Brandon Teegen on the CX Infrastructure Gap Nobody's Talking About

The contact center is ground zero for agentic AI, and most enterprises have an infrastructure gap they haven't fully reckoned with yet. In this episode, Shawn sits down with Brandon Teegen, VP of Solutions at Cresta AI, to explore what happens when AI stops just answering questions and starts taking action on your customers' behalf. They dig into how to build an AI agent strategy before the volume tsunami hits, why the Klarna story is less about a bad decision and more about a missing playbook, what Cresta's "Agent Operating Center" model reveals about the future of human-in-the-loop supervision, and why agent-to-agent interactions will force enterprises to rethink authentication, brand governance, and margin models from the ground up. If you're a CIO or CTO with a contact center in your portfolio, this is the briefing you didn't know you were missing.

What You'll Learn:

• Why the shift from AI answering questions to AI taking actions is the CX infrastructure gap most enterprises aren't prepared for
• How to sequence your contact center automation strategy — internal use cases first, customer-facing second, and why the order matters
• What the Klarna story actually teaches us about AI deployment risk (and why "dip your toe in" beats "flip the switch")
• Why personal AI assistants like Siri and Claude are weeks away from driving 2-3x your current contact center volume — and what to do about it
• How Cresta's Agent Operating Center model redefines the supervisor role when AI agents are handling the floor
• Why agent-to-agent interactions will require entirely new authentication, discount thresholds, and brand governance policies
• How AI agents are becoming an extension of brand identity — and why marketing needs a seat at the contact center table
• What the CFO conversation about AI token costs needs to look like before spend outpaces the savings

Transcript

SPEAKER_00 0:12

Hey everyone, and thanks for joining the Amplitude of Tech Podcast. I'm Sean Cordiner, Chief Marketing Officer of Amplitude. Today I had on the podcast Brandon Teagan. He's from Cresta. And we had a conversation about not AI taking over the world, but AI taking over your contact center. This is one that you have to listen to. Hope you enjoyed it as much as I do. All right, Brandon Teagan, welcome to the podcast. Thanks, Sean. Excited to be here. Excited to have you. Maybe just take two seconds and tell everyone who you are and what you do.

SPEAKER_01 0:46

Yeah. Brandon Teagan, based down in Atlanta, Georgia. I worked for a company called Cresta. Prior to joining Crusta, or I've been for about five years, I've worked at large companies like Salesforce and Okta as well as small companies called Dynamic Signal. And I've worked in pre- and post-sales roles, but currently at Cresta, I lead our solutions engineering team, global team with about 50 SCs around the world. And what does Cresta do exactly? Yeah, so Cresta is AI for customer experience. So think of a customer service use case. You call in, Cresta powers AI agents that are automating and handling those interactions, but Cresta also powers the human agents. And so if you don't want to talk to a robot or you've got it too complex of a task for an AI agent to handle today, that gets handed over to a human agent where we're powering them with agent assist. And then because we have human agents and AI agents all on the unified platform, we give visibility and insights across 100% of those interactions, as well as things like quality and QM across every single customer interaction along the full customer journey there.

SPEAKER_00 1:46

Got it. So that was well rehearsed. I think your marketing team will approve of that now. I do want to ask though, you when we talk about AI and human agents working together, what does that actually look like?

SPEAKER_01 1:58

Yeah, so I think we're actually starting to see it right now in the context center quite a bit. And I think we're seeing it a few different lenses. Like one is humans and AI working together. You probably use ChatGPT or Cloud today in your pro Summer Life. We're starting to see like many enterprise applications of that same type of interaction. But it's also starting to shift a little bit of like, I think if you look back, maybe the last two or three years of your early adopter of JatGPT, like AI was very helpful for answering questions. Now it's becoming a lot more helpful for performing actions. And so whether you're building a deck, you're creating a one pager, whatever it might be, now there's actual agentic workflows that are happening. So we're starting to see the same thing in the contact center as well, where it's no longer just, hey, AI is helping to transcribe what's being said or discussed in the conversation. But instead, it's well, what are the actions that the human agent's trying to perform? How do we use AI to drive a workflow that creates a lot more efficiency in that process that ultimately just makes the agent a little bit more effective?

SPEAKER_00 3:00

So we're hearing a lot about agentic AI as being kind of the next phase of this adoption curve. And we're seeing a lot of pilots in certain areas in the business, contact center being one of them, but there's a lot of opportunities to automate using AI in your contact center. How do you prioritize which ones to start with and which ones to address later?

SPEAKER_01 3:25

Yeah, great question. I think I've always said this about the contact center, which is to me, the contact center is a giant science experiment because literally everything is metric. Like you know how long agents are sitting there waiting to take a call. You know how long they're speaking on a call, you know what hold time is, you know what ACW is, you know what entire handle time is, you know what cost per call is, et cetera. And so part of the reason I think if you think of AI today, right, like AI for coding automation is probably the number one application for AI automation within the workforce. I think a close second is AI for customer service. And part of that is because it's just really easy business value story, right? It's easy to build a business case and an ROI model to be able to effectively say that, hey, if we give 50% of our team this tool to use in 30 days, are they able to do the work that they were doing before in less time? Is our cost per call dropping, is our handle time decreasing. And so I think the contact center is kind of a unique place where there's just so much data that exists and so much that's already quantified, that it just makes it really easy to find initial starting points there. And I think the the second piece is that so there's one aspect, which is like AI applications, building the business case, determining where to start. There's the internal aspects of again, like summarizing call notes on behalf of an agent. You can fully automate that. That saves the agent time, that saves the business money. Then there's also the customer-facing aspects. It's like, are your customers ready to engage with an AI agent? And if so, there could be even more automation savings for the business. But you have to make sure that your consumers are ready, that those journeys that the customers are calling in to go down are actually fully automatable. So when I think of like, well, where do we begin our journey as an organization with AI? You're gonna think about it in both lenses of are there internal applications that have some type of positive ROI or business case that can be built? Also, customer facing, different ballgame, different set of requirements, garbils, et cetera, that are needed. How do you make sure that you're thinking about all of that as well? So you're not starting in the place where there's a much larger monetary opportunity, but you're doing it again, maybe at a point in which your customers aren't ready, or maybe your systems and processes and uh applications might not be ready to fully automate that today.

SPEAKER_00 5:51

So is there a heuristic that the industry uses to help value the difference between increasing the velocity and supercharging the power of a human agent by using AI versus replacing the agent? I don't mean let's fire people or not hire people, but I mean let's let's make those people more effective by uh containing some calls that are low-level, easy calls and allowing them to focus on the more complex type of calls. So if you're looking at you know two opportunities that appear to be equal and you're thinking, where should I start here? Should I focus on the uh end user facing customer experience aspect of AI, or should I start looking at how I can help that human agent perform better?

