Building a Smarter AI Enterprise Strategy: Live from NJ SIM

Scott Kinka:

Well, here we are, team and welcome to another episode of The Bridgecast. We’re thrilled to have you with us today with the listeners at home, and for the live studio audience, two we have our studio audience here, our live studio audience for those of you who might be listening on Apple, YouTube, or Spotify, our members are part of an organization called Senior Society Education Management. This is the New Jersey chapter. Can I get a hoop for New Jersey in here? Alright, there we go. For our listeners and those in the room, don’t move me. I’m from the Philadelphia area, so I understand. It’s okay. It’s hard to win. I know. But anyway, that’s good. We had to get that out of the way. Today, what we’re doing is just a project of discussion with a group of some of the smartest people that I’ve met in the last couple of weeks, certainly in this room. Talk a little bit about what is on the mind of IT leaders related to AI. I think largely many cases we can get tech, we understand what we’re trying to do, the challenges that we have as IT leaders is exactly how to deploy, how to meet a line of business reform, how to make the CEO happy, how to return and effectively in our line on the capital that we’re investing in such business. We’re going to spend quite a bit of time on that today. The panelists to the question that we had earlier are some of the smartest minds that we know around AI. Let me see, this accidentally happened, or maybe they already know the slide was going to be. That’s fantastic, and I love the way that worked out. First, he’s been with you before, and some of you should know who he is. But Gary Sorrentino is the Global CIO of Zoom. Gary, tell us a little bit about yourself. 

Gary Sorrentino (Zoom):

My name is Gary Sorino, Global CIO for Zoom. Been in technology 45 years, retired once, waiting for another one to happen, but so great to be in technology now because we are finally with AI and other things, we’re changing the way people work and so I don’t want to miss that in my career and so that’s why I’m still plugging every day. 

Scott Kinka:

Gary was at the Zoom sales kickoff all week, and technically, it’s not even over yet. It was late last night. It’s still going on now. We managed to pull out of that. You took a 1:00 AM flight last night to get here, so I was looking for him to be upstairs to sleep. So thank you, Gary, for the video. Thank you, appreciate that. Next to Gary is Jim Marron, who’s the Original AI Sales Director for Talkdesk. I will skill this under on who he is or what he does. Jim, tell us a little bit. 

Jim Marron (Talkdesk):

Yeah, Jim Marron has been at Talkdesk for six months and has only been in the AI space for four years. Prior to that, I was climbing the corporate ladder at Oracle. Everything I thought I wanted to be and do all the right things, and I left to join a crazy startup AI company. We had tremendous success, and that leads me to this crazy AI world. So I’m happy to talk to everybody here. Great crowd. 

Scott Kinka:

Fantastic. Appreciate it. Thanks Yalim. And then I’m going to get this, and we practiced it, Yalim Eristiren, did I get this? Absolutely. Very close. All. So Yalim is a managing record for SESTEK. They are a Turkish company. He’s running the North American division, and you’re in Boston now, but tell us a little about you. 

Yalim Eristiren (SESTEK):

Yeah, I’ve been in the technology industry for over 25 years now. I’ve worked at Siemens Intel for 12 years and recently joined Ssec two and a half years ago, and I’m trying to drive this adoption on the AI side in the North America leg. Yeah, fantastic you all for being appreciated. 

Scott Kinka:

So what I like to do is just start with a little bit of table setting to show that we have a contest by which to set the conversation. I recognize that at the top of that slide says 2024. You recognize that the analysts don’t put up the 2025 data until about halfway through the meeting. So let’s just call this pertinent. This is from Infotech, who is a research partner of ours. Top five priorities for CIOs. This is the survey, not their thought process. And I just want to step through and augment the business with general AI. I just realized that there’s general AI, the bottoms, but I’m not going to play or is AI accelerated speech of recognition from right-sized AI governance update, vendor risk assessments, exponentially increase innovation or exponentially improved customer service. Now you may or may not argue with those, but just to highlight here, at least three, potentially four of, these top five are all largely bound up in the same topic. The two are clearly AI. Most of the customer experience efforts we see right now that are led by CIOs are about how to make the context more efficient with AI, with software, etc. So we’re tying those together. So with all that said, first Jim, I’m going to come to you first. You talk to CEOs all day, right? Wrong in different ages out. What are you doing? Does this make sense to you? 

The Reality of AI Hype vs. Practical Use

Jim Marron (Talkdesk):

Yeah, I don’t want to steal everybody’s thunder with agent AI. That’s the hottest topic right now, but I think for the most part they hit the nail on the head. The one thing that’s not on here that wouldn’t be necessarily a topic, it’s just like who do you trust when it comes to AI? I go to so many different conferences, we all do, and you guys all do too, and we have so many different conversations. What happens is there’s a line of business that gets basically assigned an assignment to go after. You have to have a certain percentage of AI in your line of business or automation, whatever the case may be. And when they say automation, it’s the hottest technology. What do you have? Gen AI, agenda AI, whatever the case may be. These folks and these people don’t know how to solve their business problems with AI. So what happens is their guard goes up, they lose trust, and everybody’s saying the same thing. Agent AI, gen AI, we do this, we do that. Very similar talk track, especially if you don’t know the business of AI. So my approach at Talkdesk, and for the record, at Talkdesk, sorry, we’re a CAS platform. I’m on the conversational AI side of the CAS. So think of a contact center. Think of everything we do as enhancing the customer experience. So we have underpinnings AI, we have our own N-L-P-N-L-U that’s driven with our own technology. We bring in a third-party LLM, and we have out-of-the-box solutions to make conversations really smart, really seamless, really efficient. And the whole end goal is we can measure ROI, ROI is such an overused word too, but in the contact center, when it comes to AI space, you can actually measure it.  You can know exactly what your cost per conversation is, and with our technology, we can drive that down between 50 and 80%. That’s my business, that’s what I do. I’m in the ins and outs that every day, what happens is these folks in the contact center, for this example in customer experience, they don’t understand AI like a CIO would. So it’s our job to unify everybody within the company. If we have an evaluation, you have CIOs, you have marketing, you have obviously customer experience, different lines of business that are curious about different things. My job is to unify ’em, explain to ’em and be a trusted consultant on my talk desk to achieve their goals. 

