints of, kind of how to stand these things and stand these things up and really roll them up out at scale. But, like. What was the main frustration, was it just like people not having great English or kind of like, yeah, dive in, in a little bit. Dive into a little bit more. Do that for the audience.
Scott Newman 05:09
Yeah, I will, you know, it’s funny, because I had it with a client. I had a conversation the other day, you know, about English and accent, and because we, you know, we’re in in 18 different countries today, offering many different options and language support. And, you know, they were concerned around one of our markets and how Americans would view accent and and there were sort of some preconceived notions around that. And as I explained to them, and you know, I’ve said for years, accent is not the source of frustration when it comes to customer experience, the source of frustration is because your problem is not getting resolved. You know, if, if I talk to an agent and you are professional and courteous, and you resolve my problem, and you come across as an expert, accent doesn’t matter, but the minute that the customer experience is poor, you know, it is always the first thing that everybody latches on to, right? It’s about accent. So it’s more about resolution than it is location. So, you know, English was never really the big thing that I was always trying to solve for, I think, with a lot of the contact center partners. Back in the day, it was a very commoditized service. You know, it was like we had the saying in the industry is like putting cheeks in seats. You know, how many bodies do I need to put in the seats? Because I need to Bill my client this number of hours. And you know, when agent a doesn’t show up, who’s the next person I can just throw into the seat so I can get a billable hour. And those kinds of methodologies and mentalities don’t produce really good customer experiences for people. So really, what I was and the other part that I thought was difficult is with a lot of the companies we work with, I never really interacted with senior management at any of these companies. I was just too small, right? You know, 50 seats here or there for a lot of these companies. If you’re not spending 200 million plus in revenues on an annual basis, you don’t speak to anybody above a manager level, and so when I wanted to focus on strategic initiatives that for us were really important, I didn’t have the audience in front of me to actually be able to have those kinds of conversations. So for me, it was about creating a model that was focused on quality delivery, making sure that we in over invested in the training and leadership development, that we were building good managers and a good culture of, you know, bench strength of leadership over time, and making sure that we were accessible in our business, that, you know, a lot of us, we love being involved in the strategic side of the business and But unfortunately, a lot of our jobs tend to become very tactical in nature. And so I wanted to build a model where the leaders in the company were always had a big chunk of their time spent on strategic initiatives, both internal and external, for our clients, and having that level of engagement. So those are the really the big parts that I was focused on, and sort of how to solve for that. I became really frustrated in the market with well,
Kyle James 08:05
and it’s funny, because hearing you talk about that, I mean, I think leads right into what you’ve kind of built now, and kind of, to give the audience a little bit more flavor, I recently wrote a research report, kind of talking about your calibrate project or product that you’ve rolled out, kind of for these call centers to help educate, train, provide feedback, get call scores and things like that. And it sounds like, you know, getting that strength of bench, like, that’s that it solves so many of these problems that you have and kind of wrapped up into one solution. So, you know, just so everybody’s like, kind of updated. Like, give us a quick overview of, kind of what is calibrate and kind of how, you know, how customers use it, and, yeah, just like, really what it is, I guess,
Scott Newman 08:47
yeah. So, you know, ever since AI has sort of come hot and heavy onto the onto the scene, everybody’s looking at ways of, how do you invest into it, and how do you leverage it? And in our space, it’s really unique, because we are in the human capital business, and a lot of customer interactions still need to be handled with, you know, humans. But you know, where we’ve spent a lot of time is we say, hey, the products that we’re going to build ourselves are going to be things like that we can integrate with best in class, AI technology solutions, because we’ll never have the firepower of a lot of these big organizations that are exclusively working on, you know, large language models or, you know, other types of things like that. And so we’ve, we’ve said, you know, how do we leverage this to to provide a better experience and better coaching for our employees? And so that’s where calibrate came in. And the idea behind calibrate is fairly simple on the surface, very complex than actual execution. But the idea is, if it’s a call, we take a voice call, we transcribe it with 9095 to 98% accuracy into text, and then after we create sort of a custom scorecard or. Evaluation card, where for that particular client or line of business or call, it’s completely custom to what their business needs to evaluate, we’re able to take that transcript and then with creative prompting, we’re actually able to drive scoring and meaningful feedback for the particular call, and end up automating the entire QA process within our within our space, we have been able to do this with calls, with text, with chat transcripts, with emails, so we can process all different kinds of interactions. And then we’ve also been able to, you know, find ways to bridge the gap from the actual evaluation and getting objective measurements that we can use for benchmarking and different things. But also, how do we help agents, you know, frontline employees, become better at their job with real time coaching, right? How do we help them see what they did and how to improve and give them opportunities within calibrating, and those have been some of the features that we’ve been heavily focused on and building today. I
Kyle James 11:02
love it because I hear so many companies that I talk to talk about pulling these call transcripts, because it seems like such an obvious low hanging fruit, because there is massive amounts of data created by every single company that has conversations on the phone, and everybody does all day long, and if they’re not pulling call transcripts and bringing it into their data warehouses now and doing things with it, like it that, that’s That’s step one. That’s obvious, but what you’ve really done is, like taking that and, like, actual brought out actionable steps and tools, and you don’t even talk about the coaching aspect, right? Like, you could also coach these people with it, which I think is just super cool. They get off a call, it didn’t go well. They know it didn’t go well. They see the score that didn’t go well. Well, they can go do a retry, essentially, right? With AI and, like, have that conversation with it, which is, like, super innovative, but, you know, the text there now to do that. It’s not science fiction anymore.
