From Alexa to Agentic AI: Conversational AI and the Future of Brand Communication - with Susan and Scott Westwater

1
00:00:00,240 --> 00:00:55,782
Daniel Nestle: Welcome or welcome back to the trending Communicator. I'm your host, Dan Nestle. I've always thought of myself as an embracer. I embrace new food. I embrace it. New gadgets. Embrace it. New tech. Absolutely embrace it. But if I'm being honest, I haven't been as embracing as I've puffed myself up to be. I mean, I have a podcast, so why aren't I doing video? And I'm all over AI, but why am I simply paying lip service to voice chat? Maybe it's because I've always been a snob for the written word, or maybe it's because I'm having trouble figuring out if voice will be additive to what I do. And it's hard to conceptualize how I connect voice to with everything else I do. Now I will say I've begun to dabble.

2
00:00:55,846 --> 00:01:45,130
Daniel Nestle: And voice chat, let's call it a part of conversational AI is far superior than what I had imagined. And I'm glad I didn't try it out six months or a year ago, you know, so much so that now I'm ready to dive in and learn all about it. So lucky for me and for all of us, I've met today's guests a few months ago and have been dying to have them on the show because they've been all in on conversational AI since the BC before ChatGPT days. Together they found and run Pragmatic Digital, a strategic marketing agency that helps small and medium businesses leverage cutting edge technology for tangible business results. They're doing what I aspire to do. They're making AI accessible and practical for their customers, guiding them toward a demystified and productive AI assisted future.

3
00:01:45,710 --> 00:02:18,302
Daniel Nestle: They're strategists, they're technologists, teachers, public speakers, authors of Voice Creating Useful and Usable Voice Experiences and more recently, voice marketing, harnessing the power of conversational AI to drive customer engagement. It's both a pleasure and an honor to have them on the show. And I keep saying them because clearly it's more than one person. The AI power couple themselves in the flesh. Susan and Scot Westwater. Hello guys, how are you?

4
00:02:18,326 --> 00:02:19,770
Scot Westwater: Thanks for having us. Great.

5
00:02:20,230 --> 00:03:16,900
Daniel Nestle: It is good to see you. And you know, for our listeners benefit. We met at Macon, the Marketing AI conference in back in September, I guess it was in beautiful Cleveland, Ohio, which was beautiful those days actually in September, you know, and I have to say like that was my first exposure to this world of, you know, people who are making their living with or appended to generative AI and AI in general, you know, and I didn't know what to expect because you know, I had recently left the corporate world and been really getting into AI. But now that I'm on my own, I'm looking for my peers and people that I really connect with. And you know, I think that like it was an amazing conference. Hats off to Paul Reitzer and Mike Caput and Catherine and everybody there.

6
00:03:19,200 --> 00:04:00,654
Daniel Nestle: But finding your tribe, finding the people that you know, that you really kind of like, holy crap, I can learn a lot from these folks and you know, potentially partner with and you know, boy, oh boy, what a positive multiplier of my understanding this is going to be. Susan and Scott are among those people and those, you know, that I met, that I feel that way about. And little quick shout out to Dave Armano who told me to look out for Scott and Susan. I mean, you know, it's a big community but a tight knit one at the same time. That's too much preamble. I just want to just thank you really for being here. And you know, let me, let's just start off by like, hey, who are you guys? And you know, you are.

7
00:04:00,742 --> 00:04:26,310
Daniel Nestle: I called you the power couple of AI because I mean for a couple reasons. First of all, I think it's true. Second of all, you're the only couple of married couple doing this that I know. But how did this all happen? You know, you have a long career from various directions, you know, and you decided to go all in on this stuff, you know, prior to it being du jour or whatever it is. So what's the story?

8
00:04:27,170 --> 00:04:28,266
Susan Westwater: I guess I'll go first.

9
00:04:28,338 --> 00:04:29,738
Daniel Nestle: Go ahead, Susan, you go first because.

10
00:04:29,794 --> 00:05:19,428
Susan Westwater: Well, the story is very different for the two of us, which is kind of the fun thing. And that's one of the things of. I do like to point out that we're still married after doing this for seven years now or eight years. Eight years now. Eight years now because yeah, if you'd have told me that I was going to get married, have a family and work with my spouse and found a company, I probably would have told you that was completely the wrong person who you were doing the reading for. But here we are. Essentially we both came from a career of. My career actually is started traditional if we call it that anymore. Gosh, if there's bc I guess it's downright prehistoric of pre web and being a college student who lucked into working at a small agency, promotional marketing Advertising.

11
00:05:19,524 --> 00:06:07,536
Susan Westwater: And going down this path, I had this delusion of grandeur. I'd be a creative. And I say delusion because after I met some incredible professionals, copywriters, I learned that coming up with five headlines about cocoa mix that didn't sound completely horrible was impressive. Like, I was sort of like, I can write plays, I can do a lot of things. Obviously I can write books. That is not something I can do. So I quickly had to say, all right, well, now what do I want to be when I grow up? So I went down this wonderful path of working with some incredible mentors and just working my way up through essentially going along this whole path of learning about proper marketing, brand strategy, promotions, all of those things. And then when digital happened, being like, this is interesting, let me keep doing this.

12
00:06:07,608 --> 00:06:49,000
Susan Westwater: So I would say it's been a story of just curiosity and research and resourcefulness that have come through. I then got the opportunity, over time, you kind of start to say, all right, what else is beyond there? I actually went to the corporate side for a while and really doubled down on content strategy, headed up a content team, learned all the fun things about practicing in corporate as opposed to agency side, where quite frankly, I think kind of pirates at times getting to do those things. But seeing all of these things work. Now, the funny thing is my love of voice actually comes from. And this is where I know I'm precocious, but this is where my double English major shows up is. Is one of the most.

13
00:06:49,380 --> 00:07:31,994
Susan Westwater: I'm a few credits short of being a medievalist and I always keep needing to go do I always want to go back? But what I studied very much for was in Renaissance Florence vernacular and along those things. Now one of the interesting things about during that time is we always think about the art, but we don't think about the writing and the language. We went from a time of where basically spoken word was the only way of news because only the elites knew how to read especially. And then once the printing press happened, still it was for the elites. It was in Latin and one of my most favorite books. And I always. I cringe when I say it, but it's the truth. My favorite books is the Divine Comedy because it is in Vernacula. It's one of the first printed.