SPEAKER_01 6:39

Yeah. So I think at the end of the day, it's effectively like, what metric are you trying to optimize? I think if you look at I think it was like Klarna a few years ago, that they cut their customer service team, I think it was like several hundred, maybe like 700. They did that because they could almost fully automate all of their customer service calls with AI. But it so poorly degraded their CSAN score that it they actually decided that they go back to hiring a bunch of people because they weren't confident and their customers were enjoying the experience that they want. And so I think that's a simple example of just, hey, like what is the KPI that you're trying to optimize? Because if you pull one lever, it's gonna have some impact on another lever as well. So I think it's really important to think about that. And then I think the second piece is in terms of the balance between what should AI handle versus what should a human handle in a contact center. I think there's an interesting paradox, which is like right now, the vast majority of customer interactions are handled by human agents. In the future, whether it's 10% or 50%, don't know exactly what it might be. And it's gonna depend probably a little bit on industry and vertical and uh those types of uh components, but it's gonna be smaller, but that smaller percentage that humans are gonna handle are going to be highly complex. And again, it's gonna be the things that can't be automated for a number of reasons. And so I think that's where it's okay, cool. It's not just thinking about, well, what are the easy to automate use cases? We can't just attack that with an automation strategy. We also have to be thinking about what does that mean for all the humans who are gonna be in the contact center handling way more difficult calls? Because they're not gonna get the easy ones, it's just gonna be the hard ones back to back to back. What is that gonna mean in terms of their performance, their satisfaction, their ability to support customers? And then finally, the the last piece that I'd say as well is that I also think there's a lot of there's low-hanging fruit with AI automation use cases that like probably don't really need to exist. There's so many companies who are right now just focused on like Wizmo calls. So if you're a retailer, someone's gonna call in and ask about status order. Okay, great. We're gonna automate that with an AI agent. Like, fix your app. Put that on your website. Like, there's other ways to solve this that aren't literally just building an AI agent to handle those types of tasks. So I think about it through those different lenses of what is the impact going to be as we start to build our automation strategy. Are the things that we're automating actually the things that we should be investing in? Or should those conversations never have hit the context center to begin with? And how do we help to get ahead of that? I think those types of dimensions are really important in just determining kind of what the appropriate starting point might actually be.

SPEAKER_00 9:25

Yeah, I just did a podcast episode with our president, Adam Renner. And that was one of the things that he pointed out, which I think is important and probably not said enough, which is AI doesn't have to be the answer for everything. In fact, if there is another answer that's not AI, you should really explore that option before you implement AI. But I can't let the Klarna thing go without coming back to it. Like, you have to wonder how decisions like this get made. Like, how do smart people get into a boardroom and say, hey, let's fire everybody and let's just automate this process and let AI handle our customer service? Like that is a completely predictable result that the experience is going to be degraded. Don't you agree?

SPEAKER_01 10:05

Yeah, I mean, I think the reality is that the the business case is incredibly strong. There's also many channels where it just feels low risk, things like messaging or written channels, where again, like, is it a poor experience to get a response to an email instantly, even if it's automated for waiting 24 hours for a human to get to it? I don't know, pros and cons to things like that. And so I think that's a little bit more complex than just saying, like, oh, that was that was silly. They shouldn't have done that. But I think what I would what I would say is that they're on the far extreme of like, well, they could have cut 70 people instead of 700, see how it worked, done a little bit more A-B testing to just make sure that there isn't a huge drop in the most important key to add to the business, which for that segment it was very much focused on CSAT. And so that's why I think it's not necessarily that they shouldn't have done it. I think it does just show that you need a strategic partner to help you go on this journey. It's unfortunately not just flip the switch and say, okay, great, we can get rid of our team and this is fully automated now. Instead, it's how do we dip our toe in the water, prove out that there is a strong business case or customers are ready for it, that are system supported or application supported, and that we're ready to do this now and we can do it in a safe, smart way. I think that to me is the big learning of go through the logical progression of do this the right way instead of do this the fast way, even though there's a really compelling business case at the end of the story.

SPEAKER_00 11:31

Yeah, I guess that's my criticism is it's you don't put on a Superman cape and then walk to the top of the Empire State Building and jump off as your test case, right? Like maybe climb on top of a chair first and see what happens. 100%.

SPEAKER_01 11:45

It's a it's a good scary analogy, but a good analogy for sure.

SPEAKER_00 11:49

Yeah. So I want to get out of the enterprise for a second and out of the contact center and talk about humans briefly. So we were talking a little bit before we hit record on the podcast, and my dog came in, my dog monkey. She's made an appearance or two here on the podcast. And it made me think like people are using AI as friends, as company, but they the younger generation, especially, that's like their number one use case for it. My generation, Gen Gen X, seems to be we use it for search. I know that's how I personally use it uh mostly when I'm not using it for work, but they're using it for companionship. And that's growing, right? That's a growing space. So I'm wondering from your perspective, do you think that AI is the new man's best friend? Or is my dog safe?

SPEAKER_01 12:39

I think your dog is safe in certain aspects. Hard to play catch with AI. But no, I think it's uh there's a risk, right? And I think that like it it's interesting. There was actually today, Sam Altman was interviewed by Time magazine, and he was talking about the fact that not as many white-colored jobs had been displaced as he had expected at this point in time. And part of the reason that he he said that that's the case is that he didn't actually realize the amount of like emotional value that people weigh on having actual personal relationships uh with our coworkers. And so that to me is kind of relevant here in for insofar as right, like it's one thing to be able to just talk to someone, it's another thing to be able to experience something with someone. And so I do think in terms of reliance on companionship, and like there's definitely risk, and we've started to see it in terms of relationships that people have built with AI. But I think the reality is that that's gonna come under more and more scrutiny. There will be more guardrails and constraints that get built around what AI can do to help prevent that getting out of hand. But like any new technology, the more people that use it, the more outliers and strange stories that you'll hear. And so it's definitely not to say that it's gonna go away anytime soon. It will probably hear more stories of crazy things that are happening in the world of AI. But I think just in terms of the overall companionship and the reliance on it, I think um we'll probably start to see it plateau at some point. But then again, just as the proliferation of the technology just becomes a little bit more commonplace around the world, like you'll you'll get those stories every now and then. But I think your dog is still.

SPEAKER_00 14:20

All right, I'll let her know. I know it's like impossible to feature cast with this kind of thing because it's just evolving so fast. But if we if we set the time frame in kind of a short period, six months, a year, how do you think AI will continue to be integrated into people's personal lives?