Scott Kinka:

We’ll get a little bit more into one of those parts of the conversation that I found particularly interesting in our prep call, which is around RI, who’s responsible for that right now, AI’s kind of being plus on it, but in many cases without use cases. So who’s going to drive dollars? We’ll come back to that in one minute. I just want to ask, lemme come to you. You mentioned on our pre-call we’re talking about trends, and I’m just going to get right to it. Let’s get at an agent, but while you’re answering the agent trend right now, what is it that you’re saying, I’m going to ask you a car, right? Do we even know what to do with generative AI and we’re already onto the next topic? Just put that in context.

Yalim Eristiren (SESTEK):

Yeah, I think the pace of innovation in the AI space is pretty high. That’s the reason why we are having a lot of confusion. Sometimes we think that we understand the concepts, but sometimes we don’t, and we learn from each other as well. But if you take a look back to the conversational AI space, up until 2013, there was no AI, almost no AI in the conversational space, the introduction of the Bert space of transformer models getting into the NLU space, we started having those 2013, 14, that’s where we picked up the technology and started with our own NLU. And life was good up until the generative AI came into place. So now, how are we going to leverage generative AI? So we started thinking about it and we started augmenting our solutions with generative AI capabilities to make it, as Jim mentioned, smoother, more natural, more conversational, and the experience and the end goal over there is improving the customer experience actually. So at the end of the day, for any company, if you don’t have a good customer experience, you’re not going to be in this business for so long. You’re going to go out of this, the competition will beat you and get your customers, your churn will go higher, and your retention will be going lower, and you’re going to lose your business. So the customer experience is key to keep those customers thatare  on your side, and these tools, I call them tools, we are just developing tools for the companies, for enterprises to keep that customer experience at a high level. But the challenge and expectations are high. That’s the reason why these technologies are extremely critical to getting the customer experience at a very high level. So generative AI was very important song like three years ago, maybe three and a half years ago, first introduced with the chat PTN, we started playing with that open AI, Microsoft Azure, and the others came in then Amazon, Google, we started playing with those. We started talking about large language models. Then we said maybe not every time large language models are right, let’s talk about small language models with fewer parameters. So maybe not on the cloud, but maybe on-prem. We need a financial institution. We cannot let our data be exposed to the public, so let’s keep it on-prem. So all these conversations we started having, all these discussions we started having and all of a sudden we said, okay, we have a lot of tools, but how are we going to be able to deal with this? Then, the agentic AI came into the picture. So agentic AI, what presents is first of all automation. One of the speakers, I think Gary Jim, mentioned that automation is key. So at the end of the day, if you don’t automate your processes, you are not going to get efficiency and productivity out of that. So you have to, that’s an endless pursuit, I would say, to achieve higher levels of productivity and efficiency. So that is one part. And then with that, that’s auto automation. The second thing is, we are getting into the middle of the process so much. So how are we going to make these systems work in an autonomous way? Are they going to be able to work autonomously? So those AI applications are actually the AI AI agents, absolutely.

Question from the Audience:

So when you say autonomously, which hangs on? Okay, alright. So when you say autonomously, okay, isn’t that predictive AI? Okay, because one of my concerns is you take the human out of the decision-making, and that’s typically predictive AI. So that’s really where you benefit from the automation. 

Yalim Eristiren (SESTEK):

Not necessarily, not necessarily predictive. So it’s all about the goal that you’re setting the goal orientation with the agent. We put at least while defining the agent AI, what we are saying is automation, first of all, it should automate it. The second one is autonomous, and the third one is goal orientation. That goal orientation doesn’t need to be predictive. So you can set the instructions for the genetic AI and it doesn’t need to be predictive, it doesn’t need to foresee what’s going to happen, but 

Question from the Audience:

Generative creates new predictions, makes a decision. So if you’re predictive, you are making a decision for a human. So that’s what my concerns are. So when we’re in an air school, we’re using predictive AI and the reason we’re doing that is we want to make decisions and increase the efficiency of the process. So a human doesn’t have to, I won’t say interfere, we take the human out of, okay, so the predictive one, that’s the one that I see most of the efficiency and productivity. 

Cloud-era Mistakes Repeating in AI Deployment

Jim Marron (Talkdesk):

Can I chime in there real quick, please? Yeah. So yes, the answer to that is yes, but there are two components to that historical context. So, I’ll just give you an example of the system. If you have an AI platform and you have a billion even-numbered conversations, it learns from those conversations. So it learns from historical context. That’s the first component of that. The second component of that is that you have workflows that are guided, although it’s predictive, it understands that you have prebuilt workflows that are predetermined to get you to what you were saying your end goal. So that is where the calculation of the risk mitigation comes into that, in my opinion. So I think your concern is okay. You take the human out of it, and this thing is just going haywire. All I had an incident on, you have something to say, but let me finish my thought real quick on a plane right over here. I use GPT all the time, but now I fact check it all the time because it’s wrong a lot, right? And so that’s a gen AI model. But where I’m going with this is you can still predefine your workflows. That’s where the proprietary nature of what we do and probably what you do too comes into it, and that’s what the trusted partnership comes into it. What’s your goal as a company? It’s predefined. So although it’s oriented and it’s predictive, it’s still guided in terms of where we’re looking to get, in terms of the goal is predefined. 

Question from the Audience

So the top two over there are generative, right? So the CIOs didn’t ask for predictive. So my thought is that predictive is really where your efficiency and productivity are potentially certain, but they’re not there yet. Those five top priorities.

