Creating an AI Roadmap for Your Contact Center from the Ground Up!

00:00Hi guys, Tom Laird here. You've heard me break down So, this was a struggle for me to get down here. I live in Erie, Pennsylvania. It's about 18 degrees with, we did get rid of our snow, but we normally have about two feet of,., snow right, So How the heck do you employ it in your actual contact center? So this is my. Flex page right. Here. So you know who the heck I am. So I'm Tom Laird. I have a BPO Expivia is about six 700 seat USA BPO work with lots of different from financial services to health care to retail. We also have an AI customer experience startup called auto QA O2TO where we're fully automating QA in the contact center and I'll talk a little bit more about that. But again, I think for this being a nice customer event or people that may be coming on to nice.

01:57I am one of the members, I-I think there's only two BPOs on the executive advisory council. So all the new technology that's coming out, we get to play with it a little bit. We, we get to tell them if it stinks or not. I want to kind of talk a little bit with that stuff today with you guys too. And this is just quickly on Expivia, like I have every tool. So, at the end of this, If you guys have any questions on some of the things that we're going to talk about, be it workforce intelligence, workforce management, for non-, BPO, for BPO, for an internal contact center, all the tools that we have, and I put here too, I am enlightened, interested, right? to the AI tool. We have a ton of third, third party AI tools that we're utilizing now. And I want to talk about that stuff too.

02:44So again, this is kind of a, what is real? What I believe this is my opinion. This is one of the awesome things here too. Nice that, hey, come talk to him as a customer. I'm like, absolutely. They never say, hey, talk about this. Hey, make sure you bring this product into our thing. I sent my deck to Morgan and she never even said anything. I don't even know if this thing is good or not. So this is all my opinion on stuff from a entertnology call center guide to technology call center people. I only have I think Liz went her full a lot of time. My decks. Like 27 minutes. So if Morgan, please, she's not going to yell at me. Please. If you guys, have any questions as we go, um,. Especially

[Gap 03:23–03:53: no reliable transcription]

03:53When I'm looking, looking at my own organization, This is the plan that we have found to work the best and I think hopefully out. How to set up the AI plan and then at the end we're just gonna talk about stuff. I H I, H and maybe that will help you at the end. So I need a little filler, so Morgan didn't yell at me. OK, so this is like the slap in the face. This is what we have kind of come up from a consulting side speaking to a lot of different analysts. And we are planning in the a BPO space, so the outsourcer space. Which doesn't mean anything to you guys here, because it's going to be you guys too. to agent reduction in the next three years. Now, that's not because of the generative AI chatbot chat bot, right? And I struggle with this talk.

05:15A lot of times because we have totally different depth or thoughts about what AI is, right? Because I guarantee the vast majority and you hear we're either are here on your own or sent here to figure out how to get on a road map to become generative AI to lower our agent headcount. And then, you know, you lie to yourself and say, you can improve the customer experience, right? Right, with AI. So, there's so many things that I, I think, can help with this number that can actually improve the aging experience that have nothing to do with the generative AI first touch point chatbot, although that's part of the deal too. Okay, so when we talk about that full meal deal, First touchpoint generative AI LLM chatbot that's where Everybody wants to go and we have some examples of why it's stinks right now Liz gave some examples. [pricing remark omitted]

07:03Or let's talk to the vendor that has the, my chatbot.isreallygood.ai, that's what everybody has now, right?. Everybody's.ai, throw one of these things on there. Then I love this one, and you guys all saw this. This is, I think it was Ruby. The DPD chatbot, where they basically made the, the out of the customer made the chatbot talk just mad smack on the company. Just, you know, the worst, the worst delivery firm in the world, they're unreliable, based on the prompting that they gave. And then I did, this is family friendly, so I blocked that out, but it was was f yeah, I'll do my best. I think it's a lot of ways a ways, but that doesn't mean that AI is a ways a ways that makes that a ways ways. Right? There's a lot that we can that we can do right now. So I want to, give you R-like tangible, literally things that you can take and actually hopefully look to implement.

08:08So howossession.io How do you start? Again, my, my chatbot is great.ai. Every single vendor you talk to, says, hey, we can get stuff up really quick. I bought it up really fast. Sorry, and I'm not so much talking about the CCaaS players. I'm talking about a, third-party guys that are using, using ChatGPT, they're using these different LLMs to quickly give, give you some type of garbage, right? It's not plug-and-play. There is work that needs to be done, but it doesn't have to be a ton of work, especially if you're road mapping and have a plan. Be. careful what you buy in 2024. Nobody really, we all talked about, about AI until ChatGPT came out. And then it was like, we have to have AI. It's a- it's a year old a year. So again, do we need to get on this path? I totally agree with the six the nine months.

