Latest / The Agentic Wave / Brian Blackader | Waiting to adopt AI means failing in AI
Transcript
- Brian Blackader Agentic Wave: The danger is waiting in the technology to finish, right? It it won't. You'll hear things like, Well, we have a system upgrade or a day to cleanups scheduled to finish in twenty thirty-five. Then we can think about a gentic but the world will have moved on by then, right? You're gonna be toast. The other extreme is actionism, right? Like we're gonna do thirty to fifty pilots, which is gonna allow us to say we did thirty to fifty pilots and we have a thousand agents. I call it vibe spending. Instead of vibe coding, it's vibe spending. Almost half of folks told us the costs are higher than they expected. At some point the CFO is going to come and ask you, where's the return on my investment? It'll be very difficult for you to have a good answer. Intersection of AI and CX, it's the agentic wave, a shift so huge it deserves its own podcast. I'm Latney Conant, Chief Market Officer at Parlowa. Get ready to hang 20 as we discuss the latest customer experience trends, advances in technology, and most importantly, how you can confidently lead your team to ride this wave. Today I am joined by Brian Blackadder, a partner at McKinsey and Company. Brian, welcome to the show. Thanks for having me. Super happy to be here. So let's get things kicked off here. Again, you're day in and day out working with customers, trying to adopt AI in the service center. Some are crushing it, some maybe not so much. The ones that are getting it right, what are the patterns? What are the ones that are getting it right doing? This is something that we've spent a lot of time having to look at over the last couple of months, trying to figure out what are the winners doing differently. because as you said, not everybody is getting it right. In fact, the ones that are really getting it right and are are having the impact, I would say, are in a minority. We went out, we interviewed about thirty folks that are operators, that are in scale-ups, that are with the hyperscalers, that are with BPOs, to kind of have a candid discussion about what's going on. and then we followed that up by surveying about five hundred, about five hundred different folks in the space. And what came out was pretty clear. number one, those that are getting it right, that are actually having impact, they have cross-functional teams. Right. So they are putting folks together from across disciplines and not just making it a IT experiment or an operations experiment. It's really a team that sort of goes across the organization. We also see that if you look at our our recent state of AI research, which is not just focused on contact centers or customer care. But business in general, we see a similar thing there that three-quarters of high performers fundamentally redesign workflows and they're doing this with cross-functional teams. Aaron Can I dig on that for one second? One could argue that every technology project ever has been a cross-functional team and an effort between IT and the business. What what does the relationship look like that's different now? Is it more dedicated? There's probably a handful of things that we're saying. The first is it's starting with a real business or customer problem and not what we see in a lot of places is we need to get agentic AI put in place and then we do it in forty eight different places. what's the business problem that we're trying to do? So not just getting jiggy with AI. It's not it's it's not that. And there's and there's plenty of that out there. And we we've seen plenty of ex two if you go back two years ago, it was you know Nobody was paying attention to what it costs. They just wanted to talk about how many pilots we have going. That's that's changed, right? second is going into in deep into the underlying workflows, right? And and redesigning them, not just trying to take a process that may or may not be working well and and trying to identify it, but really trying to redesign and reimagine what that looks like. The third is putting ops or or customer experience, whatever you want to call it. It's called the contact center. product data and engineering teams together under one leadership team. And so do they come out of their current org to create a new org? That's sort of what you're saying, right? Yeah. What what whether where you puts put the lines in boxes, I think one could debate. There's never really a right answer, but what you have is a leader who, at least for the the purpose of getting these getting these programs off the ground are working under one leadership and aren't being pulled in four different directions. Right? The, the, the next thing, and this is always surprises me because it seems so obvious, but it's it's actually quite rare, is they're establishing measurable outcomes. Right. So going back to my point number one, that you start with the real business and customer problem. Okay, we want to reduce, you know, a certain type of intent coming in that a human has to handle. You know, how are we going to measure that? Right. And that they put they put KPIs in place and they're able to be able to track day by day and week by week if it's working. And then this is continuous improvement. It's not that you you know you come in, you deploy the technology, it works, and then you can sort of leave it. No, this