Scott Kinka:

Well, I think we’re painting, and that was kind of my point earlier about whether we really even know what generation is at this point? We’re asking, this is the technology for a conversation that we constantly find ourselves in. Lemme go get into engine data. What are our use cases in generating right now? Come with what we’re doing on predictive. It’s a great, great microcosm of, I think, the broader challenge. This is just a swath of thousands of CIOs going, and this is what I have to do. Generative is weird because that’s the word that shows up as much in their fee as any other methodology. We’re being honest about it. And that was really kind of where we were going with the next chunk of, I think you kind of gave a little bit of a good segue if you will. So you’re another set of data from CIO magazine, and I’m going to come to Gary because it’s got some great thoughts on this. This is from IGCI magazine end-to-end that. And this is a little bit about what’s happening out there, and this is what pops off the pagn. 88% of CIOs are starting AI projects. Here’s where the maps are backwards. 50% are getting tools in users’ hands, but 42% are still identifying use cases. The entire thing is backwards, effectively, at the end of the day. So Gary, you and I have had this conversation a bunch of times, and I think this is a great parallel to cloud eight, 10 years ago, corporate levelICc, CEO calls me, Hey, my IT friend, go get some. I got to tell the board, I got to tell the investors, I got to have something out. We had to have some cloud eight years ago. So the question here for Gary i,s are we, aren’t we doing this to repeat at the end of the day tech for tech’s sake here. And I think the advice portion for everybody in this room is how you control that narrative so it doesn’t get out from underneath you as an IT leader. 

Gary Sorrentino (Zoom):

Well, let’s unpack that a little bit. Yeah, a couple things. So first, just to go back to the last slide, I reiterate that that data was from the beginning of 2024. We didn’t use the word agent AI, otherwise it would’ve been on here, right? Anytime you read a statistic that says 2024, it takes them 18 months to get that ready, right? So you’ve got to look at a timeframe there. But to what you said, yes, in some cases I just presented to a board and they said, look, we told our CIO to go buy AI, lots of it. I said, does he buy it by the court? The pound foot? He goes, What do you mean? I go, it’s a technology. It’s like HTML. Did you tell him to go by Coldwell? He goes, No, I don’t know what that means. You really don’t know what AI means either. And AI is not a solution, it is a technology. And I’m really surprised that that last page didn’t say employee experience because we suddenly forgot about the humans that we’re trying to service, right? Yes. We all talk about customer experience, which is sellable. Okay, you can measure that, but really, it should be about employee experience. And that’s why a lot of the math doesn’t work here. They are going out. Remember, we gave people Excel with the idea that they were going to do what with it? They were going to figure it out, and 30 years later, they’re using 3% of that tool. Well done, CIOs, well done. Okay? That is a total waste of money, okay? Because one, we gave it to ’em and we didn’t teach ’em how to use it. You know why? I think you said it. What was the business case that we gave him in Excel. There was none. We went out and bought it. It was part of a productivity package. And I think that’s what we’re going to see right now. Now we need to get in front of this, we need to turn it around. This is the time when it has to align with the business. Now I didn’t say my background. My background prior to that was I worked for JP Morgan, I worked for Citigroup, and I worked for four of the largest banks in the world for 25 years. We think differently. And I hope everybody starts; we think about the end result before we put something out to the users. I hope their percentage, and I hope my users back then, were greater than 3% in Excel. But I think this is really where it has to almost admit how much you really understand around the business. And I think we have to go inside ourselves and say, am I ready to support this business in a profitable ROI with this new technology called AI? Because I understand what skip generational wealth is. I understand what scorpions do. Whatever your business is, could you run the business or could you run it right? And I came from an environment where the guy you all know who ran that last bank of mine, JP Morgan, he made it very clear, if I walk into a room and I can identify business people and tech people, you all fail. You fire tech people, every acronym you use, because you want to be smart, you’ve got to make sure you teach the IT people, business people, business people. You need to teach the IT people all those words you use. So I think until we get this straight, we are almost doomed to put a technology out there that costs a lot of money. Second, I’m also a CIO. If you say you fully understand it, I’d love to see you present it to me. Alright? I am on both sides. Okay, so I was a CIO as a buyer and now I’m a CIO as a seller. And I clearly know that as a CIO for a buyer, I would probably understand this much of this technology. I only understand more of it because now I have to explain it to my peers. So I had to learn more of it. So I really do think we need to change the model. That’s what we talked about. We need to change the model here a little bit. We need to ask the right questions. I always leave this with CIOs when I speak. Tell me what you don’t know, and tell me what you don’t know. Just ask your team what we don’t know? And I would love to hear that answer, and tell me when you have enough of that answer, that is a hundred percent. And so I think that’s the failure that we’re going to see unless we rightsize the problem and we get a yes, we’re still going to make the investments. Look, we sell AI too, right? We give it away in one product, we include it in another product, we’re actually giving most of it away now. Hey, easy, sorry, but we’re giving it away across the platform for version one. You know why? Because we knew you weren’t going to know what to do with it. So isn’t it better to figure it out on zero dime than it is to figure it out on x dime, whatever it is. So that was our theory. If we put it in the hands of people, they’ll be, they’ll get to all the compliance stuff. I know we’re going to have that conversation, but at least if it’s in your hands for free at no cost, you’ll experiment with it. Make people curious. I guarantee that if you can put it in the hands of your users, they’ll think of better business cases than you will. 

Changing the IT-Business Relationship 

Question from the Audience:

We actually have a question. So you’re basically telling me nothing’s changed in 20 years. No, I’m telling you they know they want something. They’re telling us for the last 20, 25 years we’ve understood the process, optimized the process, and then found the right technology to automate. 