08:59You need to be on a path. It doesn't need to be full, ready, given self-service I am the best, my dream job, right? Because when I go consult for a company, they'll be like, hey, will you come on this demo with us? We're doing this with self-service.ai company. I'm like, yes, I will go on that demo with you. And we'll see this demo, and it's the greatest thing ever, right? And then I'm like, all right, I'm either gonna be like a level 12 pissed off customer. Right, let's see how it reacts to that. They all stink, right? Or I'm just gonna, I'm gonna try to trick it. I'm gonna it's gonna work to a certain extent. So don't don't fall for demos I don't think most people do anymore, but especially when you're talking third-party guys, right? outside of the See cast which they've taken a slow methodical pace to this where I've just seen so many people who've made mistakes who then come to me like hey, will you come fix us?

10:14I'm like, no, dude. I can't fix that. That's terrible. You got to get that off because there's no there's no helping it all right I think this is important especially when no one raised their hand that they're on a path this is what you need to do like today if I can open this okay you guys see this too this is I did this deck myself and I was really proud of my animations like look at that pretty good so think about integrations today I think that's the number one thing other than the next thing and I'm going to talk about so So all of your data, whether that be, you know, let's take a credit union, for an example, like they have their core, let's make sure the core is integrated, they probably do because they're probably doing some type of self-service online banking. Let's take the customer-facing data and a sales force, whatever they're using, let's integrate that into the telephony.

11:10Let's get every single data source, at least from the customer experience side. I totally agree with Liz, like this needs to be a company-wide deal, but when we're We're talking just about CX. Let's make sure that in 2024, you guys start the path. If your integrations stink, into your CCaaS, into your telephony platform, that's a huge piece of this. The other, other thing I don't have on here is, we don't have voice yet. Really, for AI, when you're talking about ChatGPT, when you're talking about those type of tools. Transcripts can be difficult. Now, CXone I, they just kind of came out the way that you can get transcripts, but think that through as well But when, when you're looking at Agent Assist, when you're looking at, you know, how can AI help from a self-service model, it needs to know all your stuff, right? It needs to know your tools.

12:27It needs to know how you service customers. It needs to know codes that you use. Everything that an agent would need to service that customer, you need to have in a set of data. Hopefully it's a KMS or something that you don't have a bunch of Excel spread spreadsheets that you guys are just plugging Stuffed to start in 2024. no plan say, hey, I want to lower head-count. There's huge, huge dynamics at work here with different areas of the organization. And there's different tools that match what those do. So coming up with almost a company-wide thought process of, what are the outcomes? Are we looking to increase our customer experience? We need MPS to be different. We need to rise that. But I also need to lower my head-count by x. have to come up with a company-wide understanding what you need, which will incorporate your roadmap of what tools you go after first.

13:51disconnects, then they want one tool, you get another tool, they get the tool, you hate it, you don't use it, it's a disaster, and I see that all the time. Think through security. New security, new privacy issues, what data are you going to give any type of LLM that you're going to use? Are you masking that data? Is it HIPAA compliant?, new security thoughts, new policies, procedures I- think is something that we're not talking about either. Like what is what. What is the deal with that right? How do we how do we figure that out? what is the policy for that? How do you QA AI? Not about, not that You have a chatbot, that you're utilizing. do you QA it with like normal like these are questions I asked I asked myself. It's part of one of the reasons that we're starting up like our little AI QA platform But think about that in your organization. How do you know if it's quote-unquote hallucinating. You just assuming that it's right because it's AI AI.

15:18Or I wouldn't. Right? Another thing to think about, how we're QA-ing AI. And again, I see this a lot. People have mistakes with that, but then they end up hiring like three or four people at the QA. What they're doing, because they're doing, because they're doing, that it almost like defeats the purpose of the whole thing. So again, don't get caught with, you know, not thinking things totally through. God, I just, and I keep hearing the demo thing. Don't take demos at face value, like just sticks with me. If you have no plan whatsoever and you are just starting, I think this is a great roadmap, right? This is exactly how we kind of thought, the thought process that we did before we went out and we bought software and came up with a plan, integrated all the stuff that we could possibly integrated and prepared for this.