is a a a sort of muscle that the organization needs to have to continually improve because also the reasons why people are contacting you and the problems that you have as an organization are going to develop over time. And you need to have that muscle to be able to identify them. and to be able to to adjust to how you're operating. So a couple of threads to draw on there. I want to go back. So what what I heard about the team is they're centralized with the right roles and quasi-dedicated. So this is their somewhat full-time job. They're not they're you're not putting a JIRA ticket in with IT to get help on your AI project, right? The everyone's shoulder to shoulder in the same org, we'll use the org, the term org loosely. The other thing that I think is important to think about is with SAS, you designed it and it worked the same. Right? If if you said this is the filter or this is the workflow, it it just executed on that. And so with but with AI, the cool thing about AI is it's going to do things maybe a little differently each time. It's also I think the challenging thing for people. And so this is why that post L live going live is almost the smallest part of the project. Does that make sense? And maybe that's what people are struggling with. Yeah. So going live with something and getting it to work in one case or not work in many cases, it that's that's the easy part, right? It's about how do you continually improve it and how do you industrialize it so that you can go from one use case to the next. And and what we often see is that for those that are let's what we put in the laggard category is there'll be having 48 different science experiments that are not connected to each other, all maybe in some ways trying to s solve some of the same underlying issues that you need, whether it's with data or systems or workflows, where the ones that are getting it right are sort of thinking through that. Okay, what do we do first? What do we do second? What do we do third? And if we what can we reuse from the first thing that we do that's going to help us on the fourth or the fifth thing? Yeah. One thing you talked about was just the the intent data and how rich intent data is. And really like prosecuting that data to come up with your roadmap. that's the real things customers are calling you about. So, and I think you had some great stories about people that truly found great business problems, sometimes even outside of of service that was just lying in that intent data. This is something that, you know, is so fundamental to being able to not only be able to make a gen tick work, but to in To be able to make sure that your training is right, that you were that you're having the folks that are working with your customers are are able to have the answers to the right things to be able to fix broken workflows. Never mind about applying technology to it. That's the next thing. But having this, this very we talk about level one up to like level five or level six and ten. So just maybe I start by giving an example. So level one would be like it's a billing dispute, right? You could almost you don't need you don't need any analytics, you can pull that from the IVR because people pressed one, I have a billing. Level two would be like, I have a billing dispute, but you want to go deeper. Like the customer doesn't recognize a specific charge or he needs to identify a specific transaction or determine whether something is valid or pulling information from another system. So then you're getting down to a much lower level. And I mean, I even had the experience earlier today speaking to an insurance company where they only have that for sort of first level, right? And it where is it coming? It's coming off of the IVR. Right. And that's That's not enough. That's it's it's not enough. It was never enough in the past, but it was super hard in the past. When you didn't have technology to be able to go down and listen to 10,000 calls, you would have to do samples. Now you can saying you can do it at the click of a button is simplifying things a bit. You need to have the intelligence behind it. You need to have people that know what they're doing. But let's just say you can click a button and you can start to understand these things. when you get down to that fourth or fifth or sixth level intent, the things become there are things that That you find out that weren't obvious when you're really look looking at the first level. I'll give you an example. We were working with a consumer electronics company that has like home devices. and you could imagine what the first level intents were. It was like, I have technical problem, I have a billing issue, I have a whatever. What didn't come up was like batteries and sensors in the first or even the second or the third level of intent. When we got down to the level five or the level six intent, what you could see is that There was an issue with batteries, which c came up with something like 20 or 25% of the calls that were coming in. And so it was like, boy, if we could just, you know, either address the root cause or genttify the resolution of those of those inquiries, we could really have a fantastic customer impact and save a lot of effort that we don't need to have. But that only became obvious when we were able to get down to this very granular level of intent about why people were calling, because it doesn't show