14
00:07:32,122 --> 00:08:11,086
Susan Westwater: There's a million reasons why. I mean, having been in Florence and seeing where, you know, the launching of a thousand ships and all that nerdiness, I won't bore anyone with any more of that. The whole thing of it was though, it was amazing. To see this trend of where were in a digital world where everything was still in one language. It was literacy, there's accessibility, all of those things. As we started to see this trend all the way back in 2017, even with Alexa and even with the Google Assistant and things spoken word suddenly became this way of being able to transfer information in a much more democratic method. You didn't have to be able to read. There were opportunities to put things in language.

15
00:08:11,278 --> 00:08:51,760
Susan Westwater: It was starting to get back to the roots again of how information was kind of pulling down some of the gatekeeping. And I really geeked out about that, of how we do that flow of written and spoken. And then also the whole idea of the nuances, which I'm sure we'll talk about today, of what happens when your brand starts speaking, when your brand actually has to have a real personality, not just a really neat thing that's in a SharePoint drive somewhere. These were all really interesting, rich experiences that come into play, which is what really geeked me out. Now, Scott was a lot. I would say I was more of the heart and emotion of the possible. I would say Scott was much more of the rational nut to put to.

16
00:08:52,140 --> 00:09:38,480
Scot Westwater: Well, I mean, my background is in visual design and user experience. So back around 2015, 2016, a lot of people are talking about Amazon, Alexa, Google Assistant, Siri. I started doing research and started digging into how people are actually interacting with these devices, because, as sue mentioned, our careers predate the Web. So we saw the advent of the Web and trying to convince brands that the Internet wasn't a fad. As a matter of fact, I worked with David Armano at one point and had a creative director tell me, focus on print. That's gonna be your bread and butter. This Internet thing's a fad. Granted, 20 years ago, the world was a very different place, but kind of seeing that trend and embracing the technology, went through this, like, three times in our career, between the web and then social and then mobile.

17
00:09:38,640 --> 00:10:23,452
Scot Westwater: So here was this other moment where it was like, yeah, this definitely seems like something that people are going to adopt more and more. It reduces friction. It removes literacy out of the equation. You don't have to know how to use a computer. Literally, as long as you can ask a question, you can get some sort of response. So that's when we really started focusing in on, all right, what is conversational AI? How can it actually help solve real business challenges? Whether it's marketing or comms or customer support and service sales, there's a lot of different applications of the technology. And the other thing that people tend to forget is chatbots are part of conversational, voice assistants are part of it. Voice Mode and ChatGPT, they're all different flavors of it.

18
00:10:23,476 --> 00:10:55,044
Scot Westwater: You're physically using your voice to ask a question and getting some sort of response from a system. So we just saw it as a force multiplier and making it easier for people to access information not only internally within an organization, but language support and being able to have content in multiple languages is now a possibility. Five, seven years ago, it was almost impossible unless you had unlimited budget. So from an accessibility standpoint, it makes a lot of sense. From a doing the right thing standpoint, it makes a lot of sense. And from a business standpoint, it makes a lot of sense.

19
00:10:55,142 --> 00:11:51,668
Daniel Nestle: Oh, for sure. And, and you know, just hearing the, the litany of the kind of high level uses that you've just talked about and you know, and I've been there with you, like, I understand, like I've gone through the same transitions, the same process. A little closer to Sue, I think, because, you know, I started off in, in a kind of, you know, getting into copywriting and then went in from there into corporate and had a long corporate career and you know, PR comms and marketing, a lot of B2B stuff for many years, and then kind of, you know, wounded my way around, around the world and back into agency, then back into corporate, you know, how it goes. Right. But the whole time just kind of looking at these, you know, transitional or revolutionary technologies and dealing with that something.

20
00:11:51,724 --> 00:12:45,692
Daniel Nestle: I think, by the way, that anybody who's not Gen X and you know, maybe boomers, I suppose, although, you know, debatable, but anybody who's not really Gen X or very, very early millennials, I mean, you know, they're never going to experience anything like this, like these kinds of like. Why don't you get super excited about this transition? I've seen it before, you know. Yeah, this time I'm really excited though, about AI. But really what I wanted to ask you though is like you kind of lead into my first big question because I think folks think conversational AI. Oh yeah, that's Alexa, you know, that's Siri, and now of course with voice on ChatGPT. I don't know how many people are taking advantage of that, but can you just lay out the conversational AI ecosystem?

21
00:12:45,756 --> 00:12:57,680
Daniel Nestle: What does that look like and what are the areas that we should be looking at in our world as creators, communicators and Marketers.

22
00:12:58,340 --> 00:13:38,796
Scot Westwater: Well, there is Amazon, Alexa. Google Assistant is still there. Samsung has Bixby, you've got Siri from Apple. So those are the known entities. But I think the bigger opportunity most people don't think about right now is voice enabling your mobile app. And so that is actually a fairly well, depending on what you're trying to do. I will couch it with that. But depending on what you're trying to do, it can be a fairly easy lift to add to your existing infrastructure. There are platforms that exist that allow you to add almost like a voice mode on top. And the idea there is you don't have to worry about navigation, you don't have to worry about knowing where in the app a certain feature or, you know, the.

23
00:13:38,828 --> 00:14:16,500
Scot Westwater: If you're trying to do tracking or whatever it is, you can literally just speak it into existence. So that's a huge untapped area that we have seen some brands having success in having virtual avatars, where it's like a virtual influencer, where you've got their likeness, you've got their voice and being able to interact. So you could do that from an influencer standpoint, but also from a customer service and support standpoint. There's a lot of those types of applications that people aren't necessarily thinking about either. But Susan's been involved. I was initially involved, but Susan's been heavily involved the past couple of years in some consumer research.