Gary Sorrentino (Zoom):

This is exactly in technology. I’m going back to you, said, so this is the cart leading the horse. See, you were told what technology to use. You made a decision whether it was FORTRAN or COBOL or whatever platform you used, but at the end of the day, you were told the solution before you were told the problem. And that’s what you’ve got to get in front of a CIO, right? You got to almost. I’m going to say bad words, so slow the train down a little, right? You’ve got to slow the train down a little. I don’t know. Everybody’s measuring success. I’m like success, you didn’t even know the problem, right? But you already got success figures up there about it. So I think that’s where we got to as CIOs; we owe it to our businesses to understand this better and to sit back and actually internalize how much of this we really know. 

Scott Kinka:

Let me add one thing that we kind of came up in a pre-call that was interesting, and we talk to CIOs all the time, and the role we talk about the last 20 years, typically the line of business hates solving problems. And ultimately, which funny, thank you for the awesome segue, I’m going to get shit for this one half hours afterwards. Ultimately, okay, the business that was using the technology that they asked you to deploy was ultimately responsible for the ROI on the investment that you made. You just had to compute lights on, you deployed, well, you needed to change management around it, but at the end of the day, they were using it or they weren’t. Now something happened by the way, cloud, obviously it’s all about cloud. We ran that way. But something happened between that one and this one, and it was COVID, and I’ve been to the CEO’s all the time and we were like, man, I worked with the CO before COVID.  I reported to the CFO before COVID, but there was a day during COVID where the CEO grabbed a CO and said, What do I do? Do? And now that’s happening again because we all already have that practice. But the problem with this one is that it’s just technology that cannot answer. So the line of business now has almost no responsibility for how they’re going to use AI. How do you drive? I ultimately, the question, I’m going to come back to these guys then, and anybody can jump on this, but let’s talk a little bit about that. Say what you just said.

Gary Sorrentino (Zoom):

Yeah, it’s tiered SMBs, that’s the case for sure, right? Go up the ladder, law firms, that’s not the case, right? Healthcare and healthcare’s not the case. The business is driving it with technology partners. As you go up the line, when you get to the financial institutions, when you get to those Fortune 100 and things like that, they have a different model on this. The business is responsible for ROI, and technology is just keeping the lights on when it comes to AI, they’re buying it and keeping it on. They’re worried about security, and they’re worried about compliance, but the business is responsible. So this is a multi-part question. Depending on the segment and the size of the company, that has really changed during the pandemic, the smaller companies, kind of the, if you’re a CI for an S mb, you became God, right? You know everything. But as you go up the line, the partnerships between CIOs and business, the business is now responsible for this. So this is not one answer. 

Scott Kinka:

No, certainly not. Let’s work out for fun. Let’s have fun. I saw chuckles when I made that. You got asked to deploy AI and control how you were going to use it. Maybe talk about your own business. But if you feel like that that’s true, you or peer of yours is being asked to play AI in your business right now without necessarily being told how to use it, you raise your hand. Are you okay with that right now? A couple of you certainly maybe not necessarily willing to address it, but I get there’s a couple of you here. We definitely hear that and I agree with you. Does anybody else, you guys have an opinion on this, how in the products that you offer or are you establishing this? Let’s morph it a little bit. Give us a use case where you’ve got the monster for AI that RI that you can push in as an answer to the question of go get me some AI.

Jim Marron (Talkdesk:

I can talk on make somebody a hero. If you work for a big company, and again, I’ll talk about contact center. Where I live right now is that find somebody on the low end totem pole. And this is not a knock on anybody in a head of contact center. And this is a true story and I think we’re going to get the customer story, but I go ahead and say it anyways. It’s just that somebody that is bringing a new disruptive technology into the organization and every organization is very different. There’s no one size fits all to this approach is going to work for this company. And so we’re going to do it across the board. You have to identify a champion and an advocate for this AI who’s going to be talking positively about your product, about your platform, about what you’re going to do for the company for X, Y, and Z reasons. Going back to the ROI, you have to define that. You have to define a success criteria. These are measurable outcomes with AI in the contact center and reducing call volume, understanding use cases, understanding how to offset some of these things. Very measurable versus what I used to do in the ERP space where it’s like, hey, rip and replace your QuickBooks. SAP with Oracle or NetSuite, great six-month project, tons of money. You guys understand, I’m sure half the room if not everybody in the room has done that before. This is so measurable. So you have to find somebody who cares enough to be speaking positively and selling this thing for you behind the scenes. And Scott, you asked a question before, how many of you, what was the question you just asked again a minute ago? 

Scott Kinka:

Every begin deliver AI without any context on how to use.

Jim Marron (Talkdesk):

My question is how many of you feel pressured to have AI? Because if you don’t have AI, you are falling behind. And if you all don’t raise your hands, you’re lying to yourself because this is a God’s truth, right? You have to have some sort of AI to keep up with your competition with your peers. I don’t care if you like it, if you don’t like it, I have my stance on it. I won’t say it on the podcast, but with that being said, it’s going in one direction. If you don’t follow that direction, you’ll be left behind and that’s black and white. So that’s what I’ll say with that. You have to identify that early on the conversation and somebody who truly cares as an advocate. And that’s how you not just win business. I’m in sales. Yeah, I’m trying to win business. My success is dependent on your success. If you’re not successful as a company, I’m not successful and our business is not successful. I can’t say that enough. So it’s identifying that early identifying use cases, success criteria and plan of action. 