16:09I believe that if you just did this this year, a really good spot because the technology that's coming is coming. It's really, it's almost cooler. What's about to-have. We and talk, talk about that too. Some cool stuff that's happening. Maybe in the next two years. So you're falling behind if you don't do this if you just sit on your but just start down the road. All right. Now, let's say you did this or let's say that you're in a spot where we want to purchase some stuff. We want a really good thing. What should you purchase? And again, this is my opinion only. This is what I believe-believe is real today. this is what I would start with and I wish I had better animations for it but I don't. So the first place that I would start with is your legacy tools that through again we're assuming that you have kind of that that KMS and everything's in in place.

17:06Workforce Management who who knows what Workforce Intelligence is on CXone. Who uses Workforce Intelligence on CXone. Love you guys. All right, we're gonna talk about that either. It's the most underutilized, I think, AI tool now in the whole suite Right. Come on guys. Let's grow up. Let's get some analytics, right, right? I think that those type of tools get you thinking down the right path. Okay, so we're cool with that. What are some of the now the tools that are Knewer that I believe are the most mature that I would say. Hey, I, I think that that could help you depending on your use case, not everybody, but depending on your use case is, is agent assist, Realtime, Realtime, Realtime Analytics, Realtime, Realtime Analytics, Supervisor Assist, Realtime, Realtime Analytics, Supervisor Assist Obviously the more more complex your contact center is, the more that these tools are a help.

18:37If you have a quick 2 minute and 30 call, and half of your calls are password reset, example, but you get what I'm saying. I wouldn't get any of that. I wouldn't do that. That's not most of us us. Most. Most of uphe have complex situations of how skills are set up, how the calls come in, where agents are asking a ton of questions and slacker in teams all the time. And we can now automate that depending on if you have your back end and you have your KMS and you have all that stuff built out well. Agent Agent Assist is amazing. For us being in a BPO, I do not have, and I'm looking at the CXone and the Enlightened, and the, the we use a third-Party Agent Assist. I will tell you, in every single RFP that a have seen in the last 18 months, Agent Assist is not something that companies are asking for.

19:31They're demanding it. So that means agent assist is becoming very mainstream. It's something to look at. It's not that difficult to employ as long as you kind of have some of your backend this is the question I always get on that as well. Depending on your use a case like I have seen a 15 to 30 second, 30 second is aggressive. Lowering of average, average handle time. If you have a really good, and say you have a five to, to seven minute call. I think you can lower it about that much which is again if you have a hundred agents if you have 10,000 agents if you have 5,000 agents if you have 5,000 agents it's a big difference and you had nothing to do with. With generative chat with generative chat bots you just lowered it by helping your agents doing that which also could lower head count. Okay so I thought, this was just totally ingenious right I got my green that's the good stuff this is like well it's common And that's like, oh, God, right?

20:30So then I call the GPT tools, I don't know if that's not really the official name form,, but that's kind of what I call, right? So auto summary, that should be in the green. That's why I kind of have it first. I think auto summarization is great, especially if you have longer after call work work. If you don't have long after call work, then I think it's a waste of money. But if you have two minutes of memoing and then you got to fill out forms and your after call work or your wrap time is over a minute, that's crazy, man. Especially times in the amount of agents that you have when it can instantly summarize what you have through to do with your CRM. Amazing tool, super mature, ready to go. All right, I'm going to talk about Auto-QA, because there's a lot of, with Auto-QA.

21:21I think we are a, and again, I'm not, I'm, I'm not, I'm not here to, I'm not, I'm a pubbing my, what we're doing, but we have a company and we're trying to basically use chat G, GPT, to fully automate the, the forms, right? I don't want any proprietary scoring. I want this form that if I know, an agent has an 89 on it, 89 is, and I want that same exact form. And there's a lot of auto QA, that's coming out, Basically, they're giving you a proprietary scoring. They they're basically giving you sentiment scoring. And I think that there's 2041, you never heard that. And I think you need to be careful with the auto QA stuff. What what it can do today now in the in the year. [QA coverage remark omitted]

22:23[QA coverage remark omitted] Maybe you guys can prove me wrong on that. But if they're using some type of LLM or ChatGPT model, the fastest that we've seen the APIs is about a minute and a. Half to two minutes per call, for each score, we can do a bunch of them, but be careful when you hear those type Ask what that score really means. Now, I can do sentiment scores on our analytics. 100 calls, well that's great, but it's not scoring calls, right? Be careful with that, something that you're, I have started to see more is education and role-playing, with kind of these GPT models. So, you know, having an agent go through a library, a ticked off customer, a happy The reason you're seeing it be not fully deployed is because it's mostly just chat right now, right? Because voice isn't totally there yet, expensive there, but you're gonna see that too. So that's like a tool that don't worry about that now, but next year when there's it's voice, that thing's gonna fly.