SPEAKER_01 14:37

Yeah, so I think back to the beginning, right? Of like to me, AI for most people up until now has been answers, not action. I think the next six months, six to eighteen months, it's gonna be AI powering actions, not just answering questions. And so this is the big fundamental shift that is gonna hit most large enterprises. Right now, if my internet goes out, I'll complain and moan about it. I'm not gonna call my internet provider and ask for a discount or a refund. But if Siri could make a call on my behalf, complain, ask for a refund, or Gymni is gonna send an email, or Claude's gonna take care of it for me. All of a sudden, I think personal assistants who are actually becoming like really intelligent and smart and powerful, writing emails, sending text, making phone calls, that's a complete game changer. And it's also a huge benefit to me as a consumer. And so I think that to me is like the coming tsunami that's gonna hit many large enterprises, especially the contact center, which is if AI is actually able to help transform the consumer from AI as the tool to answer questions to AI as the tool to perform actions, those actions create more communications, more customer interactions, et cetera. Then all of a sudden, like the world has completely changed. And if 10% of your customer base were the ones who would drive 90% of the interaction to your contact center, like now it's gonna be 50%. Holy smokes, are you able to actually handle all of that volume? And so that to me is very much why I like to push most enterprises to be thinking of like, imagine two to three X the volume that you're seeing today in your contact center. The vast majority of that is gonna come from personal assistance, human agents, agents on behalf of the human, not humans themselves. And so that to me is what's gonna fundamentally change. And I'm optimistic we're a week-ish away from WWDC with Apple, and we're gonna talk about Siri and some enhancements there. I'm not sure if it's gonna be right now in Siri, but 100% in the future, Siri is not gonna be as useless as maybe it is right now. And that's gonna just fundamentally change what the context center landscape is gonna look like, I think, for many uh many companies who probably aren't ready for it quite yet.

SPEAKER_00 17:02

Yeah, so this is why we wanted to talk to you is because we had you at our technology leadership conference and you were on a panel and you brought this up. And that's actually something I hadn't heard. Now, maybe I'm living under a rock, I don't know, but I hadn't heard this prediction before, and it makes total sense now that you say it, right? So, I mean, what is the impact going to be? Like, it do you think it's going to be a tsunami and people need to have this capacity built right away, or it's gonna jam things up? Or do you think that you're gonna see the gradual increase of traffic and you'll have time to build out that infrastructure?

SPEAKER_01 17:36

I think it's gonna be quick because I think it's gonna be so easy that everyone can use it. And like my barometer, maybe this is like offensive to my mother, but like my barometer is if my mom is able to use it, then I think it means that there's broad applicability of like most people will be able to leverage this. And so I think that's where, again, it it's because it's series on billions of devices. So like it's the proliferation is there. Chat GPT is used by billion plus users. Like everyone is on these tools already. As soon as they can tap into sending email, sending a text, making a phone call, like the game fundamentally changes. I also think that those companies will probably be somewhat under pressure of like, well, C or E can't be placing 150 calls a day on my behalf because like the Apple will get sued and things will happen. But I think the reality is that like tech savvy people will adopt it very quickly. The vast majority of consumers, I think, will be very close. I don't think we have years to prepare for this. We might have months to prepare for it right now, but I do think it's happening relatively quickly. And by the way, it it takes months to even put an AI agent in front of your customers and get that ready. And so that's to me like the urgency is you want to deploy agents in a smart way to handle customer interactions, three to six months. Okay, cool. Like knowing that's at some point what you'll be able to start to handle some of that traffic from. Okay, cool. That means that like we just have to hope that it's six months out before Siri is making calls on my behalf, because if not, many organizations are gonna be in tough spot. And by the way, that's only the organizations who either currently have a solution or are actively implementing that solution today.

SPEAKER_00 19:21

Yeah, and if your timing prediction is roughly correct, it's even worse than what you described because best case scenario, you're looking at a three to six month procurement process as well, even if you use a company like Amplix to help you through that process. So you you're stacking six to twelve months on top of um you know, or into that timeline there that people are gonna have to deal with. So, what are the specific areas that you think it's gonna impact in the tech stack that people should be looking at right now?

SPEAKER_01 19:49

Yeah, so I I mean, I think for one, it's definitely going to be uh and maybe um how to put like this is where I think uh in let's say two to three years, I think the vast majority of the way that people interact with enterprises is just gonna be fundamentally different. In the past, if you had a problem like Google company name, customer service number, you dial it. That's how you get in touch with them. But I think what we're starting to see a little bit more of is in the app or on the website, there will be a button to click to engage with that brand. And that's a little bit different because I think it also changes the dynamics in terms of like the CKS, right? And so I think if you look at the CKS 100%, like they're gonna see more traffic, more volume, they will be impacted, making sure that you've got the right CKS who can help you scale is really important. But I also think you're just gonna have a lot of these touch points and interactions happening outside of the traditional channels. And so that's where I think the ultimate burden is gonna become on the systems. And so for anything that you need to actually, for any customer journey that you need to support, an action that needs to be performed, what are the systems that sit in that process? Do you have a CRM? Great. Is it on-prem or is it in the cloud? If it's on-prem, does it have any APIs? Can we do any customer data lookups? Yes or no? Okay, great. If not, that's gonna be really difficult to build anything solution to help actually automate that customer journey if the CRM is somewhere in the critical path. So I kind of just look at it through the lens of like, what are the most common journeys that we expect people to be reaching out to our brand based on those journeys? What are the systems and applications that are in that critical path to support? And that's where I think CRMs, order management systems, anything that just has the data that exists to help provide resolution or answer the question is really important. The other bit that I'd call out is like the elephant in the room is like the knowledge base. Because I think that's the big challenge to you, which is like, how do you train an AI agent on what human agents are saying in interactions today if your knowledge base is not like LOM friendly, right? It's like, okay, cool. Like if an agent can't reference it, if it's structured incorrectly, if it's a bunch of videos, it's a bunch of things that just isn't rag ready, like you're gonna have some challenges there. And so I think that's where I look at the systems in their critical path of the most common journeys. I'd also look at things like knowledge to help to make sure that you're focused on improving, enhancing, creating some like rag ready knowledge to leverage. And then obviously like the CCAS and the different platforms that are kind of the endpoints of customer interaction entry, uh, I think are the things that I'd be kind of focused on and just helping to make sure that, again, you're you're ready for a lot more volume than potentially if you have the best.

SPEAKER_00 22:41

I'll throw in bandwidth as well because that could be the limiting factor on all this, right? So start looking at your circuit utilization and you know, see how much cap space you have before you run into problems there. If you're operating a centrally located contact center at least. And then what about you know, staffing up, or is this the opportunity to start to figure out how AI can handle some of that call volume that you're expecting to see come in?