Scott Newman 11:56
Yeah, that’s exactly right. I mean, you know, we call it a do over function where, when an agent, and we’re leveraging that to become simulation based as well, where, you know, when an agent doesn’t have a good call, they can actually interact with the AI, and the AI acts like the customer, driving the same outcome of the call, but the agent is now giving different responses so that they can see what the outcome would have looked like if they had handled the call differently. So if, if the customer became irate because they didn’t do a good job de escalating a certain situation, they can replay that call. What the with the AI acting like the customer? The other sort of ongoing thing that we see in the space when I mentioned simulations, and then this is, you know, our biggest project we’re working on right now is training and getting agents ready to actually get on the phones. It is one of the most stressful time periods for agents in our business. It’s you go from classroom training and you do your best to kind of get them ready to get on the phones. But the reality is that there’s so much anxiety, we see a huge drop, like a huge spike in attrition from like day one through the first week, because it’s so overwhelming, it takes agents time to get comfortable and to get over that anxiety. So we’ve been working on simulations using calibrate and AI, so that they can have more of a real life experience of actually role playing calls using AI before they actually have to take a live call. And so we can give them dozens and dozens of real life scenarios that took place, but they’re being played with AI so that the impacts of the call aren’t there, which takes the level of the of the of the pressure off of their plate.
Kyle James 13:43
That’s That’s fascinating. It’s like taking over the do over function and just generalizing it. So now I’m curious, like, let’s say, what is onboarding typically like, Is it, is it kind of a weak ramp up? And do they spend a day or two kind of in this simulation mode, like basically running the call, like having real conversations, but, but they’re not real conversations and, and because they know they’re not, and it doesn’t, you know, it’s not fourth quarter balls on, you know, the red zone, and they have to get it right, like they can, they can, you know, be more comfortable with that right. That’s exactly
Scott Newman 14:15
right. I mean, so classroom training for clients varies drastically. If some clients that train in four hours because it’s really simple. And I have other clients that it takes 12 weeks of classroom training before they ever take a live call. And then most contact centers do something called nesting, which is a typically, right? And this varies too, is the two week period where it’s a highly structured two weeks. So they might go spend 30 minutes on day one taking a phone call or two, and then they come back in the classroom, and they do, they revisit those couple of calls, what did you hear? What did you expect? And then they go back. And so we sort of ease them in over the first two weeks. The goal of nesting in most contact centers is you want to get to eight. Into about 80% proficiency. You know, it’s Elon Musk did a great explanation of battery charging. I know this is a completely random tangent, but it’s something the way that I thought about is, he said, you know, the speed to charge a Tesla battery from zero to 80% is like 20 minutes. It’s incredibly fast. And he equated it to a parking lot, if you were to drive into a parking lot at a Walmart and there’s no cars in the parking lot, finding a space takes an instant. Yeah, but these batteries, the ions, are coming in and parking in these parking spots, and so that last 20% to get the battery from 80 to 100% takes a lot of time. And that’s the same thing in our environment. The first couple weeks they get to 80% proficiency if we’re doing our job well, that last 20% takes 90 days, on average, to really become an expert at what they’re doing. And so we, what we’re looking at is ways to say, hey, like, how do we ease them into this in the classroom setting, before they take a real time, like phone call, and to give them access and real life scenarios that they’re going to experience so that maybe, just maybe, we can go from 80% to 90% in that nesting period and provide a quicker path proficiency for the employees. Employees in general, I think, want to be successful in every industry. They want to win. We as employers just need to do a better job giving our teammates a better chance of success. And so this is one of those things that will hopefully help them get over the the learning curve well.