up. in what you're able to pull from your standard systems like the IBR. This is one of my favorite stories. I think this is so critical because so first of all, people talk about CX that we we and I do I'm guilty of this too. I throw the word CX around, are we talking about CX? Are we talking about service? But I love that using the data from service to go back and retroactively inform CX. And I think for so long, people have used things like surveys and they haven't really been taken seriously. Absolutely. You might have gotten that information before by going through and speaking to 25 different employees in the on the front line. And it would be sort of anecdotal. Then you would listen to some calls yourself, but now you really have the proof, right? You can you can use the you can use the data that's there to really make the business case and it it becomes less of a guessing game. I even think back like five or six years, you would Do what I described anecdotally and it became a little bit of a bet. Are the things I'm going to do, are they going to have the impact? Now you you you actually have the ability to to to know that you're focusing on the right thing. And then the risk is in the implementation of it, not in did I choose the right case to go after? Yeah. Let's eliminate risks, shall we? It's it we're it's already a risky proposition. So there's already enough enough that can go wrong. Right, right. So I'm pretty passionate, and I believe you are too, about The future of the CX and service practitioners. Yeah. And where this ends up for them. And so you led with it not being an IT experiment. Totally agree. Then we talked about today, it's it's really a cross-functional team. As the technology gets better, which I think it will and easier for practitioners to use, is the future org that it really is the new hybrid workforce and that Service practitioners are running their own agents? are we going to be able to get to that? It's a it's a great question. I don't pretend to predict the future, but what I can tell you is that crystal ball. Come on, Brian. First of all, with all the talk of AI, what we still see is that in many organizations volumes are still going up. Right. And this is this if we if we go back to the if we go back to self-service, if we just use this as a as a It's not exactly the same thing, but I think it's useful for context. If we went and we looked at the number of contacts, human-to-human contacts in minutes that happened, I think between 2010 and 2024, grew two percent per year, despite self-service. And if you go it seems a bit antiquated to talk about it now, but we said back then there was gonna be need for the human because everybody's gonna be able to self-serve. What we saw is that self-service grew six percent per year. So three times as fast as as the human-to-human interactions. Why? it it it created all sorts of new demand, right? And we're seeing that already today. That there are there's demand out there that didn't exist before. And I can I I believe that's going to continue to happen. I have a an ongoing discussion with a wearables company. you know, I got this thing on my wrist here and like the AI they have on it isn't great. And I'm like trying to give them feedback. And of course there's somebody the AI is trying to solve them, but there's somebody in the background there that when I'm escalating the things that's having to handle it. That was something You would have told I was crazy if I would have s mentioned that this sort of demand existed, three or four or five years ago, but it exists today. So I I do think there's new demand that's going to be coming out. But to to go back to your question, I think your question was more about like, are are service practitioners ready? and I think the answer there is no. Right. what we've what we what we hear from from our survey data is that only 20%, and again, this is a survey of Roughly 500, 480 or something leaders in this space, only 20% say that their workforce is fully prepared. And the plans aren't there either. Only 15% are sort of thinking about that and have concrete plans about how are we gonna how are we gonna be able to upskill our folks. It's interesting. We just did a survey. I was like doing the math in my head, and 70% were said they were not ready, which is basically the exact same thing you found. Pretty close. Yeah. There's also and some interesting numbers that come out of this when you when you ask the question about are you gonna need less or more people, the forward looking answer is always we expect to need less people. When you ask about like what happened in the past year, almost the same amount, and it's like 50 or 60 percent. So we've actually had to hire more. Well, I think there's a learning curve. I I feel like we've underestimated, and this this is not just a CX thing. Like I think about my team and adopting AI. We're like, go forth, adopt. Here it is. You know, and and you should be so efficient right now. And it's like we never really think about, okay, let's give you a day or maybe a two days to actually learn it. you know, there's that y you need to you need to have time on the bike to learn how to read the bike ride the bike. And so I think that that whole learning curve has been sort of underestimated with this. Yeah. But what what I do think is that the technology is good enough now. Whether the cost