Program Close with Bob Laliberte & Zeus Kerravala | The AI ROI in Contact Center Summit

Well, welcome back. I hope you enjoyed those sessions. I know Zeus and I enjoyed doing them. Um, really want to talk about, you know, the fact you saw four different perspectives, but the conversation consistently moved beyond basic automation towards endtoend governed resolution. And I think that's really what's going to be important for organizations as they're moving forward and adopting this technology. Zeus, what do you think? >> Yeah, I I love the sessions, Bob, and I think it's solidified in my mind that uh we're all growing to the same lighthouse, right? We're converging on that common destination, but we're taking much different routes to get to get there, right? There's data management issues. There's platform integration issues. There's multi- aent automation. There's um uh you know security concerns and implementation discipline, operational transformation. There's a bunch of things that have to happen for this to work. Uh we know where we're going. And what's good though, I think, is that there's different paths to get there. No two customers are alike. They're going to want to take different journeys. And what we saw was a pretty good variety to help customers choose whatever path they want to go down. So um uh and and uh let's drill down in some of the takeaways. Bob, I think, um uh the first one that I took away is that resolution really here and resolution quality is the new unit of value, right? Bots were often measured on containment or deflection. Uh agentic system should be judged on whether the customer's needs were completed, right? and completed accurately across the full journey. >> Yeah, absolutely. I couldn't agree with you more. I mean, and I think this this actually ties into maybe a little bit of a a broader scorecard. You mentioned not just resolution, but the quality of resolution has to be there, the you know, the customer effort to to get to that, the seessat, the employee productivity, uh, and you had mentioned earlier cost, all those all those things really matter. Um and so you know the growth use cases, retention, recovery, uh qualification and proactive service, right? All these things help to expand the business case beyond just savings. >> Yeah. And I think it it highlights the fact though that this industry desperately needs new metrics uh with which to measure success by. Right. We've used the same metrics for success for the last 30 years. every contact center manager I talk to does want a better way to measure AI success, but um I I think that's that's something for the industry to work on. It's because it's lacking right now. >> Yeah, absolutely. >> Yeah. >> Um and you know from my second takeaway would be that um you know really that it's that having that context and connected data are really foundational for success. uh if you look through the the videos I Cisco made uh you know context the control point Zoom emphasized connecting interactions and knowledge talk desk focused on orchestration across their system and 59's reinforce integration as a condition for time to value so I mean all of these are really just variations on on the same lesson that need to come across >> yeah and um they are uh but again the journey to them is slightly different so um you know one of the The interesting the byproducts of the context and connecting data being foundational is that um and I've heard other people talk about this like AI is going to make you do things faster right but if you've got a broken process you're going to get to that bad destination faster that's all it's going to do so I think if you have and in fact uh we've this for in data scientists we say good data leads to good insights right but silos of data or fragmented data leads to silo to or silos of data lead fragmented insights and we don't want that right. So you know that combined with stale knowledge you know really this industry filled with a lot of inconsistent workflows all that limits accuracy it forces customer repeat themselves and it grows customer frustration and so I think it's critically important Bob that context exists across channels different tools different systems and most importantly that human digital agent handoff >> and it and it really speaks It speaks to the need to ensure that you've got you've done the the homework prior to deploying the technology so you know where that data is and you know how to connect it and integrate it. >> Yeah. No, and it's uh measure spice cut once right. So um yeah and I I think one of the great takeaways we got from this is how to get started and I ask companies that have implemented a caucus all the time. How do you get started? And the answer is what we got out of this in you just have to get started right now. With that being said, don't try and boil the ocean. I I've described this journey as uh chip shots not moonshots. If I send a moonshot to create this AI driven contact center tomorrow, it's going to be so overwhelming that I won't get it done. So uh the practical advice was um uh was was uh that we heard was consistent across speakers, right? Choose a bounded high value high volume high value problem with the clear owner implement it manage it measure to the baseline measure the value and then use that as success metric to to go forward so uh I do think that um it's tempting to try and transform the entire entire customer journey you know with your 1.0 battle or loose but it's release and but it's very risky and uh so start small learn some lessons and then expand from there. >> Yeah. And I and I think one of the other key takeaways from that is the fact that when you're doing your first deployment, right? Think about you don't want to become so