24
00:14:17,120 --> 00:14:39,870
Scot Westwater: We have data that shows how people are using Amazon Alexa, for instance, and the types of things that they're looking for, what they want to be able to do in the future, and start to look at a lot of the potential use cases. I will do a little spoiler alert. It's not too dissimilar from what they're trying to do in other channels. They just see voice as an easier mode in a lot of cases, and so they want to continue with it.

25
00:14:39,910 --> 00:15:27,010
Susan Westwater: But yeah, and I think that's definitely with the ecosystem. You just actually sparked a thought about what we're seeing in the AI world. And it's actually this funky time of where we have. We actually created a diagram once of it, although I will have to dust it off because it's a year old and I feel like anything over six months at this point is dated. Yeah, it's the whole idea of voice as a product. So that is your Alexa skills, that's your Alexa games, all those novelties, those things where you can monetize it right then and there. And then we have this whole idea of voice as part of a brand experience or conversational AI as we think about that and that whole thought of, here's how I can use it to answer questions easily at any time.

26
00:15:27,130 --> 00:16:17,440
Susan Westwater: Here's how I can be able to plus up how someone's unboxing. I can enhance an experience, I can give accessibility to celebrities or my own brand spokesperson or spokes icon, any of those things. These are all ways that we can make it so that you can kind of that whole idea of reach out and touch someone, you can actually reach out and talk to them. And that creates an incredibly rich experience. When we think about customer experience, especially when we think of, you know, it's five years ago now almost to the day, and a couple months went from being able to do stuff at shelf, we've been able to do stuff in store on premise that all disappeared. Well, then what happens? How do I build that brand? And what's really kind of interesting about Voice is a moment where to engage with it.

27
00:16:17,480 --> 00:16:58,356
Susan Westwater: You have to initiate it. It's not disruption, it's invited. And it gives us this opportunity to have that conversation. If we do well, if we don't do it well, then it's monologues and it's annoying and people just shut it off. But there are ways of having that back and forth and engagement that actually allows for this very rich experience where then we do win a heart. Not just. And we kind of move from the whole idea of I'm not just a provider, I'm an advisor or I'm that expert or I'm that guide or whomever I'm supposed to be, who I want that personality to be. Which is why we have to be intentional then, by the way, about who it is we want our audiences to speak with.

28
00:16:58,508 --> 00:17:40,462
Susan Westwater: And that can work from a B2B perspective just as much as it can work from an external customer engagement perspective. Because we're able to kind of COVID off on these two things. And while, yes, these products are really cool and I think about all of the AI building that's happening around us, the one thing that I always want to tell people is I am not asking you, when you start working with us, that you're going to have to learn how to build these things. That's why you're hiring experts, right? That's why you need like, quite frankly, if you knew how to do it yourself, you probably wouldn't be calling me, you'd be reading our books. That's what I like to think anyhow. But the reality of it is we're thinking, how do we make this an easier way to engage?

29
00:17:40,606 --> 00:18:16,588
Susan Westwater: How do we make it so that we're covering off on the important parts and then how we make that happen. Because these things all have to be consistent. Because just as no one, you know, we learn this through the web and social and all of those things. And by the way, social is another place where conversational happens. Their conversational experiences. Sometimes it's a back and forth, sometimes it's through the bot, but they're all things that are happening in front of everyone. It's just a public conversation as opposed to the one ones. Yeah, all of those things happen. We have to think about being consistent then. And is that the same person, you know, or same personality and those particular things?

30
00:18:16,724 --> 00:19:14,882
Daniel Nestle: Yeah, the idea of assigning personality to a bot. Right. To essentially, you know, your voice, your interlocutor in whatever, I guess, medium that you're choosing. You know, it's hard enough to choose a logo or a, you know, I don't know how many discussions you guys have been in where it's like, you know, you're trying to pick that. What do they call it, the, the animated logo or the sound logo for a company that goes at the beginning of your video, it's like literally one and a half seconds long and you hear 40 different versions of dissonant chimes and you're like 36, you know, like, yeah, there's no science in it, but it has to represent the brand in some way with voice. I think it's in some ways probably a lot harder to kind of get that mascot or that avatar decided on.

31
00:19:14,906 --> 00:19:48,218
Daniel Nestle: Maybe it depends on the size of the company and however many fingers they have and however many pies. But I can just hear the discussions in the pr, you know, in the weekly PR meeting, which is like, well, we have a new voice avatar, but, you know, he or she needs to sound less masculine, more feminine, or needs to be from. Needs to sound vaguely Singaporean or, you know, something. Something like this. So, you know, I guess the good news is that you can make that happen. But to decide that is. Is a, is a effort.

32
00:19:48,394 --> 00:20:33,388
Scot Westwater: Well, the good thing is we have in our back pocket. I mean, I used to work on commercials, so like casting voice talent for or specific spots. It's just something you do as part of that process. Right. Well, that is still relatively new to the AI space. So there's kind of going back to the way we used to do stuff and basically coach the voice talent and say, I need it. To be performed this way. Now, you can either take an actual human actor or actress and digitize their voice, or you can actually create a synthetic voice that's wholly ownable and just kind of a standalone. But to your earlier point, like, voice and tone now become literal. Like, is it an American accent? Is it British? Is it Australian, Is it male? Is it female? Is it urban, suburban?

33
00:20:33,484 --> 00:20:41,116
Scot Westwater: Like, all of these things are now part of the consideration set, which ultimately needs to ladder back to your brand attributes and what you stand for.

34
00:20:41,268 --> 00:21:19,546
Susan Westwater: Yeah. And the interesting thing is that gets to be a pretty. You have to be my favorite. I guess my word of the year is going to be intentional of where folks are kind of like, oh, I didn't think about that. And we've done enough of this that we know what questions to ask to start thinking about it. Because it does become those things that when you think about, what is it? When you react to a person like, how will I. What is this relationship to begin with? Am I the expert that you're going to. Am I a servant? Am I. Are these. All these different things? Am I a colleague? Or am I that friend? And, you know, for one particular brand that were working with, it was, oh, is.