SPEAKER_01 23:05

Yeah, so this is definitely where I think it the future is gonna be AI agents and human agents handling call interactions at the same time, customer interactions at the same time. And I think that's why, I mean, at the the event that we did um at the CAVE which was awesome. That's why I think my mindset of like however fast you're thinking of your agenc strategy and putting AI agents in front of your customers is probably not fast enough because every single type of interaction that can be automated, that should be. And that like that's the reality is that you need to get there as fast as you can. But it's not a flip switch and all of a sudden, like, great, every WISMO calls now handled, every FAQ is is now handled. Like the it is gonna take a while to figure out does the information live in a system that can be accessed to answer the question that is frequently asked, yes or no? And then how do we also make sure that there's the change management with the consumers of when people call in, will they even trust your AI agent to handle the question yes or no? And so that to me is where like there's a number of different like long polls in the tent. It's just really important to help to make sure that you're thinking about having some type of AI agent strategy in place because there's no way you can just step up with the human agents. And I do think that is where, again, like they're gonna be handling harder, more difficult, more complex interactions and being able to just say, Oh, the agent went down that handles the easy calls, now kick all this over to the human agents as well is like probably not a good recipe for employee satisfaction and like not creating some attrition risk within the content zone.

SPEAKER_00 24:45

Do you think slight left turn here for a second? But I'm the type of person that if I have a problem, something's misbilled or whatever, if it's below a certain amount, I'm just gonna ignore it. I just I won't bother. You know, if I buy something from Amazon and it's you know 20 bucks and I have to go to the UPS store to return it, I might just not return it, right? I'm just I'm just that kind of guy. But in a world where I've got an AI agent that can go after these companies that you know I'm looking for that that refund or that credit or whatever it may be, that billing resolution, I'm more apt to do that because it's not taking time out of my day and it's not adding frustration to my to my experience of my day. So a lot of companies kind of build breakage into their models, right? It it is an appreciable part of their margins. Do you think that AI agents could potentially impact breakage and could that cause some underlying marger margin pressure?

SPEAKER_01 25:47

Yeah, totally, right? Like, and I think you're spot on with the vast majority of people, if there's a minor inconvenience, aren't gonna pick up the phone and make a call. If I could just ask Siri to do it on my behalf, tell me once it's done, and tell me how much money I've saved. Totally, totally would do it, right? And like I think you'll also get a lot of instances where it's like, okay, like I'm moving and I want to cancel the service because I'd rather use it. And then it's like, actually, we got it 50% off. You want to keep it? Yeah, sure. Okay, great. I didn't have to do anything and I just saved a bunch of money. And so, yeah, breakage is going to increase. I wonder as well if like maybe there's a different way of categorizing it, right? Where like there is gonna be revenue impacts 100% because of all these interactions that previously didn't happen. Uh, but then I also wonder if like maybe it becomes more difficult to allow maybe I create the process, which is if an AI agent calls in, my discount threshold is 10%. But if a human consumer calls in, my discount threshold is 50%. And so I wouldn't be surprised to see companies somewhat quickly adapt to try to reduce gaming the system, but the reality is definitely going to be like there will be instances where the system is gamed a little bit more. And I think at a minimum, there will be significant impacts. I think the the extent to which companies like reactive with updating policies and procedures and all of that based on like, is it a human calling or is it an AI agent calling? I think that will be like a whole nother ballgame, whole nother playbook that needs to get built out. And so there's some type of operating procedure setting expectations of like, hey, if someone's threatening to cancel, but it's not a human, it's their assistant. Okay, well, like maybe the discount, the discount threshold and the retention offers are significantly different than it would have been.

SPEAKER_00 27:42

That's really interesting. So play with the incentive structure or the disincentive structure, right?

SPEAKER_01 27:47

That's it. Yep.

SPEAKER_00 27:48

Yeah. So all these agents are going to be calling in. You're suggesting that people build out agentic AI workflows in their business and you know, spearheaded by an AI agent on the front end of it. So what are the risks of having an agent-to-agent interaction in that way?

SPEAKER_01 28:07

Yeah, I mean, I I think for one, and this kind of gets back into like having the right guardrails, proper testing, and e-vails, all of that becomes significantly more important because if you put the an agent the front end of your customer experience and says the wrong thing, like you're in a world of hurt. I think it was Air Canada, right? Like a year or two ago, offered a refund to a flight that wasn't eligible for a refund, they got sued and still had to give the refund. Like minor example, but you can imagine a much worse example that results in class action lawsuit against that organization. And so I think that's one, right, where it's not necessarily agent to agent, but the agent at the extension at the forefront of the customer experience, whether it's human agent or agent calling and interfacing with that agent, having the agent hardened and ready to go is going to be really important. And then I think from an agent to agent perspective, this is also where it just gets a little bit more interesting because, again, like that, the discount concept, right? Of like there should be different processes that exist for agent-to-agent interactions, and maybe it exists for human-to-agent interaction. And so I think how you define that, how you operationalize that, where those are happening. So again, maybe it's not happening through your call centers. Like the agent might not pick up the phone and initiate a call. Maybe it goes to your website, maybe there's a click-to-call button. Maybe you engage with it that way. I think the channel switching is also going to be very interesting, where maybe there are AI, there are agent-to-agent interactions that just happen on channels that human agents or human consumers can't get on, right? Like there's no transfer that would happen to a human agent. There's no transfer that would happen from my personal agent to myself. It's pure agent to agent. So I kind of think about it in those different areas of just kind of like, where is that going to take place? When it takes place, how do you make sure that your brand or enterprise is protected and hardened? And so that you're not giving out information that's going to result in some type of issue or impact to your business. But that to me is where kind of the agent-to-agent piece becomes really interesting. Cause I think it's just going to happen kind of in an avenue or venue that isn't seeing a lot of traction today just because there isn't necessarily the need to have that space ready. Yeah.

SPEAKER_00 30:25

You would think though that you would need to account for um the fact that a human may need to come into the conversation from either end at some point or maybe at both ends, right? So I'm not from the CX space, but I when I had a marketing agency, I had a client that was a BPO. And at that time, this was a number of years ago, but the the the trend was omnichannel, which I think is probably table stakes today, right? I don't even know if I've heard that word in in a decade. But you know, do you think that we're gonna have to relook at the idea of omnichannel from the perspective of being able to kind of zero out from an AI experience on both sides to be able to connect a human to a human?