curve quite works yet, I think, depending on the case and what you're comparing it to, is it's maybe not quite there yet, but that that that will solve itself. But the things that I talked about before about change management, about about having the business case, about you know, all of the sort of good management practices, if you want to call them that, of being able to deliver impact, those are the things that are still preventing. the that are still preventing, let's say, this from happening at scale. Yeah. But I I believe it's only a matter of time before that happens. Now whether the whether you see this inflection point of human to human contacts going down, whether that happens in twenty eight or thirty five, that that's that's a little bit of anybody's guess. But I do think the technology is good enough, but it's the the the change management getting people to adopt it is the part that's that's so far missing. So let's unpack that because there's change management for the end customer. There's change management for the internal team too. Like they're both aspects of change. What would you say are some of the top three or four things that great customers have done to address getting over that change hump? Yeah. So I think w when we think about change management adoption, there there are three groups that you need to think about. The first are the employees. The second is I would say IT. Not that they're not employees, but they're a different, a different stakeholder group. And then the third the third, not to forget, are customers, right? So for for the employees, they need to understand how their work improves. that this is not that that not simply that AI is coming, but how is it going to improve their work, right? You're not going to be handling the routine tasks anymore. You're going to be there for your customers in the moments that matter. you need to involve them in the workflow design because at if at the end of the day, the folks that are on the front line are the ones that understand what people are calling about, how the thing works, trying to design that from the headquarters is a bit of a recipe for a disaster. It has the people that are involved in the front line need to be involved. And then you need to give them a route to challenge and improve the system because as we said at the back, the leaders are the ones that are not just trying to apply technology to the problem, but rather trying to reimagine what the what the workflow looks like, how the operating model is. And that is that's where that's where the employees can help you. If you think on the IT side of things, we talked earlier about having these cross-functional teams. IT needs to work side by side with operations and not just be an enabler to the business, but really be a business partner and be working with them. It the winning model isn't just business writes the requirements and IT builds, right? Hey, this is these are the requirements that we have. And then IT goes out and procures or builds something. And then The third is related a little bit to my first point is the engineers are need to sit in the operation. Right. And that's often where we see the bottleneck is engineering talent, but the ones that are getting it right, the engineers are sitting there in the contact center along with operations. The last group on the customers, you know, we we shouldn't forget that the ones that we're serving at the end of the day and adoption is ultimately their decision. Of course you can force them, but they're gonna vote with their feet. some of the data that we have, 57% site preference for a live person. So that's definitely an adoption barrier. 51% lack cite lack of trust with AI. and four, you know, about half say they they they worry that they won't be able to reach a human. That being said, what again, this is survey data. What people say and how they actually act are quite different. And we have examples where we see that AI is actually preferable when we're surveying people afterwards. So it's sort of the post-call survey. We we had the example that I'll give you here is with an energy provider where in the particular geography where this client operates, they they have the customers have to go and read their meter. And so you as you can imagine, everybody is calling at the same time. You of course you can do it in the app, of course you can do it on the website, but humans are humans and they a lot of people just pick up the phone. And so so what you see is this is a very specific use case. It's a very specific problem. People are waiting forever because you can't hire enough people. To answer these calls. And so on every metric, whether of course on cost, it's better than the human, but also on the speed of resolution, on waiting time and on customer satisfaction, people are people are actually happier with this. Why are they happier with it? Because their the call gets answered right away and they didn't have to wait six or eight or 10 or 20 minutes. And so that is a a use case where you can see that customers might say when you survey them in a theoretical way, hey. How do you feel about AI talking to AI? If I hate it, I don't trust it. And I all the other reasons. But when when when you're calling to give the the the meter reading on your electricity, you don't really care. You just want it to be over and done with. You have to take into account