35
00:21:19,618 --> 00:21:50,200
Susan Westwater: Is this that know it all friend who does everything better than you can? Is that who we're talking about? Or are we. Is that too off putting? Who do we want that to be? And other times it is like when we talk about gender. And there's been so much studying that's already been done on all of this. I actually had a presentation a couple years back from someone who was working on how do we create a non binary voice, taking the best aspects of male voices and the best aspects of female voices. And by best, I mean the things that resonate, the things we hear.

36
00:21:50,240 --> 00:22:28,618
Susan Westwater: Because when you start going down the path of why people react to voice, we have evolutionary things such as we are descended from the person that being that said, when mom said, get back in the house or get back in the cave because the saber tooth Tiber is coming, we're programmed to hear our mother's voice. We're programmed to hear all these things that we've heard. So how do we lean into it? Do we lean away from it? What do we do? It was interesting because when they did that study, the preferences still came forward and said what was shared was like, we still haven't figured that out yet. So it is then when we say, all right, is it a male and a female? Do we Think of teams, do we think of things that way?

37
00:22:28,674 --> 00:23:08,136
Susan Westwater: And that's actually where one of our co authors, who I've done quite a bit of work with, and Scott has as well, of sonic branding and audio branding. It goes beyond the jingle. It goes into where we start to get into the fun world of HCI and audio cues. But there's a moment where we can be intentional about those noises and those sounds and triggers just as much as, like Apple was thinking about it the first when you restart, it's an upward note. The same thing, you know, the same things apply when you start thinking about it. You can get really geeked out, to be quite honest. It's also where you get to be creative again. We have all this science and technology. This is where we kind of get into the art of it, which I find really fun.

38
00:23:08,208 --> 00:23:08,860
Daniel Nestle: Yeah.

39
00:23:09,230 --> 00:23:13,318
Susan Westwater: And having those conversations, it just keeps.

40
00:23:13,494 --> 00:23:52,640
Daniel Nestle: Like layer after layer is peeling away in my head as you speak. There's this whole. Okay, this whole front end where somebody's interacting with the voice. So how that interaction happens, that basically dictates the immediacy of the experience and that dictates the intentionality of a lot of things, right? Yep. But fundamentally, right, the. The conversation, the voice AI part of it is still the LLM, right. Or the. Or whatever language model is powering it. Is that right? Is that a good way to think about it?

41
00:23:52,680 --> 00:24:31,592
Scot Westwater: It's a piece of it, yeah. So the way we used to traditionally do is we would have to map out intents and literally say, if someone says this is what they mean, and here's all the synonyms that go along with it, and here's our response to it. So where LLMs come in is they're really good at matching language and intent, so we don't actually have to do that as much. But really, we're relying very heavily on natural language processing, even more so than LLMs. So there may be an LLM component of it, but it's not the primary driver of the experience. There's also, like the concept of RAG within generative AI also applies on the conversational side.

42
00:24:31,696 --> 00:24:34,740
Daniel Nestle: Retrieval, augmented generation or generation.

43
00:24:34,820 --> 00:24:36,480
Scot Westwater: Yeah. So having.

44
00:24:40,060 --> 00:24:45,604
Susan Westwater: Because, let's be honest, acronyms and abbreviations are the devil. Because it can mean so many different things.

45
00:24:45,692 --> 00:24:45,956
Daniel Nestle: Yeah.

46
00:24:45,988 --> 00:25:16,614
Scot Westwater: But it's a fancy way of saying a knowledge base. So uploading your own information, whether it's something as simple as a Google Doc, something as complex as an actual database, but ultimately it's your information, your proprietary information that's not part of the larger model. And that's actually going to likely be where a lot of that content and a lot of that information is being powered from. So that's kind of how we've moved even four or five years ago into adopting generative AI. But it's really helped make the process of creation a lot more efficient.

47
00:25:16,742 --> 00:25:19,290
Daniel Nestle: So it's like you're creating small language models.

48
00:25:20,510 --> 00:25:23,090
Scot Westwater: That's a great way to look at it for each.

49
00:25:23,470 --> 00:25:48,894
Daniel Nestle: I've been making no bones or no secrets about this, but I'm crazy about notebooklm. Right. Just nuts about it. And, and we can get to the Voice part of that later if we'd like, because I do have questions. But that the idea of like, of creating your own knowledge base in a single instance that is this, that is basically the sole source of what the AI is essentially allowed to talk about or give you.

50
00:25:49,062 --> 00:25:49,486
Susan Westwater: Is.

51
00:25:49,558 --> 00:26:32,446
Daniel Nestle: Is that the same kind. Is that the same way to think about a voice system? And maybe it's. Maybe it would be easier to kind of narrow it down to a practical example. Like, so recently it's different. People have been coming up on my radar. And I spoke to a company that does conversational AI for outbound sales. Right now, that's new because most. Because as you said earlier, Susan, voice is invited typically. Right. But this is actually going out there and somewhat being. It's opt in and all that stuff. But if you really want to get down to it's an, it's a, it's outbound.

52
00:26:32,558 --> 00:26:32,926
Scot Westwater: Yep.

53
00:26:32,958 --> 00:27:24,134
Daniel Nestle: It's a push. So it's crossing that barrier. So I was asking him, well, how do you. I mean, how do you regulate what they, like, what the AI like, how that conversation goes? And he was just explaining that it's. It's already quite narrowed down by the nature of the funnel and this and the task and the sales thing. And the AI knows because, you know, the system prompt, as it were, just says if you veer off of this conversation, you bring it right back to this to say, xyz, you know, so with the capability of that happening, you know, that's. Or with that kind of way that we can use the technology, it seems like now it's really taking off. And you know, I was really just curious about, like, it's not as simple as, okay, I have an LLM.

54
00:27:24,182 --> 00:27:26,374
Daniel Nestle: I could just plop Voice into it. Right?

55
00:27:26,462 --> 00:27:27,222
Susan Westwater: Yeah, please don't.

56
00:27:27,286 --> 00:27:28,010
Daniel Nestle: Okay.