SPEAKER_01 31:08

Yeah, definitely. And so, and that's why I think it's like there will probably be channels that are agent-to-agent only, and there's gonna be a bunch of channels that are hybrid in a mix of both. I think the other thing that's interesting, and we've at Crusto, we we've started to see a lot of excitement and success with something that we call agent operating center, which is essentially like you as a supervisor can be observing live interactions just purely handled by AI agents. And then what it will do is it will flag to you what we call like the virtual hand raise, which is at some point in the interaction, an AI agent might not be able to answer a question because it doesn't have the information, or uh there's a specific guardrail in place that doesn't allow it to give the 30% discount that the customer's asking for. It can only give the 25% discount. And so AOC actually allows for at that moment the EI agent to actually escalate to a human supervisor who's sitting there watching conversations. And then they actually are able to feed information directly to the AI agent to say, okay, I've approved a 30% discount. You can let the customer know that this is good to go and sorry for that issue that they'd run into, be really empathetic, was you give the response here. And so that's kind of the way in which I think this becomes way more important in the future, is you have this concept of like part of the reason that most contact centers, like everyone is there in person, and you have QA teams and you have managers and supervisors and all this structure supporting the agent. It's because it's just really hard for those agents to know everything all the time. But you also just can't have 100% experts all the time either. And so I think it's really interesting, kind of seeing more of the hybrid approach now with some of our customers, where it's like human agent interactions, AI agent interactions all happening at the same place. And then the same way it works with, hey, I'm gonna put you on hold for a second to go ask my boss how to answer this question or if we can do this. Same thing now is gonna be the case with AI agents. But again, there's a lot more volumes of how do you support that in a really scalable way? And then how do you redesign maybe even psychic contact center operations team to say, hey, if you're a supervisor and you're managing a team of 15 agents and maybe a 15 to one ratio of 15 agents per one supervisor, should that become 30 AI agents and human agents all reporting to that one supervisor and they're able to observe all of those interactions all at the same time? Possibly, right? I think there's probably gonna be some uh some change that's happening there. But same thing, right, is that I I think having good visibility and observability of AI agent interactions across the customer journey is gonna be mission critical. And back to the point of like omni channels, well, it's easy to scan emails and and calls, uh, excuse me, emails and chats and like written messages to see what's happening. It's much more difficult to do that on voice in real-time conversations. And so that's why I think it's gonna also effectively like redesign what the operating structure looks like in a contact center with QA teams, managers, supervisors, all of that, just needing to like virtually monitor the floor and walk the floor to see all the interactions that are happening today, whether they're human agents or AI agents handling this.

SPEAKER_00 34:32

So instead of calling in and saying, Can I speak to your supervisor? You're gonna say, Can I speak to your human in the loop? Yeah, yeah.

SPEAKER_01 34:39

Probably, right? At times you'll uh you'll definitely just say that, hey, like, and this is uh again, like the challenge with like the change management bit of like, I think people, most consumers will engage, and we see this all the time of like, you just talk differently knowing it's a robot than knowing it's a human. It's a little bit less empathy, a little bit more straight to the punch. You probably ask for things that otherwise you wouldn't have asked for. But we see the inverse, which is true as well. Of there are many companies who are selling products that maybe like are, I don't know, let's say like denture replacement. Okay, cool. Like maybe you're gonna ask a bunch of questions in that intake process that, like, I don't know, it's kind of sensitive. I don't want to talk about my teeth, all this stuff, just some random person sitting in the contact center somewhere. But maybe I'd share that with an AI agent because they're not gonna judge me and they're gonna be empathetic and like they're coming from a place of neutrality in their perspective of every single response that I have. It's just a machine capturing this information. So I think you'll see a little bit of both where um some people will play a bit of hardball. Other use cases might actually be a little bit better because it removes maybe some of the judgment and emotional aspects that people are less comfortable with invest.

SPEAKER_00 35:55

Yeah, just this week we put out an episode with the founder and CEO of uh Posh AI, and they operate in the financial fencer vertical. And that's what we were talking about was people are a little bit sensitive about their financial situation sometimes. And, you know, is AI a better alternative for those people because they'll feel a little bit of freedom from the shame that they might feel having that conversation with a person.

SPEAKER_01 36:18

Yeah. Yeah, I believe it, right? And I think this is also where it's still early innings, right? And I think as people get more comfortable, back to the companion bit as well of like, hey, if I'm spending a lot of time talking to AI, and there's probably a world in a year or two where like I just stand at my desk and I talk to Claude, and Claude's completely controlling my computer, where like now I'm so much more comfortable engaging and interacting with AI that fundamentally changes what I'm willing to say or do knowing that I'm speaking to an AI agent. I think the world's just gonna look fundamentally different here in the next couple of years.

SPEAKER_00 36:53

Yeah, I just saw a headline, I think, on Apple News this morning, actually, about how whispering in the office is becoming a thing because people are talking to their claws and their chat GPTs, and so there's like this low murmur and it's distracting, but it was paywalled, so I couldn't get through it. I'll probably ask Black for a summary of it. But anyway, so I know this is not your domain necessarily, but what about cybersecurity risks in particular? Because I would think that there's a risk of you know something like a uh a prompt injection attack or something like that. I mean, you're also greatly expanding surface areas.