so focused that it just becomes another silo, right? So it's important to make sure that you're selecting an architecture that can connect different data and systems, right? that has the ability to reuse governance that you've put in place to save you from, you know, having to do the same work over and over again. And and really thinking about it again, it's so this is I I think I might refer to this more as you've heard me say this before is you you're thinking about buying the book, but you're starting with chapter one, right? When you're thinking about that architecture and so forth, you need to be thinking about while you're going to start with chapter one, think about how you're going to be able to expand. And in this case, that might be from just, you know, AI assistance to workflow completion as that confidence grows. So, start small, gain success, but make sure that as you're going to expand that you've chosen the the appropriate architecture that's going to enable you to expand and continue to grow. >> All right. Um you know fourth takeaway for me uh talked a lot about governance just now but testing and governance really need to become operating discipline. So as as these AI agents gain access to all the various tools and have transactional authority, you know, just doing pre-production testing is probably not going to be enough, right? Organizations need to be able to continuously evaluate, observe, enforce policies, have traceability, and and really retest when models or workflows change. So, I think that's going to be so critical as you start deploying more and more agents that you've got that observability, you've got the ability to evaluate what they're doing, ensuring that they're complying with the policies and so forth. And you're right, it's it's going to be kind of a a continuous testing mode as far as I can tell. >> Yeah. You know, Bob, I was at an event last week and I was doing a session with the CFO in governance and I asked people in the audience for show hands who likes talking about governance and surprisingly no hands on it, right? And I I think with AI, governance has an opportunity to change the way it's thought about. Historically, uh, governance and security gets in the way, right? We're ready to rule something out, but we can't because we've got all these rules and regulations we got to comply with. With if you have the proper governance in place, you can actually move faster with your AI initiative. So, it should be something that enables adoption, not holds it back, right? And so, but with that being said, it's got to be baked into the design. It can't arrive as kind of this late stage thing that we then worry about once we've we start some of the adoptions. So, um I think when you're planning, right, you have to think about what the agent could do, what data it needs, and what evidence it needs, when it needs to escalate, and then who's accountable for the outcomes. And I think if you do that, then um and it's really about measure twice, got once, right? But you want to have the that proper foundation in place as you design these things. And that's historically where we get bogged down because we bring it in after the fact, right? So, um, for me, takeaway five, um, is the workforce model changes, right? And so, um, we're going to have digital agents. We're going to keep people, right? We talked about that extensively throughout this. But human agents will increasingly have to handle the exceptions, the complex tasks, uh the emotionally sensitive moments when uh you know human empathy is required uh where judgment needs taking place. AI can do a lot of things. It help reset your password. It can help you fill your balance. It can help you book a flight, right? But those are standard processes, right? And um so I think in some regard while we're all worried about AI taking jobs right in the context of customer experience it raises not lowers the need for training knowledge management knowledge access you know agent assistance you know and quality measurement and I think those things uh when we think about this world of what it looks like when we blend our our human and machines right um the people play a really important role but we got think about them differently. >> Yeah. And I I couldn't agree with you more. And I think, you know, when you think about it from that perspective, you know, and this has been talked about a lot and Larry Ellison has talked about this a lot, right? Supervisors must manage a a a blended workforce of people and digital workers. Um, and so, you know, you're you're thinking about the the fact that you're going to work today, you're not just managing people, but you also have those agents and digital workers that you manage. And so there's going to be a combination of skills that are going to be required for that and how to handle that. And you know, organizations are going to have to really look at how are they going to be doing their forecasting for resource need, right? Quality management is going to become really critical. Um I think you know coaching is going to be another aspect of it. Um how do you coach your supervisors to manage an agent, right? And how is that going to be done? Right? So it's a whole different skill set that organizations are going to need to adapt and be able to bring into it in order to ensure success for these for these organizations. And then obviously, you know, the capacity planning, right? All these things need to reflect how the work is going to be moving between AI and humans. And like you said, there's still, you know, a lot of good reasons why you need to have that empathy. You need to have that human judgment. and and so trying to determine that blend of driving that optimal efficiency between