57
00:27:29,790 --> 00:28:16,010
Susan Westwater: Yeah. I mean, because you started talking about those intentional things of what are we going to talk about, what don't we talk about? But most conversations, even like here is, you know, unscripted conversation. Right? Right. They're not linear. And so even before LLMs and those things. So we used to have to think about how can this conversation go off. And from a UX perspective, we talk about the happy path, right? Which is the fastest path all things go, right. We go happily to where they need to go. We spend in conversational design, spend probably what, 80% of the time getting not planning for the happy path. Cause that's easy. It's how do I get someone back to where they want to. How do I guide them without A being forceful and a jerk B, getting them without irritating the crap out of them?

58
00:28:16,050 --> 00:28:59,370
Susan Westwater: Because I've asked them 13 times to clarify how do we do that? As you would as in a conversation with someone, the people you like talking to or those types of interviews. And then also how do we. When we do. When something comes up that we don't want to talk about or if there's something that's inappropriate or. And there's two sides of that, there's off lane, off task. Right. And it's just no, we'll get back to that or how have you. But then there's also the. That is completely inappropriate. How do I handle that conversation for someone? Because I think that is the one nature of these bots. People say things to robots that you would never ever in a billion years even fathom saying to a human for sure. And it can sound like a human.

59
00:28:59,490 --> 00:29:37,170
Susan Westwater: They're still going to talk to it in a way that's a bit different. So how do we set those boundaries? But also how do we make sure that whatever it is we really want them to know or they need to know, we're still getting to that and that derailing. So those are those important things. And you still have to think about that. You still have to put those guardrails in. You still have to put that into the logic. So you may have your whole corpus of information in your knowledge base, but then it's sort of the. Here are the guardrails of where that talks to. What's helpful when we start Talking about those LLMs is it gives us a broader depth and breadth of understanding what someone's even asking.

60
00:29:37,330 --> 00:30:24,140
Susan Westwater: So instead of having to hard code, a great example is one of our partners were building a skill for. Their company's name is Cesium Sonne. It is a French name. It gets butchered a lot. And how do I get that to work? So it was Cesium and going through and figuring out every horrible way someone could pronounce it. It's 6am or all these other things, then figuring out then after we saw it launched, of looking back and saying, oh, gosh, I didn't even think someone would call it that. All right, let's make sure that it understands what we're talking about. Those are kind of some of the fine tuning and that's a very simplified example just for anyone thinking, oh, that sounds easy. Yeah, it gets a little more complex. But the point of it is thinking through each of those bits.

61
00:30:24,220 --> 00:30:42,898
Susan Westwater: That's where LLMs make our lives a little bit easier. Because some of that logic happens a bit more easily of, oh, you mean this. Because contextually it's that as opposed to having to say no specifically if you hear this phonetically spelled. This is how this works or things like that. And that's just one side of it.

62
00:30:43,034 --> 00:31:29,964
Scot Westwater: Well, and the other part of the sales conversation is it doesn't have to be an outbound. So we've done some work with home healthcare companies and I think it's less than 20% of inbound requests actually get a response. Whether it's email or calling in. Most, most organizations aren't responding. So if you had almost like an autoresponder that as soon as someone submitted a form would call them up and answer basic questions and then try to schedule an appointment. Yeah, like that's going to go a long way to, you know, helping move that prospect from informational to at least consideration, if not purchase. But it's those types of things that you can implement where it doesn't even have to proactively go out and find people on your behalf, just simply responding to people that are already hitting your website and giving them value immediately.

63
00:31:30,092 --> 00:31:32,396
Scot Westwater: There's technology now that enables that.

64
00:31:32,468 --> 00:32:26,542
Daniel Nestle: So it's that inbound case is the first one that I heard about you. Like, the outbound's like late. That's why I'm like, oh, you can do that now. Because I'd already, you know, heard about, you know, makes sense that there's just frontline customer service scheduling, you know, scheduling appointments. There's a, there's a great, a company doing work with auto dealerships and called Stella AI. They do, they do, you know, like repair appointment, like appointment scheduling and service appointments and things like this. And you know, it's essentially you're only limited by how many phone lines you have essentially and by how, by or how much bandwidth you have on the, on the phone line in terms of how many calls can happen at once and nobody's hitting the busy signals or nobody's getting the voicemail and all that's wonderful from an experience standpoint, right? Yep.

65
00:32:26,686 --> 00:33:28,898
Daniel Nestle: But thinking about the real like kind of that next level interaction, right? This is, these are the baby steps, you know, that next level interaction where you know, you develop a conversational solution or a conversational, you know, team member or a conversational phone person, whatever it is, that is actually going to have deeper interaction on a broader, more, I guess a less a, a, a, a more general topic or more significant problems, it's going to discuss more significant problems with the other person. Just thinking about how, first of all, how many guardrails and, and guidelines. Let's not even mention the ethics because I mean ethical AI is certainly a, an issue with everyone, but all of that has to be included.

66
00:33:28,994 --> 00:33:58,220
Daniel Nestle: But add on top of that, I could see the, again, going back to that PR meeting, that weekly PR meeting, I could see the comms people sitting there with hair on fire saying, all right, we're going to let loose a conversational AI to talk to somebody. And if somebody doesn't know, what if they don't know it's AI and then they say something that is either offensive or mispronounced or maybe the other person just simply mishears something, you know, or.

67
00:33:58,260 --> 00:34:02,332
Scot Westwater: It goes ahead and offers a bereavement fair that doesn't exist for Air Canada.

68
00:34:02,436 --> 00:34:07,510
Daniel Nestle: I mean, that's how that would never happen. Like something like this.

69
00:34:07,660 --> 00:34:08,322
Scot Westwater: Yeah.

70
00:34:08,466 --> 00:35:07,390
Daniel Nestle: You know, so the, the trepidation, the reluctance to sort of embrace this full on. I guess it, I guess it has to do with technological advancement and same thing with everything else with compliance. But you know, for the PR world and the comms world and for marketers who deal with, especially in the B2B space where it's far more, it just, you know, there has to be a simultaneous maybe way and maybe you can help me understand this, that we can use conversational AI to manage brand reputation. Right. To kind of get ahead of that. So do you guys have any thoughts about that? About how like what are, what is conversational AI's potential role as an assistant or as a helper with, or an even accelerator for reputational issues?