SPEAKER_01 37:30

Yeah, so I think um I think of risks in in two different ways here. One is there is bad actors who are trying to take advantage of agentic solutions, AI agent calling in, prompt injections, jail breaking it, et cetera. I think for the most part, we've generally seen, broadly speaking, the space, presta plus every other company that's in the AI for customer service space right now, pretty rare and far view between examples of like anything actually being able to be leaked out. Like most companies are doing a pretty good job with guardrails, I would say. And like I think it's like somewhat hardened generally. I think uh it's a few months ago, maybe that like the Chipotle AI agent could just answer any question that people had. And so they were using it like ChatGPT, it was funny and blowing up on Twitter or X. I will never call it X, it's always Twitter me. So I think there's there's that one piece, which is like right, like finding the right partner, making sure that you're protecting your business because they have the right guardrails in place, but also like there's the input-output piece, which is like guardrails are going to be protecting on all the inputs, but how do you also protect on the outputs? That for us is what we're talking about, things like constraints. And so designing the agents in the right way and agentic workflows in the right way is super important because it's giving a lot of access to a lot of valuable customer information. And now, as you start to see financial services, things like healthcare, like all of those agentic solutions are gonna be tapping into all of the systems that have access to your money, all the systems have access to your health records, like very sensitive information. Um, so I think it will continue to be very important from just a just general cybersecurity perspective to make sure that you have the right partner supporting you there. The other piece is that I think what is what is going to be maybe more difficult for brands than protecting the systems that you're using from bad actors, it will be identifying who's a bad actor versus like who's using AI. And what I mean by that is like think about like again back to the personal assistant, bunch of calls coming in. Like, is that a legitimate call on behalf of a customer? Or is that some like random interaction of someone trying to impersonate me as the customer? And now that you have like the ability to clone voices with text-to-speech models and like it can sound just like me as well. Like that to me is where like the cybersecurity risk in terms of the customer experience in the contact center is like wildly changing. Because, like, how the hell do you even know? Like, is that a real person? Is that not a real person? Now, if it's someone who's saying that they're calling on their behalf and maybe they have some of their information, like, what is the willingness that we will do to share that information with them? So, that to me is where like, from a cybersecurity perspective, the two big risks are protecting all the systems that will get more and more access over time. And that's back to the point on just kind of like, are your systems on-prem or cloud-based? If they're cloud-based, do they have APIs? Like, all of that is just gonna open the business up to more risk. The other bit is the volume and the confusion over what is fraudulent versus what is not. It's gonna get quite murky here in the future. And so I think that's where figuring out like what is the right verification process and how do you make sure that it's hardened and foolproofed and things like that to protect to protect your consumers or customers will be really, really important. But I'm not sure if I'm like a large company, like how I get ready for that. I think there'll be a lot of learnings here for the next couple of months and quarters. But that to me is definitely going to be a space that cybersecurity companies will be thinking a lot about and then trying to help to make sure that like maybe back to your point on like omnichannel, it's like everything's two-factored, even agent-to-agent interactions, and maybe it's the human agent or the human consumer who has to like click a button in a text message to like validate that they did approve this agent to reach out on their behalf. I think the world will have to adopt a little bit because otherwise I don't know how there'll be enough trust in the ecosystem to perform all these actions that will be happening.

SPEAKER_00 41:41

Yeah, in marketing, we talk a lot about uh user experience design. I bring it up on the podcast all the time. People are probably sick of it because these are not marketing people listening to this. Pass this to your marketing people, though, please, so they'll download it. But you know, we talk about user experience design, and that's us designing a digital, you know, pathway for journey for people to land on your website and and find their way through and be able to accomplish what they want to accomplish and get them to nudge them to do the things that you want them to do on the website. And one concept that I think is overlooked a lot in user experience design is friction, because it is not intuitive to think that you would intentionally introduce friction into the process because so much of user experience design is about removing friction so that people can have a clear path to what they're trying to accomplish. But intentional friction can actually play a really important part in getting people to take actions and to delivering a good experience. And this might be a place where we necessarily have to introduce friction into the process so that it's not a free-for-all, and you can't have one person with, you know, a hundred agents banging on the door of a single brand trying to get something done or or um you know, just blind outreach. And and if you could create a step in that process around authentication, uh you kind of kill two birds with one stone, right? Is you're you're blocking and tackling some of the some of the junk traffic that really you don't need to deal with and you and maybe is you know not real, I'll say, right? Like maybe it's just people trying to trying to get a discount because they want a discount, right? There's there's plenty of that. But also you can deal with the step of making sure that it is a valid request.

SPEAKER_01 43:24

Yeah. So it's button on, right? I think authentication will fundamentally change here in the future. And it's been interesting, but I think most from a consumer perspective, most of the times when I'm thinking about like logging into anything and get the text message, here's a six digit code, and and put the code good to go. Like, I think there'll probably be the proactive equivalent too, which is okay, great. If I'm trying to do a thing, now I have to validate that like yes, I'm actually approving and wanting to do that thing. And so I think we'll start to see kind of like a proactive authentication in addition to like the reactive authentication that.

SPEAKER_00 44:02

Yeah, I could see how, you know, uh you have the pass key on um the iPhone, right? And it just uses your biometrics to approve things so you don't put a password in and it replaces the function of entering a password in to an application or to a website. I could see how you could get a notification on your iPhone that says your AI agent needs approval, and it you has a pass key and it authenticates on your face, and you just tap the button until it you know complete the transaction.

SPEAKER_01 44:30

Pass keys, one of the greatest inventions. The first of which are like from an authentication perspective is like person who like created the autofill for the one-time code. Like yes, brilliant.

SPEAKER_00 44:40

Brilliant. It's these little things, right? Little things. Little things. So on the topic of risk, the third area of risk, I think, is brand risk. So talk to me about how this introduces the potential to do damage to the brand, or uh on the positive side, to actually improve the perception of the brand.

SPEAKER_01 45:01

Yeah. I mean, the reality is that like I think most companies will view this as an opportunity to create an identity for a brand that is communicated, distributed through an AI agent. Right. So if I pick up the phone and I want to call an airline, for instance, like the sound, the experience, all of that, like the marketing team should actually be very engaged in the process of building AI agents for customer service because it literally just becomes an extension of the brand identity. It's the sound, it's the feel, it's the personality, it's the characteristics, is is it funny and youthful? Is it serious and a little bit less excitable? Is it male? Is it female? There's so many different characteristics that go into the actual design of the personality for an AI agent. That to me is where to your point, like, and both both things are true, which is there's a a lot of an opportunity to lean in to the design process and use this as an opportunity so that everyone understands and knows what it sounds like. Right? It's like the uh who is the uh the the girl from Progressive Eds, right? Like you you know who that is, Flow username? Right. And so like everyone knows Flo, knows the voice, what she looks like, all of that. It's become an extension of the brand. With that said, it also presents the same risk, which is like it's not good, it breaks, it doesn't give the information, it's no different than what the experience was with an IVR. People lose trust in it, they don't like it, what have you. Then all of a sudden it's it's detrimental to the brand. And now you've got a loss in customer uh trust, impacts the customer uh experience, CSAT, et cetera. And so that to me is where I think most companies right now are realizing that like they should lean in, they should do it the right way. And if they do, it creates a huge opportunity for them. The inverse can also be very true, too.

SPEAKER_00 47:04

Yeah, a famous line in marketing is Marshall McLuhan saying that the medium is the message. And what that means is where your brand exists creates a context for your brand. And so, like if you're a luxury watch company and you're sponsoring Formula One, that says something about your brand, right? Just being at the Formula One race or in that context. And uh in today, like when I go through a checkout process, if no matter what I'm doing, if I'm on my phone and it doesn't have Apple Pay integration, I kind of feel like that brand is a little bit behind. And I get a little bit of a bad taste. Not enough that I'm not gonna purchase from that brand or anything, right? But that's unintended and unwanted friction that could slow me down or stop me from uh completing that transaction. So I think just having agentic AI and a way for your personal agent to communicate with a brand is going to be the medium that sends the message that you know you're a tech forward uh business who is looking to meet their customers where they're at. Totally.

