𝗕𝗶𝗴 𝗜𝗺𝗽𝗮𝗰𝘁, 𝗦𝗺𝗮𝗹𝗹 𝗕𝘂𝗱𝗴𝗲𝘁: 𝗔𝗜 𝗦𝘁𝗿𝗮𝘁𝗲𝗴𝗶𝗲𝘀 𝗳𝗼𝗿 𝗦𝗠𝗕 𝗖𝗼𝗻𝘁𝗮𝗰𝘁 𝗖𝗲𝗻𝘁𝗲𝗿𝘀

00:00Hi guys, Tom Laird here. You've heard me break down contact center technology on the podcast time and time again. Now it's time for me to help you with things like on all-things call center we'll cover it all from call center operations hiring culture technology and education we're here to give you actionable items to improve the quality of yours and your customers experience. This is an evolving industry with creative So, hopefully, we can improve your technology, a little bit, because the overall theme here, right, and you want to improve the customer experience. What's going on, everybody. It's good to be back on the pod, on the advice from a call-center And the question is, again, this ongoing thing

[Gap 01:50–02:08: no reliable transcription]

02:08but probably focus more on under 100 seats, right? but I think any 1,000 to 5, I'm talking about. I'm not talking if, you know, you got 20,000 seats, you know, all across the United States, you're not going to value out of this, Let's dive into this and I got a bunch of notes here. So if you guys see me, kind of looking off here to the right. First thing is we see a lot, a, I evolution. We see a lot of larger contact centers, I think that that I. And I think that makes sense, right? Because to get ROI on AI. When it comes to kind of that first touch point, chat bot, agent facing, right. You want and you need volume, right? It's, especially if you're using a vendor, right? It's relatively still really expensive. If you don't have a ton of volume to utilize that technology, the cost savings That you think you're going to get, larger enterprises have that ball, they have, you know, really millions and millions of interactions, that again, the numbers.

03:39and you think you're going to get larger enterprises, right? They have you know really millions and millions of interactions that again the numbers game really works for them. thing, you know, they can spend the money either on the third party and get that technology in here, or they have the giant, you know, a hundred person team, and that's probably. Probably small for some of them can start to utilize these tools on their own. And again, still really expensive, but the RI is going to be there just I just because of the volume

[Gap 04:02–04:11: no reliable transcription]

04:11that we've seen no one falling in 2024. But you see kind of these massive cost savings that these companies are getting. But again, number one, they don't M they don't tell you what they paid for it or what they the development costs of what that costs as well, again, there's there's this, this, that that's what everyone's doing. It's I will tell you the really cool thing and I have to be kind of confidential here. But, you know, I was part of the a part of the nice CXone Executive Customer Council. And there's very large organizations on that. Like it's public knowledge. You can kind of go and see what companies are on there, but Monster, Fortune 500, Fortune 100, Fortune 10 companies. And the little old me, which is still pretty cool. But I will tell you there's a Fortune 10 company that is just utilizing AI in a very small segment to test it out, for their kind of AI council.

05:05They have struggled with it just from a hardware standpoint and also from a data standpoint. They've been able to utilize a lot of their data, but it's so much data that it's not all clean, so sometimes it doesn't hallucinate. I think that's what we're starting to see now too, is again, yeah, there's the hallucination. But there's also we're confusing our own AI just with the data that we that we, just with the data that we have. So even from a larger enterprise standpoint, it's still a lot of kind of touch and go. I will tell you in that group, no one's really just But if you go right now to the My Chat, everybody is saying, we can do AI really quick, just give me an FAQ, we'll throw this on there, and that's when the garbage happens. Don't There's not a lot of really good companies out there.

06:24There are, but if they're really good, they're crazy expensive, right? Not gonna name any names. If you want to DM me, I can tell you about names. but for this, once you're going to pay me, come on, right? I'm not going to really talk about too many companies. This is what you need to do in 2024, right? Forget about purchasing anything at this second, whether it is customer facing, whether it is kind of agent facing, whether it is kind of agent facing, whether it is, kind of agent facing, forget AI until you have kind of these things in order. Number one is you have full integrations into all your stuff, right? Your CCaaS, your datasets, your CRM, anything that you guys can integrate, that's what you

[Gap 06:59–07:28: no reliable transcription]

07:28No one one has very large companies to very small companies. We're seeing this in our BPO It's a real struggle. We're trying to This the tools to hopefully try to help with this, but you need to have a robust knowledge management system, a KMS. You a repository for your data that is clean, that can be updated easily-easily, and that everybody kind of understands how it works. can't have. I've seen so many companies try to do this, but they have six different departments all. I'll throw in a bunch of stuff in here and nobody's really looking at it. This is number one hold back right now for AI for the vast majority of companies is the data. If you don't have the data to either train or have the AI read off, then there's a struggle there. So, you know, I think 2024, 2025, you a lot of companies out, but if you don't have a good KMS, forget about AI for right now, right? You're still doing AI, you and you're still working on that roadmap, if you're working on your KMS, but it's vastly important.