the experience they're coming from, right? The the other experience is an IVR, right? That's not a great experience. They'd probably rather, you know, be able to talk naturally to someone and and and even if they don't end up getting resolved by the AI. Presumably the routing's at least better and they get to the right place. so so that, I I think you're right. The survey data, it's like, would I rather talk to a concierge that is a human who's my best friend? Yeah, but that's not the experience that we're getting right now. It's not necessarily the alternative. Yeah, exactly. Exactly. And that that matters. You talked about an upgrade cycle and w and sh is this something I mean Your data is saying this is a relatively challenging endeavor, right? And that some people are getting it right, but a lot are getting it wrong. Would your advice then be to just sit it out, wait for an upgrade cycle? I think the short answer is no. I mean, if you if you are have super happy customers and you're making a ton of money and you don't have pressure to do it, then maybe. But I think the danger is is waiting in the technology for to finish, right? They it won't. So what you'll hear sometimes is things like Well, we have a system upgrade or a data cleanup. It's set to, and I'm being a bit facetious here, I'm over exaggerating, it's scheduled to finish in 2035. Then we can think about a gen tick. Yeah, okay, you you could, but the world will have moved on by then. You're gonna be toast. That's the one extreme. The other extreme is actionism, right? Like we're gonna do 30 to 50 pilots, which is gonna allow us to say we did 30 to 50 pilots and we have a thousand agents, which is also not good. Then you have a zoo of of experiments which aren't necessarily driving impact. And so, w when I I would I would say there are probably a few things that you should be doing. The first is to pick two or three meaningful journeys. And it's going back to this sort of intent data. Pick the ones where the impact is, right? People are gonna it's not just that you're gonna, you know, make the journey better. It's gonna be less expensive and and and your clients are gonna be h happy. It's also it's gonna create excitement internally when you've done something that's impactful, right? Learn what data you actually need and don't wait for all the data to be solved. It's never going to be the data is is usually pretty messy in these enterprises, right? So figure out exactly what data you need and get that clean first. Understand the intents, build the integrations, work on the handoffs then between the human and AI. This is important because the AI isn't going to solve everything. So know when, when, when do you tap out and hand it over to an expert? Learn how to measure the quality and economics. This is Super important, right? This is a hot button, right? Absolutely. It sounds basic, but everybody gets so excited that they just start doing things and say, What is it we're trying to solve and how are we measuring it's if it's successful or not? And is it making economic sense? The the that you need to put that in place up front. And then I would say the last thing is building the operating model, right? We said that we said at the beginning. Having this cross-functional team and redesigning workflows, not necessarily to just jam Magentic or AI on top of what you already have. If you do those things, then you're going to be able to start seeing, you're going to start seeing progress. I I wouldn't wait for everything to be finished, right? It's never going to be finished. On the other hand, I wouldn't, I wouldn't recommend that people just start experimenting because. At some point the CFO is going to come and ask you where where's the return on my investment? And you'll be it'll be very difficult for you to have a good answer. Brian, let's talk about this. I I call it vibe spending. Instead of vibe coding, it's vibe, vibe spending on experiments. And, you know, according to the data and the survey that I saw from y'all, people are are getting a bit shocked. They're getting some sticker shock. Yeah. We're we're also seeing that. I mean, on the on the On our state of care survey. So I think what you were referring to is the one that was our broader initiative. We saw that almost half of folks told us that the costs are higher than they expected. And it's the number one challenge in value realization. And you know, these are these are popping up in a variety of different places. But I would say, you know, token and model inference of one, the APIs, the integration. We even see like some CCAS players are starting to charge now for the integration. then there's there's all of the the costs that are not necessarily directly related to technology, but the people, data engineering, QA and testing, monitoring, security and compliance, AI engineering, then the things out of ops like knowledge maintenance, and then knowing that you spent the money on the tokens, it didn't work, it's a failed interaction, and then a human needs to pick it up anyways, right? So you have this sort of escalation. And then there's something around around vendor lock-in, though, but I think that's that's That's that's probably less of an issue. but you're you're you're absolutely right. It is, it is a it is that this stuff is not for free, right? And I think that people were so excited two or three years