SPEAKER_01 48:12

Yeah. You're either proactive, forward-thinking, innovative, taking advantage of it, or you don't have that same experience and call it's clunky, you're gonna lose consumers, they're not gonna want to engage with you 100%. Like, I think you've really got to lean into it, otherwise you'll quickly fall off. And good analogy, right? Was Apple Pay of like it can either be frictionless or it can be frictionful, and you got to choose what path you're going down and get started on that ASAP.

SPEAKER_00 48:42

Yeah, so I think you have to be intentional about the the technology decisions that you're making and the journey that you're creating. And and you know, you made this point, but be intentional about how the brand is showing up in those places because being a brand is really the sum of all of the interaction points that people have with the company in all the different mediums that you exist, whether that's the human beings in your company that they're dealing with on a day-to-day basis, whether it's your website, your social media presence, print ads, billboards, all these things, uh, they all equal how people perceive you. And that is, you know, from an outward perspective, what a brand is. So I think being really intentional about that, and and you made a good point about marketing being involved with customer service. I'm seeing that more that CMOs are having conversations about customer experience and they're getting involved in some of the tech stack decisions as it relates to the contact center. What what are you seeing on your side? Like, is the the marketing persona becoming a buying persona for you or at least a decision-making uh team?

SPEAKER_01 49:45

Yeah, they're definitely part of the buying committee. They are heavily influencing the design and considerations. It's funny, like when you think of like agentix solutions, it is difficult to both evaluate and also to sell them because there's so many different stakeholders, right? Like you've got the CFO who's looking at the numbers to understand well, what is the business case? And is there a justifiable ROI here for us to make this investment? You have IT who's looking at it through the lens of like, oh my gosh, how much work are we gonna have to do here to support this? And how do we get this integrated to all of our systems and what systems have APIs and don't that we need to build? How do we get this integrated into our release process? So do we need to integrate into all of our dev environments, or is it just going straight to fraud? How do we reduce the amount of risk there? You have marketing who again is looking at it through the lens of like, well, how is this gonna make our customers feel? What does it sound like? How does it act? What's the personality, et cetera? Then you have like the operations team who is like, okay, great, who is this disrupting? How is this impact them and so forth? And so the buying committee is quite large, I think, in general, with Agentix Solutions. But because there's so many different stakeholders here who all have a significant amount of influence over the decision, just becomes really important because also it touches so many different parts of the business. That that's the beauty of a gentix solutions, too, is that like they're meaningful and they're important, and they do actually drive a lot of productivity. But with all of that productivity also comes a lot of considerations and making sure that you're doing the right way, everyone signed off broad consensus, and it's kind of the right path forward at the right time with the right use cases.

SPEAKER_00 51:27

Yeah, and I the last thing that I'll say on brand uh to spare the listeners uh from my marketing junk, but uh it should probably be present in governance as well. It should show up in your policy, and you use the Air Canada example. The the cost for giving that person the credit for the flight that was supposed to be free was probably like 500 bucks. The cost for the legal fees was probably tens of thousands, if not hundreds of thousands of dollars. But the brand damage, they they became the people that didn't stand by what they said on their website. There's a loss of trust and and a loss of brand equity when you do that. And on top of that, you come across like an asshole because you're you're picking a fight with someone. It was a bereavement flight. That that was what it was. So it's not even like someone just asked, you know, for a credit or something like that. No, somebody called up and said, My uncle's dead. I need to fly. You know, does what is this gonna cost, or can it be free, or do I get a portion of it back? And they they answered wrong. So you don't want to come across that way. You you want your brand to look like you you stand by, you know, what you say you're gonna do and that you deliver for your customers, and that you're gonna keep that trust because trust is one of the most important things in in building brand equity. It's your spokesperson, right?

SPEAKER_01 52:46

Like it's it's that. And so I think how you coach, how you design it, how you develop it in the right way to be the face of your brand is is mission critical.

SPEAKER_00 52:54

Yeah. Um, so I'm gonna take uh a final left turn here and and talk about the cost of all this because if we're in a world where there's a bunch of agents that are trying to engage with agents on your side, there's gonna be a whole lot of token use that you didn't account for. And I I saw a meme not too long ago on Instagram because I'm old and I use Instagram and not TikTok. But uh, you know, it was it was like the the face of the CFO when he realizes that the costs of the tokens are more than the salaries of the people they let go, right? And I think that's a real thing. I think that we're already seeing some instability or fluidity in the the underlying cost structures of some of these models, and that's most likely going to continue. And who knows what it's gonna look like. But uh, regardless of how it goes, whether it becomes cheaper or whether it becomes costlier, it's the volume that's gonna end up uh increasing the cost. So how do you project that? How do you deal with that?

SPEAKER_01 53:53

Yeah, I mean, uh it's a bit of a a moving target in a number of different ways. So for one, it's interesting because like the inference cost has gone down significantly, but the models have improved even faster. And back to the point of like most people are transitioning from answers to actions, like what they're actually doing with AI significantly more complex than it was six to twelve months ago. And so you've kind of created this like interesting paradox, which is like usage and productivity is outpacing the cost efficiency being realized over time. And I think that's where many organizations, and maybe even back to like to Sam Altman's point on like, hey, maybe this is not causing as much white-collar job loss as we expected, is like, it's becoming pretty expensive to use AI. And that's partly just because how people are using it, it's just fundamentally different. Like 12 months ago, you probably weren't building any decks with AI. Now, like Claude's making pretty much every deck that is being put together in tons of what one pagers and material that just wasn't being produced previously. It was like, hey, I'd go to Claude to ask a question, get some information answered, et cetera. Uh, maybe like summarize a bunch of a bunch of text that I don't want to dig through. Um, so I think you're you're at this interesting point where like usage is becoming more complex with AI, but also because usage is becoming more complex, you're using the smarter, more expensive models. And so even if the underlying model costs are going down, like the rate of complexity uh is going up. So like token usage is increasing more than the cost per token price is decreasing. So that's a challenge. Many CFOs are gonna have to deal with. Um, if we just only charged when resolution was provided, when the task was completed, et cetera, it makes it a lot easier to make sure that you're only paying for the compute and token consumption that is warranted because it saved the conversation from happening. But the challenge is that we're starting to see there too, is that then it just creates a bunch of concern over like what is the definition of an outcome and what satisfies when that outcome is being met versus not met. It also creates some disincentivization on leveraging the best models because like to have the strongest outcome, or excuse me, have the strongest margin with an outcome-based model, you obviously want to use like the cheapest models possible to provide that action. And so long-winded way of saying that like there's no shortage of complexity around AI cost and usage, I think we'll continue to see spend increase in the world of AI cost as complexity and the tests increase over time. The reality though is I think that's where like for us, it's been very easy to do things like volume-based pricing instead, where it's like, hey, if you know you got a million minutes a month of calls, okay, cool. Like I kind of know what to expect in terms of what those costs could be, and potentially those costs will go down over time because what we're doing is leveraging models that should be coming more efficient. It's much more difficult when, again, kind of back into like more of the knowledge worker and producing materials and running tasks and building reports and web pages and applications, et cetera, like the complexity of that is increasing, token consumption is going to increase, costs are gonna continue to increase. And so there'll be a delicate balance there. I think many CFOs will be dealing with of how do we continue to invest in AI, but how do we not create some type of like AI sprawl in terms of what those costs mean to the business? And then how do we also not invert the ratio so it's like, well, damn, all the humans are actually way cheaper than what all of the costs are that we're spending on AI today.