agents and humans. Uh because I know a lot of times when when I get on and I'm talking through something at a certain point I'll just be like I need to speak to a human please just get me to a human and so u the ability to recognize that and redirect people so that they can have a positive experience is going to be super important. Yeah, you brought up an important point too with with the supervisors because a lot of these tools are built for the agent and helping them be more productive and helping them be smarter and more accurate, right? But what about supervisor, right? So, make sure that as you're rolling these tools out, the the the managers and supervisors understand what's happening so they can coach better, right? That they can understand where AI is working, where it's not working, right? And I I think that's uh um that's something that doesn't get talked about enough. Yeah, absolutely. And I think, you know, um it was a lot of great points that we just brought out, a lot of great similarities between the organizations that we heard from today. I think there were also some some areas where they where they differed. So, I thought we' we'd chat about that for a minute or two. You know, looking at Cisco, you mentioned they're very large organization, right? Their lens tends to be more architectural, tends to be more governanceled. Uh organizations like talk desk as you mentioned, right? brings up that end-to-end and industry specific automation in a lot of cases. Zoom and its partner, right, really emphasized that integrated tools and testing and pragmatic deployment. And you know, 59 framed the the maturity journey and really provided a lot of customer proof of the value it's delivering for organizations. >> Yeah. And while the difference in these companies maybe can you might think make things confusing, it's actually a good thing. So I think gives buyers a context and evaluation criteria that they can use to make a better decision. So the right choice for every organization will be unique to them, right? A lot of it has to do with what you currently have installed, where you're coming from, where you're going, what kind of workflow complexity you need. Is simple, you know, you have a lot of simple interactions, a lot of complex ones, high net worth individuals, low net worth, right? Different industry requirements, what kind of resources you need to implement, right? Does it need to be partnerled or not? and whether the priority is the service efficiency or is it uh customer experience uh improvement or is it just growth of the brand right different brands have different needs and I think it's good to see you know this a different variety of vendors like this uh approach through different lenses and that's obviously good for everybody so creates a bigger pie >> yeah absolutely and so so to kind of bring this home Zas What do we recommend for organizations? What's their action plan coming out of this? >> Yeah, I think first uh and this came through loud and clear. It's choose a journey, right? Don't boil the ocean. Choose a journey, baseline it, document it, and then measure it. And so this includes uh you know, volume of calls, transfers, hail time, resolution, customer, customer effort, um um you know, quality measurement. There's a bunch of different things you can you could measure. I'm saying pick a journey, find the the key metrics for that and me baseline it and then measure against it. >> Yeah, I I think that's a great way to start. And then I think I think next what's what's really important is for organizations to be able to to map out those workflows and the processes they have in place, right? understand the data knowledge, the different systems, any policies they have, where are the human escalation points that are required to resolve that journey from beginning to end. So really document that and have that ready to go so they can work with their provider to incorporate that. >> Yeah. And uh last I'll say it's important to run a controlled deployment. Don't let this get out of hand. Don't let the scope get too big. Um, I think you want to be able to uh see where you've been, know where you're going, but then also importantly test a lot of the edge cases. We know a lot of the core business has a you think of the 8020 rule, right? That most of interactions are going to fall in the 80. Well, test the 20 as well, right? And test those continuously. Pair results the baseline and then expand only when trust and quality hold up. And I think if you do that, you'll have a successful deployment. >> Yeah. Absolutely. Well, look to wrap this thing up, uh, you know, AI ROI in CX won't be determined by the number of bots deployed really clearly, it's going to come from getting to better resolutions, being able to have more capable employees and more efficient operations and responsible execution at scale. >> Yeah. And I want to obviously thank all our all the guests that we had. It was a great summit. I want to thank everyone for watching this. And I'll I'll leave you with this. Uh I think sometimes with the new technology, we think, well, that's risky to do that. Right now, we're in a period of time where I think it's riskier to not use AI than to use AI because the world is moving that way. So move and move with urgency, but do so in a measure scaled way that can provide you the the the the data, the evidence, and the discipline to be a to have success and ultimately build customer trust. >> Yeah, I I think that's a that's a great way to to close. And I think um I'll I'll paraphrase um Jensen at Cisco's AI Summit when he said, "Everyone's talking about human in the loop. It's time to make sure AI is in the loop." And we need to make sure AI is in the loop now in all the processes. >> Yeah.