23:42Okay and then quickly we have the the chat box and, and the,.

[Gap 24:17–24:47: no reliable transcription]

24:47F H F F I'm always trying to say what to a what to a what to a what to a what to a what to a. I don't have any French speaking here in Pennsylvania, but to be able to have this, this is coming. bSo full-voice so that you can speak hundreds of languages and act like you're having the same. Same conversation when you're actually speaking a different. Different language predictive analytics. That's that's kind of a, I think that's obvious, right? Being able to all the different types of the I'm tired of dashboards, I don't want to... Look at data. I don't just go and sort by our average handle. I, I just want, to say things like, hey, what and You know, the longest handle, you know, the. The longest handle handle reps, boom, that's there.

27:14Hey, how many, can you give me a sample of calls that, for when, customers use the phrase too expensive. Yes, your day, boom there, right? So having kind of that prompting mentality, translate into the data that we want, right? That's coming, that's almost here now. even more as we go, as we look at using our head to kind of tell it what we want. I've picked off that voice real-time sentiment, real-time sentiment. And when that customer is yelling, we really know that they're yelling because we're a way of using voice and we're not using transcripts for any of the analytics. And every time I see IoT, I don the I think the idea of today, but that's not what that stands for. But you're starting to see devices talk to themselves. My refrigerator's light bulb. It goes out. I set up some parameters where it can reach out to a bot at the at Honeywell or Yeah.

28:59So I think you started the agent level I think you start with the tools that you have now I think you start with the legacy tools that are the most mature. If you want to actually buy AI tools right You start working at the ChatGPT type of things to eliminate Time if that's what your deal is if you're looking for CX and an improvement to that the agent assist in those type of tools I think you work to the customer level, meaning that, questions, just rambling on here. No, I'm really that good the awesome. Okay, so the. Reiterate before I get into some of the cool tools and how to utilize them. If you're just starting out, which seems like a lot of you guys are now build-build out your KMS, focus on your integrations, start to use the legacy AI tools and use them differently and I'm gonna show you how to use them differently, and it's awesome. Let's, talk about that.

30:27We hear from nice, from nice, from nice, from nice, from nice, from nice, here's your set of to 8030, which I still don't know why 80 Well, I don't have a budget for that.- I say, okay, what we can do is we can turn on analytics I hadverified negative sentiment at the beginning of a call when they wait three minutes some type of lowering of that account. Like that's what AI does, right? K And you can say, okay, okay, listen, we're going to roadmap. We're going to be smart with this,. But there's other things I can do, off so that you can do the thing right. H that's one thing that I love. We use analytics as as well. I incentive and again, maybe this isn't important for what we're talking about here, here. But again, remember, my last thing was anything I want to talk about. And I think that this is so cool right. we're actually paying, paying and it's sending our reps off of positive sentiment. So so many of And me, we go to have our agents and there's a cross-sell, there's an upsell, there's some type of sales component.

[Gap 33:17–33:47: no reliable transcription]

33:47totally crazy, how that totally builds a different type of culture in your organization when they know they're getting paid off of being nice and now we have a way to actually show it other than, they QA'd my call, how was nice on that? I swear to God it was nice on that call right? Now that you can take kind of all that away. And now I'm going to talk about workforce intelligence. The most underutilized tool in the whole., whole CXone suite, one of the things that I would start with as an AI tool and all of you have it. You don't need to pay anymore for, if it's not on your CXone, go tell your team, go tell. Tell somebody like, hey, this needs to be on there and they'll put it on. and it's core and it's kind of babyish is where you send you an email or send anybody an email if you know service level drops or handle time goes. That's the basics like baby part of it.

34:39The cool part and how it's getting better and better and better is AI is being infused into the engine so that we're moving agents in and out of skills based on reset service levels right that's the one another awesome use case so I can guarantee service level for a customer that has a sales let's say a password reset I don't know why I keep saying that but sales password reset regular customer service and let's say a shipping issue No one uses this to the point. Nobody talks about it. It is absolutely unbelievable. It is awesome. And there's so much you can do with it. Go play with it, okay? So that's if you were again, if you had no idea about what you're doing, okay?, hopefully that's enough, tangible, especially when no one raised their hand And so when we method especially when I started to see customers want tools.

37:44Like Agent Assist for me, I have to have, it's the cost of doing business now, right? But it's something that for an internal contact center If you do it right, it will lower your handle time, It will raise your CX because customers have information if you're are doing it properly. I think we were so methodical with it that honestly I'm a pretty good voice on what st companies in 2019 to 80% of every single booth was was my top grade AI dot AI. And I think that there's a lot of people that have fallen for it. They've had a poor experience with it. They don't trust it. And it's kind of like to look at a full sub-service model yet. Now, you talked to me in five years, that could be totally different. But if they're just starting, just be very wary of that. I really think, and again, they've not told me to say this, the CXone and NICE has been extremely methodical.