ago, and also the economics were different back then. There was a lot of like all you can eat sort of models out there. It's not like that anymore. and so we we are seeing that as a as a hurdle. and again, I think if history is any any indication of what we're to see here, I I I I would put my money on the technology eventually solving that. But in the meantime, we're all being measured on a quarterly basis. Yeah. What do you think so a a couple of things. One, what's your take on pricing models and outcome based pricing? Does that is that the solve? It depends on the you're gonna it's a typical consultant answer, right? The it depends. But I look I a as you alluded to at the beginning of our of our of our conversation, the idea of outcome-based pricing, you can almost set your clock by it every five years. It it's just sort of in vogue. I know. It's almost impossible to do though. It's very hard to do. It where typically where it works well if we're thinking about the world of customer care, it typically works well where you can say the outcome I can attribute to the vendor, whether it's a technology vendor or whether it's a service vendor. If I could say, look, it's it I can I can very clearly see it's because of what they did. So this has typically worked very, very well in things like sales and retention, right? where where somebody is is is calling in, you know, take a retention hotline on you know, in a telco, right? I can measure things like what's my save rate and what's the R poo that I read. That I I retained. And the telcos typically reset the policies and what they're offering almost on a weekly basis, depending on what's going on in the market. So you can really say that 90% of the outcome is really is down to the performance of that particular provider. That's where it works really well. Where, where, where it starts to get a bit muddy is when we start talking about outcome based pricing for you know, I'm gonna charge you, you know, per. successful call. How do you measure that? there's a lot of and it becomes the you start to have a as time goes on, the discussion of evidently comes up. Is the is the result better or worse because of me, or is it because you, my client, if I'm the vendor, improved a system, improved a process, changed the system, changed the process, had an outage which drove the call volume up. And so that that's where it becomes a little bit more difficult. Everybody's talking about it again at the moment, and maybe now is the time that that that people are actually be able to, that it's actually going to work. But up until now, what we've seen is that it's been, it's been difficult to measure. I still think there's a place for it. And it's where you can very clearly attribute the improvement or the outcome in this case to the effort that the vendor has put in. I also think, I mean, we opened up talking about getting to your intent data and looking all the way from level one to level five. And reimagining the process, that makes this an a very simplistic outcome-based pricing model. It breaks pretty fast, I would think. Yeah. So if if in a world where a vendor is able to provide the technology, the system integration, the labor, the design, everything else, and the client sort of says, Here, take everything, then And they read then they do that redesign. Sure, why not? Then I think that what you just described could could work really well. But that's not that's not the reality of of how things operate. Typically in the enterprise segment in particular, the client is working with multiple technology vendors, they're working with multiple BPO vendors, they're working with multiple system integrators. and so the at the end of the day the ul the the the overall outcome still sits with the business and not with the sort of landscape of partners that they're working with. Yeah. I wanna also go back to cost. You know, so much of operations and service has been cost takeout, cost takeout. And now we're applying AI and the original assumption was, hey, the you know, let's let's try to be more efficient, aka takeout cost. We just said we're not that's not exactly playing out. Is is it really recasting to be like how do we make more lift out of this? how do we get to the more value add, upsell, revenue generating use cases? should should those be more the focal point? It should be both. I I I do think it's gonna take the it's it's gonna take cost out, right? And that's inevitable, right? Now it other intents might pop up elsewhere and the overall cost might stay steady. It's gonna depend on case to case, but I do think that that AI being able to help on the cost side of things is is inevitable. What I think most people are missing, and I think this is what you were alluding to, is that there's a whole nother element of it around loyalty, around revenue generation, that around just creating quite frankly a better customer experience that that that that makes this possible with with the technology today. And so if you think about out the idea of outbound, right? You might have to you you might have opportunities to call customers and offer them an upgrade. Think about in the airline business, if you have you know, empty first class seats, you could call your loyal customers and offer them an upgrade. Right. You might not have that you wouldn't have have had the capacity to do that beforehand and never mind the operation