Kyle James 16:30
And what I think is another really interesting example of this is like, these are frontline employees, right? They’re not senior executives, they’re not very they’re a lot of times not super technical savvy, and they’re getting exposed to AI in a way where, like, it’s collaborative, it’s helping them, and it’s not really scary, right? Which is what has been like, the response and feedback of that, because that’s typically a cohort that doesn’t get to play with they’re not to chat GPT, asking questions and doing stuff all the time, like some of us are, but like, what has been that experience that you’ve seen and heard about them getting introduced this like science fiction technology, right? Yeah.
Scott Newman 17:13
I mean, I think my answer a year ago would be different than it is today. I mean, a year ago, if a frontline employee heard about some sort of AI tool that we were working on or deploy a client was deploying immediate negative reaction, shut down. My job’s going to be eliminated. That sort of that was the early reactions to things. But you know, the project I’m talking about, calibrate, is just one example of dozens that are being already live and deployed in the contact center environment, and most of the development and tools are there to design to help the agent do a better job at their provide a better customer experience, right for whoever they’re handling a call for or or interaction. And so whether that be like what a product that is known in our industry as Agent assist, where it’s actually connected into the phone line and it’s listening in real time. And if you know Kyle, if you called me and I was, I was a travel agent, and you said, hey, you know, I really like to travel to Hawaii. I’ve never been before. What islands? One of the best ideas places I should go, the AI is listening and immediately is popping up talking points or suggestions or recommendations, so that the the agent can read them and, you know, kind of give some guidance that is founded in, you know, some sort of fact. And it’s not a realistic expectation that an agent that’s sitting in, you know, we’re anywhere in the United States or in a foreign country is going to know all the nuances of the islands of Hawaii or, like, the cities of of New Mexico, like, that’s not a realistic training expectation. And so these assist tools that they’ve been exposed to are received well, because it’s allowing them to seem more like an expert, and when they can act and like, have the feeling of an expert to customers, the dialog is so much easier, and you don’t end up with these friction, you know, points on a call. But I really think that they’ve seen the value in it, so today, more than ever, you know, this year, as opposed to maybe last year, they hear about it. It’s not a scary thing anymore, right? It’s more of, hey, this is something that’s going to be a valuable tool for us well,
Kyle James 19:22
and I’m hearing you give that example, and I’m like, Maybe I’m just a nerd, but I’m also sitting there thinking, like, I’m curious about that, right? If someone’s asking me about Hawaii, and there’s something popping up, and here’s telling me about the little islands with a little bit of description off, and I can read that to the customer. I’m learning about it too. And then like, oh, that island sounds interesting. And then it populates resort options there that you Oh, this one’s got five stars. This is the kind of rooms it has. It’s all inclusive. It’s this or that. Like you’re getting to share that and feeling like you know you’re doing your job in a rewarding way, but you’re also not struggling, trying to do Google searches and all this. It’s just like coming to you, and then you put. But click a button, I imagine, to kind of book something for them and help them right?