Yalim Eristiren (SESTEK):

Anything. Yeah, just one thing. I mean maybe this is my personal opinion, but over the past 25, 30 years trying to sell to enterprise different technologies, different technologies stacks, I have never seen line of businesses taking this much role in the decision making. So I’ve been always talking to CTOs and their teams. I’m always talking to CIOs and their offices, data center projects or CRM projects. We here now they have taken the lead to have this conversation with the vendors. I am pretty sure you’re experiencing this too. So I think what they need right now is because they don’t know the technology. So once we have those conversations, the first few sessions we go through the education piece again, this is that, this is this, this is that, this is how that works. Do you guys already know that? You have teams that are already, so what they need is actually a thought partner. So that’s what I’m seeing in this. I’m sorry, this is what I’m, did I make my, sorry, this is headphones. So yeah, that is what I’m seeing over the past. This is a change because AI is so sexy to everyone now. Everybody wants to learn it and it’s at fingertips. So they can play with churchy media and they say they know about it and they start talking about it. They hear a lot on their LinkedIn fees, on their Twitter fees or whatever, social media. They read a lot and they watch a lot and that’s the reason why they think that they can do it. Yes, they’re driving the business value. I a hundred percent agree with Gary about the business value, but at the same time, they need you to make a well-educated decision.

Employee Experience, Productivity, and Accountability

Gary Sorrentino (Zoom):

I would say. But just add to one thing is that it’s also part of their business goals now for the first time. So now when they’re being rated as heads of groups, heads of departments, how to keep headcount and how to increase customer experience, how to make the employees not quit and that’s now part of their goal. So as part of their goal, the business leaders have no choice to be part of this measured on something. And I agree with you said they need to lean on the trusted advisor. That’s the CIO type, right? You’re a trusted advisor because while your bonus is based on that, so is theirs. And so for the first time you get a partner who’s in the same room with you because for the first time you’re most motivated by the same challenge and the same incentive. And I think that’s changed a little bit. They weren’t measured on whether you got ’em the latest pc. They weren’t measured on whether you upgraded DERP that wasn’t on their goal. How do you use AI? How do you improve customer experience? How do you improve employee experience is on their goals, but it’s on ours also. First time I think we’re sharing it. 

Question from the Audience:

So when you say it’s on their agenda and your agenda, it’s not use AI. It’s basically how you reduce, how do you increase productivity? So that’s how you question.

Gary Sorrentino (Zoom):

No, no. So what he said was it’s not about using AI, it’s how you increase productivity. That is exactly what it should be. Agree. Okay. Unfortunately it says how have you deployed AI and what percentage of AI have you deployed in your department? And if it doesn’t say a hundred percent, you’re not getting your bonus. What you said is perfect. It should be how do you measure business value, right? And it should be increasing business value by X, increasing employee value by X. That should be the goals. 

Question from the Audience:

So the people who are in those positions, the people in those positions don’t deserve to be in those positions because it should be how you increasing productivity. 

Gary Sorrentino (Zoom):

I’m on your side of defense. Come up with us. 

Scott Kinka:

I think lemme throw something on top is going to be a general question and I would love for you guys to respond. It’s not in an, but it does bring up an interesting point about the measure of productivity inside of the business, which is we’ve lived since the days of factory shifts in this mode of I have an eight hour day and the person who works 12 is the person who’s going to get ahead regardless of how many individual they actually produce. We now have this challenge goes over into HR in businesses that are AI iterate. What happens to a developer in your group who gets two days worth of code done in two and a half hours and the other person’s chopping wood all night long? Can they take the rest of the day off or are we expecting a full day to get 10 days worth of work done? There’s a whole paradigm challenge in there to start to thinking about when we’re thinking about productivity that way outside the real. Is that anything to add onto that? We could be on this one forever so we probably put after this, but comment, 

Gary Sorrentino (Zoom):

I think you’re correct. I think have to change. We have to change the model. We’re all learning. That’s the best part about this. We’re all learning and learn with the business, right? We’ve invented titles, right? Who has A-C-D-I-O in their company, right? Yeah. Responsible for business transfer. It is probably an ex CIO who figured out he has to get from the expense side of the house to the income side of the house, the resident side. That’s the goal today is we’re all learning. And so the business is now hiring someone dedicated to making their bonus or their numbers better and they’re hiring a trusted advisor who was probably an XCIO to move over and do those things. And so we’re seeing the model change and which I agree, we’re seeing the model change on how we deploy and we’re seeing the model change on how we measure. 

Scott Kinka:

I think this is a pretty big part. I just want to throw this one out there. This is why I was up here earlier, but I didn’t point on the training upskilling comment, a 51% saying they’re training and upskilling employees on gen AI. Let’s talk about this AI literacy conversation as a Carlo to this statement about productivity, right? Because I think when you get to the employee experience guys thing you were talking about it’s incumbent on the business whether HR is driving or whether it is driving it to make sure that one, you get a good measure of productivity, but two, we don’t allow alone those who are doing work at common figure out how to write a good prompts to succeed beyond others who maybe could with some training like is AI literacy just a reasonable goal for this? Forget all these high products where we’re going to drive all these costs have business, we have a lot of work to do to be able to get the business aligned. But if there’s one thing to do today, is it just to start building a literacy program in pockets of the business? Get some Gen I in their hands, maybe do something around prompt writing. Do you think that’s a reasonable project for IT Industry? They that? 

Yalim Eristiren (SESTEK):

Yeah, I mean absolutely. That’s a starting point. AI literacy is number one. So you need to as a corporation, not only the it, but overall since the legal piece is compliance piece, everybody needs to get educated on the AI side. So I was talking to a friend of mine, he has a startup in Canada is working on on-prem GRC solution, not a cloud. So what are the challenges that, I asked him the question, what are the challenges he has? He’s seeing, the first thing is that if you take a look at the data governance, that’s 70%, he says 70% on the other departments and 20% on the technology side in terms of effort that’s put in to make it work. Whereas in the AI act after the EU AI act, so that’s the opposite, like the 70% on the technology side, 30% on the other stuff, legal stuff and stuff. So there’s a lot of training to be done, the governance side as well, but as well as identifying the right project where to add value. We talked about productivity, efficiency, employee experience, customer experience ly improve, where to improve, you’re going to start from somewhere. And in order to identify where to add value, you have to have educated people to identify those things. So otherwise you’re going to just go with the flow and that’s not going to get you where you want it to.