39:12They have not raced to this. They're pretty much the only C-cast player with their own attached AI model, right? Five-Nine, Genesys, all those guys, you got to slap stuff on. And you can still slap stuff on if you want it from a CXone standpoint, but the tools that they're coming in is what I am most excited with because it's gonna be fully integrated with all my stuff. So again, that's a long way around. I don't know if I answered your question, but that's kind of what scared me or the things that I would definitely be careful of. Yes. I'm loud. I don't get it. Yeah, so if you talk about live years, I was thinking about live years So, for us, we had to change, I changed my whole thought process, right? And again, I think I was telling Morgan, or I forget who, I was telling somebody, I gave a talk that was basically this talk, but it was for a group of people here that were all VP owners, or they're all call center outsources, and you should see some scared people in that group, right? Because no one understands how to change their business model.

40:33So for us, I'm totally fine now. And I have accepted that I want a customer to say, hey, I have 100 seats, but I need to get that down to 45 seats. And I'm now at the point where I'm like, yes, let's do that. Let's do it through agent led tools. Let's do it with self service. Let's do it with some type of, of maybe some type of hybrid generator bought to take away some of the easy stuff. There's tools that the CXone platform has now where I can basically load data in and they can say, hey, these are the calls that you need to turn into self-service. So I think the mentality change is the biggest thing for me and understanding that the tools that I'm gonna buy in the future, they're gonna lower my head count. My whole experience as being an owner is gonna change, But I think that's the most value for a lot of these guys and for our customers is again, to be able to, I don't like the companies that say, hey, just, I wanna move all this to self-service.

41:30I'm like, guys, it's gonna be awful. It's gonna be a disaster, right? But if there's a plan to say, hey, let's do this from the agent level, let's take away some of this easy stuff. And again, that's not rocket science. We've been doing that forever. But now we have cooler tools to do it with. And I think that's the biggest difference. Yes, the RFPs. Yeah, like you don't have it, go beat it. Like you're not getting any of our business. And that's when I said, this needs to change. And I said, and I would like to be first, like first to do things. And I made a lot of mistakes doing that. But I think in this world, it's been right. And again, to have auto summarization for a client that has, and we're not a huge BPO, I'm 600 seats, like I'm not 10,000 seats.

42:22So a big client for us is maybe 40, 50 seats. But if they have, we say, hey, we have these 50 seats, and our after-call work is normally eight minutes. I'm like, yeah, let's go, right? Because I can provide instant value, instant savings right there. So I think once I got more into it, and the other thing is most of, I don't wanna say you guys, but most customers that come to us have no clue about what, what they're doing payout. So if I can be that kind of that guiding force, I think there's a lot of things that you can come up with a really good plan, I think that there's a lot of things you really cool with your companies. If you have a leadership group, that will kind of let you have a say as well. Obviously if you guys are here.

43:15Yes. Thank you. Appreciate you. We had, we spent again, I don't want to, again, maybe this is612 I started early. Like I came, I ran a 1800 C, 1800 C BPO, mostly financial services, you know, 2008, 2009, 2010 came. [pricing remark omitted] God knows what it is now. So we said, we tried to create as much content as we possibly could in the CX and the customer service space as humanly possible to help to as many people as possible. And, so I don't have any sales people in my company. There is not one sales person, I guess it's just me. All we do, do is create as much content for somebody who is just starting out.

44:49That call center and you become the supervisor, you become that out of procurement, who then comes to me because we've helped them out. So-, I think, again, I would love for you guys to just get advice from a call center podcast. Make sure you follow me on LinkedIn. I'd like to talk to you guys more on that. I'd love to see some of the things that you guys are doing. But when it comes to like the consulting side, I think that these things are more important. Most of the consultants, how many of you guys go on LinkedIn and you to talk to the, really see the real technology. They have as a real technology. But, again, you're not. You're not going on a buying spree. You're not going on a buying spree. You're going to find out what's good for your organization I can actually help the things that you need help with.

46:00And that's the tools that, I think, that you go out with. But again, from the consulting side, it's tough out there. This is just so new. And everybody's use case is so different that, you know, depending on what buddy they have at, you know, my call center.ai that they're kind of pushing on you I think it's the kind of the way that I've seen things go here lately. Thank you guys. Thank you so much.