Scott Newman 20:03
Exactly. I mean, it takes so much anxiety out of the frontline team members hands, because I think where people struggle the most is in the unknown, right? Like, that’s where the anxiety and fear comes from. And when you have a tool, an AI tool, that’s sitting there, you know, in your pocket that is able to help you and fill in the gaps when you don’t know something. It really is such a valuable piece of information, and it really is transforming the way that we provide support today. So,
Kyle James 20:32
so now I’ve got to ask, what other tool like you mentioned, there’s all these other tools. What are some other ones that y’all are playing with or rolling out, like some of the crazy you know? Let’s talk about the future. This is exciting. I
Scott Newman 20:43
mean, so some of the things, there’s a lot of the chat bots that are online, those are all AI, 100% AI driven. That’s been what I would call more of the low hanging fruit in our space. So there’s a lot of that. There’s a lot of clients that are actually using a voice AI bot to handle calls. Best practice in the industry, really, is that they say, Hey, I’m, you know, let’s say the AI bot is, is Kyle, right? Hey, I’m Kyle. I’m an AI bot, but I know as much as a human, I’d be happy to help you today. But if you’d like to speak to a human, please say human. And so a lot of businesses today to kind of dip their toe in the water and not alienate they’re actually indicating that it is an AI bot, and you can opt out of that. I actually was talking with a client the other day. They’re seeing and all their voice calls are greeted with an AI bot, and about 45% of their customers are electing to go human. And so there is a significant adoption of that. Some of the cool things are noise like aI driven noise cancelation. So you know, typically, you know when you’re talking to somebody in a call center, because you hear everybody around them. That’s no longer a reality, because there’s an AI driven noise cancelation that with zero latency on the call is is pulling out all the background noises, dogs barking, babies crying, other people yelling around them. So that’s happening and and I think one of the, actually one tool that we have today is called accent translation. So using AI, the model has been trained on different dialects in different market, and it actually is eliminating accents from calls. So I could be speaking to somebody in India, where I may have an Indian accent, and the AI is removing that so they sound like, you know, Susie, that’s sitting in Texas. It’s a human. It sounds like a human. It’s just removing the accent. I think some of the neat things that are coming right now, which are very close, is real time language translation, where, you know, someone in another country potentially could be speaking Spanish, and it’s real time being translated to English, and you know, vice versa, going back to the agent. So you know, everybody can speak in their in their native language and have clear communications that you don’t have any issues with contact when, when you’re each party is trying to translate well, and
Kyle James 23:08
it does that in real time. It’s pretty close.
Scott Newman 23:11
It’s pretty close to real time. It does have a little bit of latency built in, but if you’re not on video, you never really see the see that thing. But that technology is advancing at a very rapid pace. I’d expect to see that rolled out pretty aggressively going into 2026, or late, late this year.
Kyle James 23:31
I want to dive in, because I recently had a conversation on the show with Sarah Ross, who is at smart cat, who is kind of doing a lot of that AI translation, and go listen to that hashtag. Go listen to that episode, if we’re to dive into that. But I think the interesting thing we got to at the end of the call was, you know, what you’ve seen in the last 100 years is, how many, and this is, I’m nerding out on a side to edge here, but I’m curious for your perspective of this is we’ve seen like, how many languages have we lost? Right? Because, you know, English is the business language, and everybody needs to learn English. But if we have this, like Babel Fish translation area, like layer in the middle, like, there’s not the need to lose all these dialects and local language anymore. I’m curious if you thought about that, like, that seems like a very good preservation thing, where we’re so used to technology like destroying things, but here’s an opportunity, because you don’t have to lose this to match everybody else. It kind of keeps diversity, you know, in ways that maybe we haven’t thought about in the past. I
Scott Newman 24:31
think the part that’s really important to us, it’s, it’s about partnership, and we are a multi cultural organization, being in, you know, so many different countries. And I think for us internally, in terms of like culture, it’s is to respect the cultures of each one of the markets that we work in, and to figure out the best way to communicate without requiring people to adapt to, you know, English or other languages. And so I think there’s such a huge. Component of that, that it just comes down to, like, really great collaboration. And you know, there are individuals, even within our organization, that try to speak English, but there’s going to be communication barriers and gaps because it’s not their primary language. And so when you remove that variable from the communication, I just think it’s gonna, it’s gonna drive such more meaningful interactions, both within companies and within outside. You know, we, we have a huge augmented talent division where, you know, us, companies are hiring remote workforce in different countries because it’s impossible to hire in certain jobs in the United States today at a cost effective number. And, you know, being able to have this real time language translation so that you can access different parts of the world where the talent you could have a doctor in in the Philippines, that is just one of the most educated, smart individuals that could do wonders for your organization. They there’s just one thing, they don’t speak English, yeah, right. And we’re there’s a whole labor pool that we’re not even discussing today that could, you know, that could be valuable resources. So I see the potential for these kinds of products to really interconnect a lot of cultures and countries that we haven’t been able to figure out before.