08:43Let's just say that you, have integrations, right? And let's say that you have a pretty good KMS, right? Then we have, have to start to think about things like, what are some of the problems that you want to solve? landish AI is, has a lot of different tools for a lot of different use cases. Like you're not going to hear this from any technology guy, but they're just going to start to sell you stuff. But let's talk about the different tools, right? if you want to lower your handle time, right? then we're looking at more agent type tools. We're looking at maybe like an agent assist, that's right. Possibly an auto summarization to get rid of your work or your, wrap if it's long, right? If it's long, right? If you have a 20 second or a 15 second after call work with very limited memoing, the don't get auto summarization.

09:30Don't do that. don't just buy a tool to, to have a tool, right? So that's one thing. If you want to look at, you know, we need to raise our CSAT in our, in our NPS, right? I think I think you can look at a tool like it like an agent assist look at an auto QA, right? Really dig deeper into what you're happening now use AI more as an analytic tool that also as an agent assist tool, right? You can also look at some things like real I'm sentiment and real time kind of supervisor. I'm I'm I'm I'm, I a or from a customer standpoint. Like I think those things are here, they're relatively mature and you can utilize them. The other thing is we're running into this with auto QA is security and policies. a lot of. Companies don't know, what their threshold is for AI, the they don't know, what is the, what is the tolerance that they have for using it?

10:36how far can they push it? Do you have to have an internal LLM? Can you have a an outgoing model where, you know, you can use a ChatGPT, you know, through APIs. You know, what is the threshold that you're willing to do. and I think that's a company that has to happen., right? If you start doing AI and you don't have a lot of these things in place, I think you're going to run into some issues. So, you know, those are some things that need to be working before you kind of roll out your AI. The next thing is, how do you QA your AI? Do you ever think about that? That's a question I ask in a lot of different meetings. I think that that's where kind of, again, an auto QA can kind of really go really well if you do have a customer facing chatbot or some type of IBA.

11:22So think that through. How are you, are you just gonna do random human calibrations with AI looking at transcripts. Are you going to use AI to do it? Are you going to use an auto QA? But you have to make sure that what your AI is saying is right. And I think that's kind of an advantage for a smaller company, right? Because there's not that many transactions. So you can maybe, you can kind of reel that in a little bit quicker than some of these larger organizations that are getting really nervous about where things could go. And one mistake could really cost them a ton of money. And the last thing I'm going to say is never take a demo for face value, right? If I show you an auto QA demo, I'm showing you live data, I'm doing things live. Most of these other companies, you have to make sure that you're asking a ton of questions that you're going deep into it, that you get a chance to play with it before you pay for it.

12:15Most of them won't let you do that, right? There's a lot that kind of goes into it. So again, before you even start to think about what AI tools should you get, look and get your integrations in order. Number two, make sure you have a robust KMS and you have a process. It's not just setting up a KMS. What is the process for getting rid of data that is not good anymore, that is not correct? Putting in new data. How does that whole process work? Really understanding what outcomes you need. What AI do we want? Do we need a customer facing? Do we have really, I see so many credit unions and banks, their self-service is awesome, right? There's integrations into the core already. They can do a ton of different things, from filling out a loan application to getting account balances to move money around to pay credit cards.

13:04They can do all this stuff. Why spend a ton of money, right, to maybe redo that with, I don't know, better voice, add some FAQs in there? So I think you have to be kind of careful. And again, there's some things that I'm missing here from maybe in a voice biometrics and kind of those really cool things. But be careful and start to think through what do we really need other than we're just going to check a box to say that we have AI. Look at your security and your policies. What is your threshold for this? What is your company allowing you to do? How you're going to QA your AI and then again, be careful on the demos. There's some scary stuff out there that is made to look really, really, really cool. Again, let's talk about from an SMB standpoint, where are we today?

13:53So, I think there's another misunderstanding of AI for most people is that chatbot, right? That bot that's going to take over and do full self-service, that's what people are thinking when they hear AI. But what they don't realize is that a lot of the tools that they're utilizing now are being infused with AI and being infused with AI features. And if you start to use these tools in different ways, you know, you can kind of bridge that gap between what we really need, what we really need from a, you know, that what we think AI is to what the tools that we have right now, right. And we can kind of bridge that so that you're still saving money. You're still having a better experience for your agents better experience from your customers. Hey, Nate, what's up, buddy? Good, good to See you back in the Philippines. Nice.

14:50But again, if you look at these legacy tools, workforce management, workforce intelligence, analytics and All of these things that we've been able to do for a long time are starting to get really I because they're being infused with AI, and you can start to use them in different ways, you know, how good this agent is, what's the analytics and the sentiment say about this agent taking and all of this data that a human being couldn't and sending that call to the right agent at the the right time, right? Not just using the next available agent, right, but actually looking at all the data, right, that maybe even an agent or a customer has in their CRM, like to see maybe if there is a, high value customer, if there is someone who's definitely having a different issue, right? moving them like, like, uh, I don't. I think that there's a lot we can do there that is happening with AI that people don't think the AI that people don kind of AI infused is really, really good.