Gary Sorrentino (Zoom):

I just think could change the way you train. You can’t go back to those 45 minute 10 question cyber trainings that we’re all used to, right? Micro learning, micro certification, Right? That’s where I see AI taking good role as you’re training people. What’d they say? When AI first came out at the beginning of the pandemic, there was a quote that said A car goes a thousand miles an hour. We gave it to a 16-year-old with no license and no instruction book. And now that’s what’s coming back here now is that, but I do think we have to use AI to train an AI and it’s all about micro learning. And remember most of us are writing code right now for Gen Zs, which we probably don’t understand. They’re our kids. We don’t understand ’em at home. We don’t understand ’em at work. But that’s why we have to change that. And I’m glad that that’s up there because I think all of this has to come together, but it has to be in a form that people can consume it and it’s not the old fashioned one. I want to add 

Scott Kinka:

On that quote though. I think they probably were like there was a large of a thousand miles an hour with no structures that was unlocked that your 16-year-old clearly how to drive that on our own. 

Gary Sorrentino (Zoom):

No, we gave it to them. Yeah, we at the beginning and expected an outcome. 

Use Case: ROI in Six Months at Contact Center

Scott Kinka:

Clearly. Clearly yes, it’s going to happen to the businesses whether we’re driving, right, not getting into the day. Well, there was a request in our pre pre-show around some use case. I’m just going to skip this slide, slide. Forget this slide. I got to up here for some fun. These guys are talking to customers all day day and we’ve been talking about ROI, decisionmaking, who owns it. What we were driving has some requests around pharmaceutical life sciences have requests around small mid market businesses. Each of you just has an opportunity. Tell us a customer story. And in the vein of the question we’ve already had, who owned it? Where did it get driven from? What was the outcome? Was it demonstrable from an RRI perspective? Tell us something we can sink our teeth into. Didn’t be super long, but Situationalized formula, I’m going to come to you first.

Yalim Eristiren (SESTEK):

To you first. Absolutely. I mean I love that because we had it just finished it. They completed the project and it’s live now. It’s a conversational AI project with PHP an insurance company and it was very successful. It was C went fast because CTO ran it. So CTO had the business view, had the business partners in the team, and from start to the end, he was in every single call and he was driving it. He was leading the teams and we didn’t have any problem in terms of backend integrations. We have done through all these, I would say complex problems we had and having CTO there, understanding the problems, understanding the business value, what to create, eased our lives big time. So definitely that was not an easy project. Multilingual, multimodal, multichannel. So a lot of things together. And generative AI, there was a RAG implementation as well. So generat AI was involved, included in that. That was a complex project and with the CTO involved in it driving it was super fast and super successful. 

Scott Kinka:

So two Things basically. Questions on that. It did what to deliver what and what was the expected RI in the business? 

Yalim Eristiren (SESTEK):

Yeah, RO. I just completed this so early to tell about the ROI piece, but what it did is, so first of all, they have 25,000 agents in the field that they need to assist. So the agents can access the knowledge base at any time they want and get to extract the information they want or prompt some queries about their business results and everything. All these APIs results. And it should be in two different languages, Spanish and English. It should be available on web chat and also through mobile app. And it should be also via text and voice. So all these multimodal, multichannel, multi multi channel multimodal. Yep. Infrastructure, it should be available and we all deliver that. And that’s I will say, I’m just back to my first point, having the leadership from the technical side paid off at the end. 

Scott Kinka:

Okay, Jim, tell us one. 

Use Case: Healthcare Application for Detecting Adverse Effects

Jim Marron (Talkdesk):

Can I bring up my notes? I’m just kidding. No, my numbers. So here’s the deal. 208% ROI in six months, that’s what I’ll say. And I’ll start by this, somebody over in that corner, I think it was you mentioned streamlining operations. Well, there’s not a single person, this room or a single company that we work for that is not measured on reducing costs. So I look at everything from how can we reduce costs. So in the context center, just in terms of volume, I look at over there, the health and life sciences use cases, any kind of scenario where there’s high volume and high touch points. If you can offset that with self-service, essentially AI, self-service, some will real issue and lower the average handling time on those calls, lower the amount of calls coming in. Let’s say your talk desk and you’re picking up that voice, the voice box picking up that call, right? It’s owning that for a certain dollar amount. We take a look at what they’re paying per dollar, and I’ll just give you, these are not real numbers, so don’t hold me to it, but they’re close numbers. Let’s say a hundred thousand calls per month and their containment rate was very low. By containment rate, I mean calls that are going to agents so they’re not being contained in an IVR or via self-service. You can measure that, you can offset that and you can find that dollar amount and reduce that based on the low hanging fruit use cases, booking an appointment, billing, asking a question about your account. You do data dips, we have plug and plays, we have prebuilt integrations, APIs, all this other stuff that everybody else has that you dive into that information, you extract it. And in real time, gen AI or gen AI, it doesn’t matter both in some cases you are having real time conversations, whether it be via chat, a voice, SMS, WhatsApp and agents are not needed. And that’s not a knock on agents. It’s not saying, Hey, we’re going to reduce agents. You brought up a really great point, and I looked at that slide Scott sent over probably five times on the plane between my broken wifi and the kids screaming behind me was that employee experience was not mentioned on that slide. And that’s such a lost art because if you have a model where you want to deliver AI and you, okay, you have everything built out perfectly, Hey, we have this all is ROI, it’s great, except for we’re going to lose all of our employees because you don’t matter. That completely eliminates the entire issue because then you have a business. So to Gary’s point is how do you let AI work with the humans and find ROI? And to somebody’s point before, it’s finding those use cases that get you there and there’s a lot of low hanging fruit. I don’t know how many folks in the room have AI implemented right now specifically for these use cases, find those use cases, start saving costs and drive operational efficiency. There’s a lot of examples like that. And to your point before Scott, 208% ROI in six months, pretty good I’d say. 