SPEAKER_00 57:59

Yeah, it's gonna be obviously tough to predict, but people should start having those conversations with their CFO now, right? Totally.

SPEAKER_01 58:07

Yeah. I mean, you gotta start to game plan for it. There's also, again, like the reality is that AI is not going anywhere. People are gonna continue to use it, the complexity of it, how they use it is gonna continue to increase. And so how to get ahead of it, make sure that there's a strategy for it, I think is important. We also use it a lot too, where there's also this risk, which is like everyone is doing these highly complex tasks using AI in sidelines. And so, like, there's lost efficiency, I think, right now. And like, there's also not a lot of incentive of like anthropic or open AI to just be like, hey, instead of everyone building their own agents to all do the same things in probably some inefficient ways, just have one centralize it, have everyone use it, drive more efficiency, reduce cost. And so I also think like that will start to change as well, where how large enterprises are leveraging AI is gonna be hopefully a little bit more efficient than potentially it's been in the past.

SPEAKER_00 59:03

Yeah, and it goes back to something we said earlier, too, which is don't introduce AI unless it's the right solution that's right, right? That's it.

unknown 59:12

Yeah.

SPEAKER_00 59:12

Yep. I know I said it was our last left term, but I did wanna you you kind of when you said that AI is not going anywhere, um, it made me think of all these graduation speeches that people are getting booed at. So why why are they booing, do you think? Booing in the sense of about AI or Yeah. Every time AI was brought up in a a graduation speech, yeah.

SPEAKER_01 59:34

Yeah, I mean, I think so the reality is a few things. One is that with midterm elections coming up, I think data centers are probably like first and foremost a thing that is highly political and contentious at the moment. And there's some reality to that. I think like when you think of AI today. So if I were to ask in Google, like, how do I reset a Sono speaker? Simple question. If I were to ask that same question to Chat GPT, it's 10 times as much energy to produce that response. And so I think the energy consumption is crazy. Same thing of like, if I ask uh Claude to produce like a hundred-word email, it's like a bottle of water is used just on that one task. And so I think a lot of people see the impacts in terms of like energy consumption and water usage, et cetera, and data centers again being a very political thing at the moment. And so there's no shortage of issues, I think, facing the world of AI, just in the in the sense of it is going to be compute and resource intensive. You also have a lot of companies like Meta is a good example of like they get the vast majority of their uh data centers powered through renewable, and so like it's not that bad at the end of the day, but I think the reality too is that like I've not seen anyone not use AI because it uses more power and it consumes more water. And so, yes, I think there's some frustration. I think a lot of the frustration, maybe it's not on like the environmental side as much as it is on like the white-collar job risk side and a lot of kind of the fear-mongering that has existed within the industry of like staking everyone's jobs, no one's gonna have jobs. Then you have St. Motlin today saying, like, maybe that's not quite the case and that's not happening at the rate that we think it would. And so, insofar as I think we continue to be cognizant of all of those things, there's all important things for us to address and answer and find good solutions for. But I think in the next probably six to 12 months, like if what Sam was saying today like continues to be the case where there's not this widespread job loss that potentially was predicted six months ago, I think the public perception, uh perception of AI is going to be fundamentally different. And also, again, like the applications of utility, the productivity that I gain from leveraging AI is that increases as well. I think so well satisfaction sentiment with it in general.

SPEAKER_00 1:02:05

Yeah, it's so funny because the the industry itself are the ones that did all the fear-mongering. I haven't talked to a single person in this industry, either on the provider side or on the enterprise uh consumer side, that has said, yeah, this is gonna take all the jobs away. Not a single person has ever said that. Every single person has said this is overstated. That's not what it's gonna be like. AI is gonna be working alongside humans. And yes, there will be some job loss, but those jobs are gonna shift to other things. I mean, you talked about a couple of things today, right? Having a, you know, someone that could have been a frontline agent can now be that supervisor that sits over top of 30 AI agents, right? So it's just a different skill set that you have to train them on. So I I feel like they they kind of created this environment for themselves of you know people being angry with them and and not liking AI. But to your point, like I don't think it's going to move the needle at all in terms of the expansion of AI. It's it's just another, like you say, we have to address it, right? We should be investing in renewable energy and have uh you know responsible ways to and sustainable ways to to fuel this. But you know, almonds use a lot of water in California, but you know, there's not a mass movement against almonds. So I don't know. I just thought it was interesting. And and you know, I I always want to look at it like consumer culture drives enterprise adoption to some extent, I think, right? Like the consumer market has to be ready for the technology that we're introducing if it's a technology that interfaces with them. And I think that's one of the reasons that you're seeing so much enterprise adoption is that the chat GPTs of the world and the clouds of the world, people got used to it. And so they've become familiar with AI and they're comfortable with AI, and they're able to get past that little voice in the back of their head that says, This is new and weird, and I don't want to do it.

SPEAKER_01 1:03:59

Yeah. Yeah. I completely agree. Like, I don't think we are where we are today, if not for Chat GPT, and I think it was late 22 that it came out. And then I also agree with just like the humorism is it's also part of like anthropic being in a perpetual state of fundraising. Yeah, it's it's like driving capital. It's all sales. Like it's all it's all sales at at the end of the day. And so yeah, it's a very interesting time. I think the world's gonna change. And maybe I'm techno optimist here, but change for the better in terms of what people will be able to do leveraging this new technology in the future. But it's disruptive, like any new technology. And so I think it'll be a lot of fun. And hopefully some of these predictions play out, but probably some of them won't as well. But uh, who knows? I don't think five years ago I thought we'd be on a podcast talking about AI and how it's going to revolutionize the customer experience. But uh, but here we are.

SPEAKER_00 1:04:53

Yeah. Um, well, I don't think five years ago anyone thought I needed to be on a podcast, and they probably still don't, but let's end on an optimistic note. Brandon Teagan, thank you so much for your time and expertise. Appreciate it. Thanks, Sean. Happy to be here.