Kyle James 26:19
Yeah, that’s yeah. And I love, like, the common theme through all these, whether it’s reducing background noise or giving readable recommendations or letting people train, is like, how do we make the customer experience better? Right? All of these do that and also build confidence in the people that are delivering it from both sides. And, you know, if I could talk to an AI bot right away, versus waiting on a, you know, like, I’ll admit, there’s a lot of times I’ll just go use a chat bot because I know I can start getting a response now, versus be on hold for 510, minutes. But if that could get the ball going and then transfer the correct data on a screen to the person, like, it’s it’s a win, win all around, for
Scott Newman 26:58
sure, we’re also the other factor that, you know, I think it’s sort of getting lost in all this, because there’s so much excitement around technology in the world today, is, is the generational shift that’s happening too, right? I mean, so you’re not only having access to all of these different technologies, and what, I think, the last time we spoke, we’ve never seen innovation and speed of development like ever in the history of our world, like we are seeing with AI today. It’s never. There’s nothing even comparable, in my opinion, to what we’ve seen in the past, in the in the last couple of years. But the other part is, like the younger generation, they will do everything possible not to make a phone call, right? They would send a text message, they’ll go to the website, they’ll send an email, but they part of the generation is that they want real time interaction, right? And so how do you realistically provide that level of support? So it there’s also generational changes that are happening right now in our space for customer experience that you can’t just solely think about, hey, how do we leverage AI for cost savings or for economics, but also like, how are we meeting the new generation of customers, where they want to be met, and giving them the type of experience? And I think that’s where AI really lends in. They grew up in that in that era of technology, right? So, like for me, in my generation, we, we learned about Facebook over time, right? This generation never knows about Facebook because it’s not what they use, right, right? And so I think it’s just as as the demands of the types of interactions and the engagement that they want, AI is becoming more and more of a realistic support solution, because it’s not scary to them. It’s actually a preferred method of communication, right?
Kyle James 28:47
That’s fascinating. Here’s a stat for you that I just heard this past week. I mean, we’re all familiar with Moore’s law now, right? That, like the transistors double every 18 months. Well, apparently, AI now is getting twice as smart every less than six months. Wow, I didn’t know that one. That’s crazy. So every six months, it doubles what it could do. So in a year, it’s four times better. It’s It’s wild. How fast it stuff’s getting better. I don’t know if that continues indefinitely, but right now, that’s the kind of transition we’re in. Is it possible for people to keep up with for
Scott Newman 29:18
sure, and not only the how much smarter is getting, but also, you know, you know, deep seek is a good example, not something we use, but to be able to now deploy these sorts of things at such a fraction of the, you know, the GPU processing needs to do that. And so I, you know, those are usually thinking of like a comparison. When the first big screen TV came out, it was so cost prohibitive that no one everybody wanted it, but no one could afford it. And it took, what, 1015, years for the price of big screen TVs to come down for people, for the average consumer, to be able to afford one. But already the AI models that are coming out where they’re using a quarter. Or less of the processing power that happened in a year, a year and a half again, I just think that we’re seeing all types of innovations around this that are going to be really substantial for us in the future.
Kyle James 30:11
Yeah, yeah. Oh, let’s take a step back a little bit, though, because one of the things that you and I really talked about in our first conversation is why, like, 80% of companies aren’t ready to do this stuff because they don’t have the foundation in place. Um, let’s dive into that a little bit, right? Like, I think we’ve talked about call transcripts. What are those transition, transactional or it’s, it’s, it’s table stakes now, right? To build your bespoke data set, but, but you know, what other best practices and things do you when you’re working with businesses, talk about when, like, hey, how do you get ready for this? And what are the things that you need to have in place to really take advantage of all this stuff that’s coming and all this compute power that’s available now? Yeah,
Scott Newman 30:50
you know, really for me and what I’ve seen, and I think businesses are starting to catch on to this, but I see a lot of businesses that have been I don’t want to sound negative when I say this, but sort of the tech stack has been cobbled together over time, right? You know? And maybe, like thinking about Southwest, how long ago that was at this point when they their systems crash for multiple days, and it was just because they weren’t keeping up to date, a lot of tech debt, and they weren’t keeping up to date on 95 wasn’t it? Yeah, exactly. But I think that’s it’s not an uncommon thing for a lot of businesses these days that they’re layering things on top of other things. I had one client that had a blue mainframe, and you know what they ended up doing was just building a new web interface to layer into that. But there really wasn’t a lot of consideration around like, what else, what kind of improvements do we need to make in the underlying database to support the types of innovation that is going to be coming through? And so a lot of companies don’t want to rip the band aid off, and so they have a lot