71
00:35:08,650 --> 00:36:00,790
Susan Westwater: Well, some of the things is it goes back to that intentional discussion and even as you were talking about, let's say we have a crisis, you have now suddenly an ability to have something that can start to field or start, quite frankly, it's that qualifying of either the lead or the qualifying of who, what they're talking to or what information needs to get out there. It's coming from. If you have crafted your voice correctly, if you have crafted those, the conversation correctly, what will happen is then they will get a warm brand embodying interaction and engagement. Now the biggest thing to remember is, number one, if you have to go to a human, that's not a fail. That's simply a, a situation of where we make sure that you do it smoothly and say, oh, I'm sorry, this is something out of my world.

72
00:36:01,410 --> 00:36:47,602
Susan Westwater: And that is why we often will say you're talking to an agent, you're talking to a virtual agent, because it then sets expectations. What's interesting to me always is there is the age old argument of I want to talk to the operator, I want to talk to the human. A human can't change the policy anymore. And actually, as we just were laughing about the Air Canada example, sometimes they get creative. That's where you have to think about those. That's where testing is important. I'm not gonna lie, every time I see a create a bot in 20 minutes, a little part of me dies. Because fast, good and cheap. You've got fast and cheap happening. Guess what you're missing? You are missing some really important parts to how that needs to be monitored. Ethan Mullock talks about the human in the loop.

73
00:36:47,746 --> 00:37:09,180
Susan Westwater: There always needs to be human in the loop. But what you're actually able to do is you're able to scale that message. You're able to make sure that they're out there and they're covering it. Also, you're covering all your channels because people aren't always going to call, they're not always even going to hit your website and hit your chatbot. They're going to go to your social bot, they're going to go to your Twitter feed, they're going to go, I'm sorry, X feed. Sorry, Elon.

74
00:37:11,200 --> 00:37:12,456
Daniel Nestle: Tweetster. Just.

75
00:37:12,608 --> 00:37:13,640
Susan Westwater: I like that. Oh, I like that.

76
00:37:13,680 --> 00:37:15,020
Daniel Nestle: Just be safe.

77
00:37:15,440 --> 00:37:53,330
Susan Westwater: Yeah, left Twix, right Twixt. But they're going to go to all these other channels. They might even go and store your kiosk. These are places where you can deploy that because you've already got your system set up where you're going to get that consistent answer and that can help field it so that you're on point, you're able to get ahead of that. So if it's Oh, I hear, you know, I heard someone's going to kneel at the Philadelphia Eagles game back when that was a big deal. I actually remember dealing with this as a corporate example. Oh, shoot. They're one of our partners. What do we have? What's our statement? Well, I'm going to make sure that statement's available in multiple places and we can talk about it.

78
00:37:53,370 --> 00:38:31,680
Susan Westwater: And I've also got a point of where I have a strategy because I know my PR specialist has already coached someone. Imagine if when you're coaching someone to be a speaker, they actually do it because they aren't going to be like, oh, I got nervous and then I messed it up. You're actually able to have a little more control. So some of our pharma clients are always like, oh, wait a minute, compliance is going to adore that we can control some of that. And depending on how we're using LLMs and generative AI, we're able to be a little more on script than a human would. And that's kind of the thing where I laugh and say, when folks are saying, oh, I want the human. Humans don't change those policies. Humans sometimes aren't even better.

79
00:38:31,800 --> 00:39:19,098
Susan Westwater: Even when we look at, like, sentiment analysis, humans are only right about between 85 and 90% of the time of interpreting someone's mood. So when someone's like, well, is it 100%? Well, no, but no one is. And that doesn't say that, oh, it doesn't have to be perfect, but it's our way of also saying our benchmark is a little different, of perfection isn't always it. That's what makes us human. That's also what makes things authentic. It's just what is that tolerance of I will allow you to stumble, stammer or mispronounce something. I'm not going to allow you to create a whole new policy. So that's where the testing and the looking through the use case and constant monitoring with conversational AI. And I feel like all AI is this now. And everyone's starting to accept that launch does not mean done right.

80
00:39:19,154 --> 00:39:30,538
Susan Westwater: It's just your starting point. And now we're moving into a different way of dealing with it. We've been planning it so that we've got our basics and our foundations. Now we're going to start to explore and see what else it can do.

81
00:39:30,674 --> 00:39:37,254
Scot Westwater: The big part of this is not to replace existing tactics. It's in addition to. It's just another tool in your toolbox.

82
00:39:37,302 --> 00:39:37,830
Daniel Nestle: It's additive.

83
00:39:37,910 --> 00:40:22,238
Scot Westwater: So for Instance back during COVID we worked with McDonald's UK and we actually added in a lot of PR crisis information to an Alexa skill because it's, you know, what are your now crisis hours, what's the menu, what are your cleaning policies like? All those types of things that people were asking. We looked at how do we do it in a conversational way. And then you know, as a secondary piece, here's a way you can make your orders and then working with Amazon to actually facilitate all that stuff. But that actually started out almost as a exclusively PR approach to address that very real need that they had to try to get that information out as many places as possible. So there's also ways you can use it in existing tactics as again like mobile app as a layer on top.

84
00:40:22,294 --> 00:40:34,716
Scot Westwater: But you can add PR content, you can add brand promotional content. It doesn't have to be my PR stuff's over here, my brand stuff's over here. It could actually live in the same place then when people ask for it's just delivered.

85
00:40:34,828 --> 00:41:25,056
Daniel Nestle: Yeah. It is about access ultimately. Right. I mean it's about being able to get the information in the way you want, when you want it. And we're being spoiled or conditioned I think as this generation or this multitude of generations as we deal with AI that's what's going to continue to happen. You know, I don't know. I think it's the more I'm talking to you, the more I'm sort of. Now those dots are starting to connect even brighter. You were just talking about how it won't veer off of if you prepare your AI, your conversational AI with statements, it won't veer off the messaging. I've taken executives through media training for hours and hours.

86
00:41:25,128 --> 00:41:25,888
Scot Westwater: Yeah, yeah.