16:30So again, those are, we can start to play with more of an AI standpoint, right? Maybe start to check the customer says about things like- auto summarization when you for auto summar when they have an Africa work of like 10 seconds, right? In very little memo. Right. So two minute after call work and you need to have perfect memos, right?, because you're a pharmaceutical company and you have to have different names for your drug products. That all makes sense, right? You don't want the agents to make those mistakes. Then we start to get into where we are with chat about a lot is the tools that we have today are not going to this Like I think, I think Nate as you're talking about with with translations, like that stuff's kind of here now Now, that's getting really, really good. You know, really the ability and there's a lot of, I don't have a problem with it. I think it's a mess. I know a lot of people are having some kind of ethical issues with it, right like that, you know, having a, you know, an agent in India or in the Philippines actually sound like somebody who's, you know, in the, I don't know- Oklahoma right and have that draw I think it's a great use of technology but I think that, to

[Gap 20:35–21:19: no reliable transcription]

21:19service level, we have calls in queue,- we have how many agents Can you also tell me what agents have the best sentiment and who had the best handle time yesterday and then what agents really A, can you pop them and can you make sure that I have, you know, five or four things that they struggled with yesterday. Those voice It's coming, you know, it's a pretty big drawback too right now with even with our Auto-QA, right? We have a sentiment prompt that I think works extremely well but it's still not voice driven, right? The technology is not just not there yet for full voice, but that's coming this year. In fact, I'm disappointed here already. There's a lot that we can do with it when it comes to looking at sarcasm, looking at wait times or silence on calls, looking at just the tone of the agent.

23:09I think that really starts to be a much higher level percentage of accuracy once we have. And look at the Internet of Things. Looking at AI, talking to a, AI. I think that stuff's really cool. Having your refrigerator and the light bulb go out. to, kind of, heard that exampleiasco, right? And then, again, just going, having machines start to talk to machines from a service model so that humans don't even have to get involved with it.. I think, you know, that's coming, and, again, looking at, like, video avatars and all of that kind of crazy stuff that's kind of the next generation of AI, I think, is there too. So again, there's so much cool stuff that's going to happen that you're going to have to have your, your, your house and right. It's people, again, I, I-I'm going off on the tangents, but you think you're getting ahead because you, you bought a chat bot, you have an FAQ and you have that on your website, but you have no KMS, you have very little integrations, but you prospectually think you're ahead. Or is the company that is working really hard to, to build out even with a one or two person team to build out a full KMS that's taken six months to do that, that.

24:26That's now integrating, integrating their Zoho, to Zendesk with their CCaaS, with an SQL server that has a bunch of sales data, having all that integrated, utilizing all of their CCaaS data to kind of think through issues and problems, that customers have. Who's really ahead in the AI game? It's not even close. So, we got to think this through logically, and not just say we got to go get a tool. to be prepared for the tools to use it to its full capacity I think that that's you know something that's you know really important let me just check my notes here yeah so again I mean just I think just to sum this bad boy up I'm just I think again I've been thinking a lot about this again I've been a lot of BS on LinkedIn where a lot of companies are trying to sling products. There's a lot of white papers out on giant, huge companies that are utilizing AI. And again, remember, the larger companies can afford it. The larger your companies have the volume for the ROI for it as well.

25:32A lot of them are still struggling and we'll struggle with the knowledge and the things that we're still talking about, so use, use 2024 use 2025 as an SMB to get your house in order from an analytic, from a data-standpoint, from a KMS, from an integration standpoint, I would say start to look at your agent tools tools to solve- to solve the questions or the issues that you have first. And then once you feel comfortable, start to move to that agent facing AI, that last, if a really want to do it right, that's. at least what we're trying to kind of inst here at Expivia with our customers to make sure that if they don't have their ducks in a row from a knowledge standpoint, helping them them, but if they do, then let's talk about what problems Again, this is like the forefront of me, right, and I get fired too, that I probably should probably should have ticked down.

26:31But it just gets me fired up when I see like, even AI driven posts, that don't help, help anybody. And they just try to sell. And they're, they're, they're going to make, they're going to make that we're going to make the same mistakes, we made with the IVR, and we made with email, we made with all of. These kind of self service, you know, messiahs that we're going to come and just totally take the agent an industry from a customer experience standpoint, but if we do it wrong and I'm starting to see the same mistakes being made over and over again, we're going to be back to C a CX scores go down, right?

[Gap 27:10–27:40: no reliable transcription]

27:40atviation at Expivia, and then check out our auto QA product as well as our All right, thank you. Appreciate you. Talk to everybody. Next, next, next week.