Scott Kinka:

Yeah. Yeah. And they base that on, what was that improvement in handled time hires they didn’t make. I think we can immediately go to this, can we let go of people? But I mean we’re really thinking more along empowering their productivity. What did they measure? The ROL? 

Jim Marron (Talkdesk):

It was actually they were sending calls to A BPO. So this is a utilities, 

Scott Kinka:

So straight up minutes of use that were going outside to answer phone calls. Were not going outside. 

Use Case: Email Classification and Agent Workload Reduction

Jim Marron (Talkdesk):

So the industry average, you pay a contact center rep if you offshore, let’s say $15, this is an average onshore, 40 bucks. You can do simple math. Average handle time on a call in enterprise businesses is between seven and 10 minutes. You can do simple math, even if you’re off a couple minutes, you will find ROI how prescriptive you want to get with it. You drive down those costs. That’s your entry level AI. And at least in the contact center or anybody who’s fielding calls and answering customer inquiries, everybody in this room has had terrible experiences calling either their cell phone company, their bank insurance. I was talking to somebody in the room before, so that’s your at least entry level in terms of what I do. And then it’s measured based on how quickly can we deploy, 

Scott Kinka:

Which is Gary a customer story 

Gary Sorrentino (Zoom):

Should we covered ROI? But I think we talked about efficiency and I think employee efficiency is probably over missed and not that we can give you another ROI story, but really about think about this. How are we making people work more? We used to say smarter, not harder. Now we’re saying intelligent, not smarter, right? How do we bring them up to scale? Think about this. We have people who are now looking at meetings and going, how many people have been in a meeting? There’s 12 people in the meeting. This is a round number and nine don’t talk. Okay, okay. How inefficient is that? There’s no ROI, but it’s inefficient. How efficient is it that the three that do talk have the conversation and we send the nine the notes and they can read it later, right? Because they’re only going to go and take notes themselves. And we know that AI takes better notes, unbiased, right? So at the end of the day, you have to look at it for more than just ROI. Now you can turn that into ROI later. Maybe we don’t have to hire as many PMs. Maybe down the road we can become a little bit more fine tuned, but we’re not really, I don’t think the first step of AI is to eliminate humans. I think the first step of AI is how do we make the existing humans more efficient in what they do and how do we get a better outcome from our clients? And I think so this whole thing about note taking, what they’re doing with meetings, the captions that you saw at the end here, the multilingual things like that. I think they are really adding efficiency. So while we say ROI, we got to put right next to what the word efficiency and the efficiency is an inward score. And the ROI is usually an external score, but you have to balance both. 

Scott Kinka:

You have a question here, but I got a microphone. So you hang on 

Question from the Audience:

Talking about case studies. I’ve got an awesome one. I work with a large healthcare organization. They’ve got probably 500 locations. There is money in insurance. I know we as users of the insurance don’t think there is, but there is if you’re doing transplants or you’re working with cancer. So think about the large language models that these companies are able to build because they’ve been recording our voices and every time we call in for the past 10 years. So they can take action on that. What are they doing when it comes to where does the rubber meet the road? One of the organizations they work with, they’re building center. They did not want to go with the big call center because there’s 800 plus agents there. They want a more intimate type experience for the patients because it’s real time. So five of us in here unfortunately are stuck with cancer and we all have different doctors, but we end up at the same building, same call center for appointments. I call in on a Monday, I’ve received my chemo on Thursday before I have a cough or I’ve got a fever, I’ve got a rash. Someone else in this room, same thing they call on Tuesday with the same problem. And they had received the same chemo on same Thursday, two other people, same exact issue. This would’ve all been thrown in a folder that never would’ve been looked at when you go to your doctor’s appointment, oh, I see a fever. Okay, cool. No, at the end of the week, the people, the managers, the doctors, the nurse practitioners can run these reports. Hey, five of the people for the people that took this drug on last Thursday now have a fever or rash or whatever. You can take action with items. Is there ROI on that? Maybe a life or two. So think about what these organizations are doing. They’re using that data properly now to save lives or to point out, hey, there’s a problem with the drug. Was the drug administered correctly? Was the drug refrigerated properly? It all ties in together. So that’s a great use case. Great use case. 

Question from the Audience:

So what you’re talking about is adverse effect, the pharmac. So what is happening is usually the traditional way to do that is adverse respect happens. Doctors to the pharmaceuticals to see this is what is happening. So they collect all the feedback. What can happen now with the AI is the decision making decision. So the people who were suffering from adverse effect, they are able to not stop the medication earlier, have multiples used to go, but now they can make decisions quickly of that stop any reaction. So that’s use case, not look at, but now you see 

Question from the Audience:

Didn’t have information. Absolutely. Let’s pull that drug. Let’s pull that administrator or we’ll put that the refrigerator. Exactly. 

Question from the Audience:

Imagine what, 20 years back. I won’t say the full name. It started with V. Yeah, it was there. Imagine how you would like. Yeah. Thank you. Do you have a question over here? I want someone assume Here, thank you. 