87
00:41:25,904 --> 00:41:29,104
Daniel Nestle: And some of them are wonderful and some of them are need improvement.

88
00:41:29,232 --> 00:41:29,900
Scot Westwater: Sure.

89
00:41:30,360 --> 00:42:01,170
Daniel Nestle: You know, and some people get it immediately and some people have a natural talent for bridging or for pivoting or you know, whatever in a non shifty way. I could upload a deck, a media training deck right now and then it will be media trained like whatever instance of AI I'm using and it won't veer off of that. It'll stick to the rules until it doesn't. I suppose you have to keep watching.

90
00:42:01,510 --> 00:42:52,032
Susan Westwater: Well and that's where with all of these things and that's one of the big things of. I was actually at a conference, the Voice and AI conference in October and was sitting. I love going to the way more technical stuff because Strategy is really fun, but hearing a technical piece of talking through, like, tolerances of what am I willing to accept? That's what we figure out before we start building. Or we should. How about that? We should. And then you use. You can use those rag systems to test and make sure that what's coming out is right. And then determining. Can I live with if it puts this word here or if it's that or these, you know, hallucinations of this level? Okay. Or if not, because, you know, hey, this is life or death. This is health care information, then. No, then let's.

91
00:42:52,096 --> 00:43:07,712
Susan Westwater: We're gonna go to a much more rigid structure, and that's the way life is. Yeah. You have to think of just those tolerances, then give your guidance of how much generative AI can I use. And that's okay, because not everything is generative AI. I know that's probably blasphemy to some.

92
00:43:07,736 --> 00:43:46,626
Daniel Nestle: Folks, but I would think that if you could. If any company that. And this means most companies, I suppose, that has a customer base and has a customer service function, whether it's a call center or, you know, or just your salespeople or your. Or your marketers just answering the phone, whoever it is, you know, that customer service function. I think that if we could. If we only ever get to the point where somebody calls in and you don't have to listen to seven recordings and you don't have to wait.

93
00:43:46,818 --> 00:43:47,362
Scot Westwater: Yep.

94
00:43:47,426 --> 00:44:02,226
Daniel Nestle: Yeah, right. That's a major thing. That's why every time I call my. My health insurance or my specialty pharmacy or whatever it is, the very first thing I say is representative. Representative, always.

95
00:44:02,298 --> 00:44:05,394
Scot Westwater: Because you don't want to. You don't want to navigate the structure, you don't want to do the navigation.

96
00:44:05,442 --> 00:44:06,274
Daniel Nestle: I don't have that much time.

97
00:44:06,362 --> 00:44:16,976
Susan Westwater: Yeah, yeah. Well. And chances are you went to the website, you couldn't find it. So you're like, fine, I'll call. And so, like, I always love the. Hey, did you know you could go to our website? That's why I'm here.

98
00:44:17,048 --> 00:44:17,280
Scot Westwater: Right.

99
00:44:17,320 --> 00:44:58,010
Susan Westwater: You know, and. And some of that. I mean, like, we could pick on customer service phone trees all day. The other part of it, too, though, are like, little interesting things from a B2B perspective of, let's say, that person who is doing the media training. And it is a human. But let's say I have media notes. There's certain approved stats that I'm allowed to use. Instead of having to go look and read that form. If I have an assistant set up internally where I can say, what is our latest? How many billion hamburgers have been served? What's our current line that helps the PR team instead of having to constantly go, is it hidden in SharePoint somewhere? Is it how we phrase that? Can I just ask it questions? That's something we're actually playing with right now.

100
00:44:58,050 --> 00:45:39,566
Susan Westwater: Is looking at research a collection of trend reports and I just can ask it questions of like, hey, is this trending down? And then what it's doing is it's comparing all of those reports and that data. So that way then I'm getting that piece from there instead of having to go through and sift it myself. Oh yeah, and we're testing. And those are things internally of like, no, you can't say that. Yes, but you can say this. Or what is the correct spelling? These are, I mean, those are, I mean when I think of the bane of my existence of no, it's not called this, it's called, you know, you flip the words. Those seem like little simple things. Or what is our policy on this? Or how do we handle that? Or again, what are those things?

101
00:45:39,638 --> 00:46:14,888
Susan Westwater: We can actually put that information in a conversational way of just asking instead of having to take the moment to type it. Now, by the way, conversational does include typing because there sometimes is information you don't want spoken. Right. Because I just want to look at it on my screen. Like I just asked about test results. I don't know if I want everyone in this car to know, in this train car to know that I actually just tested positive for Covid while I'm trying to get to the next stop. Things like that or personal information like bank balances, financial services is a whole other realm of confidential information. So you have to balance that.

102
00:46:14,944 --> 00:46:54,922
Daniel Nestle: Yeah, I would imagine that. And if you sort of extrapolate that to, you know, what a day to day marketer or communicator might be facing, you know, in their job. And you know, you're using different tools to get different things done. You know, everybody talks about how the holy grail is measurement, etc. Etc. Yeah, you know, that's not something you want to talk. You don't want somebody say, you don't want to be have a voice conversation where your conversation partner is saying, and then it travels at a 45 degree upward on the X axis until it reaches point. You know, that doesn't work. So there has to be a multimodal kind of aspect to everything.

103
00:46:55,106 --> 00:47:08,202
Scot Westwater: Yeah, it could be voice input and it comes back on the screen it could be like in certain cases, especially like in the car where you can't physically look at a screen. You get a voice coming back. There's different, depending on the situation, there's different ways we can handle it.

104
00:47:08,306 --> 00:47:38,224
Susan Westwater: You can even like this is so, this is going to sound so Neanderthal, but you can just do something like, I don't know, send a link to the graphic that you need to look at when you're at a point where you can look at it. I know that sounds crazy and very simple, but those are some of the simple things. If I've asked you to prepare this report for me, can I see this visual? Then when I'm where I need to be, you can do that. Especially if we're asking it to do agentic. Can you give me a report that gives me xyz? Go find that information and then send it to me.

105
00:47:38,392 --> 00:47:39,984
Scot Westwater: You said the latest buzzword.