of legacy things that have taken place. And so, you know, an average client will come to us sometimes with four or five different technology platforms. You know, we use this for our chat interactions. We use this for our phone calls. We this is our CRM, this is our order entry system. This is our knowledge base. And so everything is really disconnected, and they’ve never really bridged these things together, or they’re not capable of doing it. I think the reason that most companies are not able to leverage AI quite yet is because AI needs huge data sets to be fine tuned and trained to really get the large the types of results that you would expect. And so, you know, trying to integrate with a dozen different systems that don’t all talk the same language and have the same kinds of data that normalize data is a huge challenge that goes even in the knowledge basis. The stuff you train your employees on is such a critical part the AI models learning of what the right response is. It’s not just the data that needs to come out like factually what the right answer is. It’s also the context of how you want things to be answered in what sort of tone, professionalism and the other sorts of business policies and things that are there. And I think companies are racing right now to try to get this stuff together so that it can be integrated with AI, but I think the average company is not ready to do that quite yet. And so there’s an appetite for AI and like, how can we dabble in it? And I see a lot of clients that are, you know, doing proof of concept things, but that’s great, except that they’re not going to be able to roll it out because they haven’t done the underlying infrastructure requirements that are necessary to do that. So I actually have seen a lot of clients now, more over the last I think 12 to 18 months are going through major technology transformation by dumping their legacy systems and trying to migrate to new, cutting edge platforms for customer support so that they can check that box off. But that’s why I think a majority of the companies are not ready to really, truly leverage AI yet, and the maybe the challenging part for clients is that every AI vendor that they talk to, yes, we can integrate Yes, we can do that yes, because they’re trying to make software sales. Yeah, but it’s not enough to make it really the outcomes that happen because of that are not great. So I think that there’s a lot of painful lessons that are happening right now at the same time with companies dabbling in this that aren’t seeing the results that they had hoped for.
Kyle James 34:28
I’m curious, kind of going a little bit deeper than that, right? Like there’s the old way and then kind of there’s a new way. What is the difference? I’ll say that. Is it like 4x is it 10x is this one of those 10x improvements a day? If you clean up your infrastructure, move to a more modern stack where you could take advantage of this, your team will be 10x better. Is that right? Or is what do you think the percent is? I
Scott Newman 34:54
would hate to even put a number out there, because I don’t even know exactly how. Get how to get to that particular piece of information. I can tell you, it’s multiple x and I think that. So here’s an example, right? You know, Legacy way of doing things like, this is our chat, this is our email. This is our and everything’s a different piece of software that somehow integrate, you know, into like a centralized CRM. But what companies are going to now is sort of a single system of record that maps the customer journey, right? So I can go into a customer account and see what ad they responded to online, what website of ours and landing page they visited when they called in, what they ordered this month as opposed to last month, and having all that data in that complete customer journey in one central location allows you to do a number of things. So number one, if we think about cost savings, having all the data into one particular place will allow AI and other products to kind of layer on top of the human interaction to provide more efficient interaction support. There’s less navigating between windows for the agents. You get better call handling time. So I think that there’s immediate cost savings. You know, could that be 10% or 80% impossible to quantify in general, but for each each business case, it’s going to be different. But then, obviously it depends on client business, but the revenue impacts that this could drive. You know, for for decades, a lot of companies have viewed customer support as like a cost center. And, you know, that’s been a shift over time. But you know, really, customer support could be a revenue generator if you have AI and and the customer journey there, where you can make other product recommendations, or, Hey, I saw you didn’t reorder this last month. You know, can I help you with this? And you’re doing more outreach, you know, you see the customer’s preferences and likes, and you’re able to present ads or other products that will drive relevant call volume. So I, I think that there’s this whole other aspect of real, meaningful revenue drivers that can happen. So could it be 10x easily. I mean, you know, between the cost savings and the revenue increase, this is one of those ways that I think customers are going to be able to really benefit from on both ends of the spectrum.
Kyle James 37:12
Yeah, yeah. No, that makes a ton of sense. And because you’ve now got this giant database repository that you can query and ask questions about your customers and their experiences. I’m curious, how much are you seeing that data cross departmentally shared right, like whether it goes to a product team or a sales team or a marketing team to where they can query and say, Hey, what are the what are the 10 biggest issues we had last week, you know, so that we can overcome those objections in the sales process or write content around it to address it. Okay, you seeing that sort of collaboration start coming out of all of this?