Scott Kinka:

Not so much a question, but kind of to play off of what he generally here talking about. Certainly cancer research is its own beast and its own. But most organizations have what? Decades worth of data, right? 1980s, nineties, two thousands, so on and so forth. This data, 80% of the company’s data lives in documents, PDFs, Excel files, word documents, you name it. And it lives in the cloud. It lives in people’s computers, in handwritten notes. What AI can do beyond simply just transcribing and understanding and helping digest what these gentlemen are talking about, it can make this data accessible. AI has the capability through intelligent processing capabilities to access decades worth of data. Now imagine that data at your fingertips and all the decisions that can made combine that with agents that we’re talking about and you combine it with process automation that we’re talking about. And to give you an example of a use case, something that my company is actually pointing today with an IT MSP, they’re receiving 80,000 emails per month to one central supporting box. That’s a lot for a human manage, okay? 80,000 emails. What we are doing today is, or what we have done I should say is we have implemented an email classification system to intercept, transcribe and summarize these emails for this organization so that they can see all 80,000 emails in one dashboard with all their information they can make their decisions. And then we combine it with agents to take action on those emails and complete those actions. So what we’re talking about ROI and results, and we’re talking about cx, we’re talking about customer experience. What we did is we took this company from reviewing these emails 12 hours per day to 12 minutes. 12 hours to 12 minutes. So that right there should illustrate the power of the customer experience. Excuse me, employee experience. It makes their lives easier, it helps make the company more efficient, it makes everybody look better. And to their point as well, we had a project champion within the organization that helped us elevate it through the CEO to get this project done. So just wanted to play off the point with you guys.

Foundation Before Innovation

Scott Kinka:

Guys. That’s great use case. Thank you. Thank you very much. So with just three or four minutes that we have left, I knew we weren’t going to get to everything and I didn’t think it was enough slides, but thank you for your participation and that’s what kept this conversation breached. I just want to throw this out of here and I’ll read these out guys. So you get it from the state report. This is fairly recent. What worries it? Decision makers about gen AI and the workplace. Legal loss of control going rogue employees, lack of confidence in the results. That’s number one. Second group, close to a quarter of it, decision makers are concerned that the organization might be moving too fast. Gary, your point, Take a pause, figure out what we’re trying to do. This gets to your point about the data that’s not in a spot where it’s usable or here’s the other problem that may contain bad data processes that somebody followed to get something done efficiently that weren’t endorsed by the company. So there are only 34% of IT decision makers right now here. Their organization has the right data and technology around that data to be able to support a broad based AI strategy on that. So without unpacking all this, because we only have a of minutes, and I had a few questions there. Let’s just say this. I mean, this could be, there are concerns here about data hygiene, data protection, undocumented processes that we’re trying to get AI to document. Are they right? There’s been plenty of use cases out there about that. Governance, security, all of those things. So without getting into all that, I’m just going to make this simple and we’ll round robin quick. If you happen to be someone in this room who finds themselves thinking about, Hey, you have the use case, Hey, you can demonstrate the RO, I might just not have the foundation on which to build this handled yet, right? I’m going to ask each of these guys, just give us one piece of advice like where to start because it’s coming at you, right? Particularly in this middle one, the employees are going there anyway, even if you don’t have a strategy around it. So just a word of advice for somebody who’s been through it. 

Gary Sorrentino (Zoom):

I think the first thing is look, be curious. Start working with it, right? You don’t have to go from what he said was you don’t have to go from solution to actually putting it in place. We can start looking at it. We can start adding more data as time goes on. But the problem with the data is it was before our time. Some of this data was generated and we don’t know the quality of it or the context that it was created. And so I think one of the things to go through today is I have this change. I was a CISO also at one time, and I was a doctor. No, that was my, you come to me, walk in. No, I don’t care. If you’re asking me, you go to lunch. It’s no. Right. But today, ask how. Don’t ask for permission. And I think that’s how we work with our compliance people today. We walk in and we say, I’m not asking you for permission. I’m asking you how do we deploy this? Because that forces them to actually put words on paper. They can type two, right? And not give you a complete No. And I think you need to change the model a little bit internally to figure out how do we get through some of these fears? And again, did we ever, ever have a hundred percent solution that we rolled out? But today for some reason, unless I get the data a hundred percent right, I’m afraid to roll it out. And I know internally we can do some things that we can’t do externally, but I think you have to be curious. You have to try the cases out. You have to figure out what’s working and what’s not working. But it is about trial and experimentation. Don’t be afraid of it. 

Scott Kinka:

Jim, word of advice. 

Jim Marron (Talkdesk):

Kind of almost pawn off what you said, ask. Find somebody who’s actually curious about your business. Find somebody who understands your business. Find somebody who cares. AI is in every conversation across the board right now. There’s a lot of Fugazi out there. Okay. There’s a lot of people who don’t really care. Just get you in the door and ask about the business, ask about the financials, ask about the health of the company, ask about their personal credentials, ask why they care if you find that person and why I’m so passionate. But what I do is I don’t have to do what I do. I can go back to doing whatever I want to do. I love this space because I care about the health of the businesses I work with because their success is our success. And there’s a lot of people who don’t think like that. So find that trusted advisor within that company. Do your research and find your match made in heaven. 

Yalim Eristiren (SESTEK):

Got it. Yeah. I mean, the first thing is identifying it. So you might come up with a large process and transforming that process with AI to something else might take forever. So, decompose that process, split it into a digestible piece and start with that. And again, I will reiterate that one, I strongly believe in introducing yourself as a thought partner. Thought partner. It is not a counterpart to the line of business but a thought partner line of business. They need you best. And that competence is going to be built on your end, not their end. They know their business and the technology piece, and you’re going to come grow this competence and help them on the transformation. 

Scott Kinka:

Yeah. Well, thank you all. Before we go, any parting questions from the group at this point? I know this could go on all night. This is a hot topic, but we are constrained by time, so one last round. Does anybody have anything else before we go? Well, I want to thank you all for the spirit of conversation. Give yourselves around. I’d love to thank Sim, New Jersey, obviously. Thank you all for inviting us in here today. For those of you who aren’t members and you find this interesting, please find somebody who can get you involved in the organization while we’re here. I’d love to thank the panel. Thank you guys. We appreciate that as we go. And finally, as has been the odd part of this project, doing this in public, we’d like to thank the listeners at home on Apple, on Spotify and YouTube. Awesome to have you with us, and we will catch you soon on another episode of The Bridgecast. Thank you very much.