trying to operationalize that of you know, from customer value, lifetime management and to to like, you know, by the t by the time that got down to the on the floor of who's calling, the plane would have taken off. You you you you it would have taken off two weeks ago. you you you you can do that now. As far as I'm aware, I don't I don't know anybody that's actually doing that, but that's that's like a use case that seems super obvious to me. Or using using AI to be able to improve the product. I went back to the example of my wearable. I mean the the amount of data they collect they can improve the product like in in the the the the life cycle ha is just much shorter than it used to be. They can do it almost on a daily basis to tweak things. So I d I do think that this idea of moving from viewing this purely as a cost play to a revenue and loyalty play is a missed opportunity for many. Well Brian, we've talked a lot about making clients successful, something you're passionate about. We're passionate about here at Parlowa. We cannot wait to collaborate with McKinsey. Tell us a little bit about what McKinsey will be doing for Parlowa's customers and how we're better together. The the basic premise here is that the clients don't just want advice, right? They want it, they want a solution. And so in many cases, what we see is that they don't necessarily want to build, right? We we, of course, as McKinsey can help clients build a an agentic voice solution. That's not for everybody. And so they want to they want to be able to have a sort of all in one offering where we can come in and be able to bring the intelligence, the sort of business rigor, the business planning. And then what Parlova brings is the is the technology platform, right? And so we see a a fantastic opportunity to be able to work together on being able to Create tremendous client impact, right? We want to have ambitious clients, we want to have clients that that want to do things differently. And I think that that that aligns very well with how you guys are thinking thinking of things. So for folks listening who are thinking about getting on the agentic wave, if they're ambitious and they want to do things differently, this like McKinsey is a great place for them to start. Would you is that a fair assessment? feels it feels a little bit like a loaded question, but look, we we the when we think about what is a great client, a great client, first of all, you need to have a chemistry with, right? We and and we're not serving companies, we're serving individuals. So you need to have you need to be able to get along. The the second is they need to, as as I said earlier, be ambitious, right? Not accept the status quo and say, I need to change something. And I think the third thing is be open to help, right? Not you you'll have a lot of people that check the first two boxes, but if you are resistant to new ideas, that that makes it really difficult. So those are always the sort of three things that I'm thinking about when I'm looking to when I'm looking to to collaborate with my clients. do we get along? Are you ambitious? Are you open to my help? and and the and the sort of institutions collective knowledge and then can we find something that makes sense for everybody to work on well for folks listening, I think that They're probably thinking, wow, I definitely would love this guy's help. He is a treasure trove of knowledge. I so appreciated hearing the data and the research that y'all have done. that was really valuable. But then the client examples were just phenomenal. it was amazing to hear about the company that got, you know, multiple levels of intent down and realized it wasn't a service issue. It was a battery. Like, let's go fix the battery and we'll have less service issues. I also thought it was really interesting your take on, hey, we've been trying to do more self-service forever. And the more self-service we do, the more service there is to be had. so it seems to me that customers have a ferocious appetite for service, and this is an opportunity. Again, back to the cost center versus loyalty center. Customers wanting to talk to you is not a bad thing. and so hopefully with AI, this is a big opportunity for everyone to Embrace that. as we close here, Brian, any closing thoughts, two or three things that people should make sure to think about again as they think about getting on this agentic wave. So the first thing I would say is start, right? Start now, but be focused. Take the extra two or three weeks to plan it out, or four weeks, however long it's gonna take you, but but get started, right? And that and what does the planning out mean? It means don't Plan to automate everything, choose the meaningful customer journeys, get that intent data down to level four and five. It'll really open your eyes. put operations and technology together and don't wait for the perfect data because it's most likely isn't going to exist. And then lastly, measure the outcomes that what you can learn from, you can improve, and then you can scale it. Awesome. Well, I took a lot of notes. I'm sure everyone listening also took a lot of notes. Loved having you on the show. Thank you so much for your time, Brian. Thanks for having me. Brian, thank you for such an amazing conversation. It was jam-packed of all kinds of great information. If you enjoyed it, don't forget to like and subscribe.