Scott Newman 37:51
Yeah, I think with the with our calibrate implementation, we’re seeing more of that than ever before, right? Because it’s giving, you know, and that there’s only so much insight we get into some of those internal department conversations. But I think with calibrate, we’re seeing a lot more engagement on the marketing product side. Because, you know, what has traditionally become sort of secondhand complaints, like, you know, we’re we always say it’s like a multiplier in the context in our industry. Someone says, hey, you know, I’m getting a lot of calls about this today. Oh yeah, I’ve been getting that too, and I’ve been getting that too, and it compounds. And then when you go in and do like a root cause analysis, it was like one call that happened one time, but it was really impactful to the agent. So it seemed like a bigger thing. We’re now able to take more subjective feedback and turn it way more objective. And so I think a lot of that kind of stuff is happening in real time, going to the marketing and product departments, being able to get really, you know, meaningful pieces of information. The other part that we’ve been spending a ton of time trying to work on is, and I’ll just go briefly into it, but is soft skills in the contact center space. So like golf, soft skills are like tone and professionalism, empathy. Did they explain the agent? Explain next steps? Did they handle? Did they build rapport? We have probably 20 different soft skills that we’ve we’ve outlined, and we’ve been fine tuning our llms to be able to score these soft skills the same way that humans would. And so just maybe at a high level, what we’ve done is we’ve taken a sampling of calls, and we’ve had 150 or more humans for these calls, for these soft skills type things, and then you do statistical analysis to remove outliers, and ended up with your, you know, your standard deviation and your confidence interval. And then we take that information, and we’ve been fine tuning our llms to be able to bring their LLM scoring into that same sort of confidence interval range that that the humans would have, so that our. Calibri product will be able to score these soft skills within the deviation of what a human would. And so that’s been taking a ton of time and energy to really fine tune to be able to get that because what our hypothesis is with this particular exercise is, we think, you know, there are a lot of contact centers that measure a few, you know, soft skills like empathy or Paul control, or those professionalism and tone, but they haven’t been able to benchmark against it. What our objective is, if we have 20 soft skills, can we identify the two or three for this particular client that drive the best possible outcomes for the customer, and then all you go all in on training. You know, if it’s if it’s tone, we all we do is coach and train and hold them accountable to tone, because we know that that’s the number one driver for the best positive outcomes. The other thing that we’ve been spending a ton of time on and is challenging is, in our space, there’s something we call like after call surveys, which are measure customer satisfaction or net promoter score. You know, would you recommend zero to 10? And those are like the two biggest factors in the contact center space. But it’s always done by the customer after the call. We’re using traditional after call surveys to try to train the LLM so that the LLM will actually be able to provide a predictive NPS CSAT score based on what it hears in the call. And so our objective for doing this is, we think it’ll be a better customer experience to for Kyle to never have to get a call afterwards and say, Hey, would you tell us how well we did? Or, Hey, will you hang on the phone to tell us, so that we can become more predictive in that that kind of evaluation, so that those are a couple of the things where we’re spending a lot more time on right now to try to drive meaningful outcomes on
Kyle James 41:52
the call. That makes sense, if you could save the customer the annoyance of having asked them that hard question, because you could predict what they’re going to answer within some certainty, like, that’s a win for everybody. I’d be remiss if I didn’t kind of ask this kind of as we kind of start to wrap things up with all this innovation change in AI come and talk to me about how you’re thinking about the ethics aspects of it and the security aspects of it, right? Because, you know, as exciting as we are about all this, there’s also a whole portion of the population. It’s like, whoa, cowboy. Slow down. Now, you know, this stuff leads to, you know, the terminators or whatever else, like, how are you thinking about and dealing with with those aspects and and those concerns? Yeah,
Scott Newman 42:33
I mean, so in our in our industry, and, you know, the two big things are PII and PCI, like credit card information, social security numbers, medical history. So number first and foremost is that we we redact everything that can be PII or PCI sensitive in nature. You know, that’s super important. So that kind of information never makes it into an LLM model. Got it. So we ensure that we’re not doing that, and then additionally, we have a number of closed loop LLM models that we’re fine tuning and training so that the data isn’t really getting out into the public domain. You know, I think it’s really important as we’re thinking about, you know, like you said, the ethics, but the security, it’s all about information, right? And the more that contact centers are deploying technology where humans aren’t taking a lot of that information anymore, you know, in that it’s being encrypted and safeguarded and stored away in secure locations, I think that’s that’s really where we’re actually going to get a lot more security over humans actually taking that information over time. And so if anything, I would think that with the advancement of AI and some of these tools, payment card capture using web links or text messages, as opposed to reading it over the phone or typing in your social security number on a secure portal, rather than giving it out over the phone. These are the types of things that I think will actually enhance the level of security and should provide more comfort for a lot of consumers. That’s,
Kyle James 44:10
that’s good. I think that’s, that’s the kind of stuff everybody needs to hear, like there are these safeguards in place, and, you know, we’re all aware of it, we’re not ignoring it. But you know, it’s, it gets into the devil the details. Um, Scott, what have I not asked you that I should be asking you, what if we, you know, we’ve talked about a lot of awesome, great subjects, but as we kind of wrap up, what else? What else you want to share with people that maybe you’re doing or you’re seeing or you’re thinking about right now that you know just needs to be out in the into the general ether.
Scott Newman 44:41
Yeah. I mean, just maybe a couple thoughts for closing like, you know, first is, I think, and this is dialog we have with clients all the time. Ai, is really cool. It’s exciting. There’s a lot of advancements happening in the industry, but for clients that are potentially. Listening. I mean, we shouldn’t be deploying AI just because we want to have the newest, coolest product out on the market. You know, for us, we’ve always taken an approach internally, with our with our team, where, you know, if a client asks for something, our first question is, what problem you’re Are you trying to solve? Because AI today is not the right answer for all the problems, and overwhelmingly, customers still want the human in the loop interactions. You know, AI is not reliable enough at scale to be customer facing, to handle all of the queries and concerns and issues, and so, you know, I think it’s important that we take a very slow, methodical approach to how we deploy AI into customer facing environments. And so really, a lot of the times is, you know, what is it you’re trying to solve? Because there’s probably 10 different ways to accomplish that without just trying to launch something out there that potentially would have more risk and liability. And then, you know? So that’s sort of advice number one, the other thing is, I think our internal philosophy is we want to lean into AI and development and things where we can but it is all around making the human interaction better a better experience, whether it’s making our team members have a better time and enjoyable job, and creating better culture, which then turns to, you know, good customer interactions, or whether trying to, you know, genuinely just make better customer interactions by access to information. I think we really need to think about the fact that, you know this, humans still need to be in the loop today. What that will look like in five to 10 years. I have no idea, because, as we talked about, it’s developing at a crazy pace, but, you know, it, that’s where most of our efforts have gone in. How do we make the our employees experience better? This is if in our business, this is not replace AI is not replacing jobs. If anything, we’re getting more opportunities, you know, for for different kinds of development work and other types of support. And so I think it’s important to keep in mind that we can be excited about the technology and what’s coming, but you know, you know, the days of AI replacing huge swaths of human jobs, I just don’t think are upon us at this certain point in time. I
Kyle James 47:18
think that’s a hopeful message, take away, like what you’re doing is, is taking away the stress and insecurity about your job and making it more fun and being able to help people right? Because, like you said, it’s all about empathy and compassion, and the people that are good at that, that are good in these kind of roles anyway, thrive on on the success of that. If these things are enabling them to do that better, they’re going to be more fulfilled and satisfied, right? And Thrive themselves
Scott Newman 47:46
100% I mean, and you know that this, some of that is that we’re going to be more efficient at our jobs. And so you know where we might need 1000 agents to provide support. Maybe they’re all you know, able to cut a minute off of their talk time, and now we only need 900 so, you know, there might be some efficiency gains, but we’re also staffing so many other jobs for other types of needs to make up the difference. And so I really, I’m excited about the future of of the space. I think, you know, customer support should just incrementally get better over the coming years. I think consumers have higher demands than they’ve ever had before. Rightfully so. And I really do feel like for the first time in a long time, that the technology is going to be a really unique way to help, help ensure that the consumers are getting the types of experiences that that they’re looking for, and that, you know, we’re helping, our team members do a better job. Love
Kyle James 48:42
it. I love it. Scott, thank you so much for joining us. I hope everybody out there found this as insightful and a good of a discussion as I did. It feels like every time I talk to you, there’s so much that I come away with. So thank you once again for joining the podcast. How could people be of help for you? Help with you. How can people connect with you? What services can they offer to you and and, you know, ways that they can engage with you. Yeah,
Scott Newman 49:07
I mean, I’m on LinkedIn. Also, you know, one source, corp.com is our, is our corporate website. I think, you know, where, where anybody has some innovative ideas, or, you know, technology platforms that might be of interest or, you know, fit into the space we’d love to connect and any clients that are in need of, you know, top tier customer support leaning into technology and as well as, you know, back office rules, we’d love to have the opportunity to connect,
Kyle James 49:33
awesome, awesome. I love it. And for everybody out there, if you enjoyed this episode, please subscribe. Like give it a six star review. We’ll take five, but we try to push people to six star reviews. And thanks everybody for tuning in, and we’ll be back next week with another GTM innovators. Until then, everybody keep learning. Everybody keep growing.