106
00:47:40,032 --> 00:47:42,340
Daniel Nestle: I don't know whether I should have like an air horn.

107
00:47:42,680 --> 00:47:44,240
Susan Westwater: I don't know. It's all right, it's all right.

108
00:47:44,280 --> 00:48:07,344
Daniel Nestle: Agentic. Yeah, yeah. And it's because this seems, I think there's a, there's a world in which, and maybe it's the world we live in which your average person who is dealing with all these tools might think that conversational AI is agentic AI and because in, you know, many ways it is kind of miraculous.

109
00:48:07,392 --> 00:48:08,784
Susan Westwater: It's, it kind of can be, it.

110
00:48:08,792 --> 00:49:07,880
Daniel Nestle: Can be, it can be, of course and it will get there, you know, but I, I, I feel like it's, it's definitely a built for multimodal or multi task actions. You know, I, I just did, I did this great thing. My friend, Frank Prendergast, shout out to Frank has been really going overboard or I should say the extra mile with custom GPTs, you know, and with the addition of voice, you know, it, it does open up a couple different possibilities. So he's developed this kind of series of GPTs that helps you build your website copy, for example. And you know, in each it's broken down into several, like where when agents come in, you just talk to the agent, say build me a website. And just, you know, but this is in this case the first part of it.

111
00:49:08,180 --> 00:49:52,750
Daniel Nestle: He's set it up as an interview. So the custom GPT you have to go into voice mode. The custom GPT will ask you questions and you just talk, right? You talk and then the output is not a conversation. Like after the, after this is all done. The output is a document that takes everything you've said, but puts it into, like, proper terminology and into. Into something that makes sense. Cause I talk all over the place and. Yeah, you know, I think even doing that with that, with Frank's GPT, there was maybe the biggest thing that got me over the hurdle of. Of accepting and embracing voice as a. As a mode. Because I'm like, oh, I can talk all over the place and it still gets it.

112
00:49:53,210 --> 00:49:53,970
Susan Westwater: It's like.

113
00:49:54,090 --> 00:49:55,602
Daniel Nestle: It's a miracle. What kind of.

114
00:49:55,626 --> 00:50:25,926
Scot Westwater: Well, she's. She's the double English major. I was always the picture guy. I mean, I was a visual designer for years. So, like, I'm not a writer, but I've written two books. Well, guess what? We actually ended up speaking the book into existence. So we had a loose outline for what we wanted to talk about. And this was all pre chatgpt, because had we had chatgpt, it would have been even easier. But literally talking into existence, getting the transcripts and then doing editorial and kind of moving things around really helped my process because I can talk with the best of them.

115
00:50:25,998 --> 00:50:26,236
Daniel Nestle: But.

116
00:50:26,278 --> 00:50:41,780
Scot Westwater: But when I'm sitting at a blank or looking at a blank screen, it's very intimidating for me to start writing. So it was a natural way for me to just talk it out. We do presentations all the time and everything else, so it was no big deal. But that was the first. Aha. And then we're like, oh, we just used voice to write a book about voice.

117
00:50:43,320 --> 00:50:44,208
Susan Westwater: How meta was that?

118
00:50:44,264 --> 00:50:56,202
Scot Westwater: How meta was that? But yeah, I mean, it's definitely one of those things where once you get over that hurdle and you start interacting with these systems, especially as you start to see success, that's really where the magic happens.

119
00:50:56,346 --> 00:51:41,260
Susan Westwater: Yeah. And it's. There's something. There is something about speaking things out into existence. When you say it's. What is it? Bruce Lee talked about those being spells. It's the same thing when you're talking those things out. So we have to get used to and comfortable hearing our voices and talking to ourselves a little bit more than we used to. You know, it's probably. I mean, I don't know anyone who adores the sound of their own voice. Even though I've been in meetings with people, I swear to. But. But that whole idea, but kind of getting past that, but then getting comfortable, it is a lot easier for helping folks who. Again, from reading or getting that access. Like you said, it's information access and distribution. There's a lot of things in AI that ultimately. And it's because of my content background.

120
00:51:41,420 --> 00:51:43,308
Susan Westwater: It's a content thing. It's not.

121
00:51:43,364 --> 00:51:43,900
Daniel Nestle: Agreed.

122
00:51:43,980 --> 00:52:24,340
Susan Westwater: It's technology enabling content as opposed to. And I think that's kind of one of the things to think through of what am I trying to do with this AI or these things. Because there's a lot of really fun novelty things. But when we really start cooking, I feel like with peanut oil, it's definitely involving thinking about that content accessibility, how am I giving that information, and yes, how I parse it when I'm talking to someone will be different than how if they're reading it, than if they're reading it on a chatbot, than what have you. That is definitely where the art and the design comes in. Just as we would never put a brochure on a website, same thing anymore. You have to think about those nuances. Yeah.

123
00:52:27,160 --> 00:52:29,392
Daniel Nestle: I remember 2003. Does anybody else.

124
00:52:29,536 --> 00:52:30,976
Susan Westwater: Yeah. Oh, yeah.

125
00:52:31,168 --> 00:52:41,456
Daniel Nestle: The. Yeah, it's. It's interesting. You've. So there's a couple. There's a couple things there that I. I'd love to pick apart, but. Or not pick apart. Cause that makes it sound. And that makes it sound like I've got.

126
00:52:41,528 --> 00:52:42,768
Susan Westwater: I've got a critique.

127
00:52:42,944 --> 00:53:33,144
Daniel Nestle: No, it's more about, like, I want to pull it up and just understand more. But you mentioned the. The magic word. And I. I feel like, you know, as we start to kind of wind down here, I would like to get into where you see all this going, and especially with agents. So agentic AI, you know, it's not something I know. I mean, I know as much about it as, you know, any dabbler in AI who is like. Who's like, all right, I can't. I just can't wait for. Well, I'm. I'm not exactly feeling great and easy about agents coming along, but I know they're coming, so, hey, I'm all in. So now I'm like, all right, fine, let me have the thing that I just kind of turned like, just speak into my phone or just type a sentence, whatever it is.

128
00:53:33,232 --> 00:53:58,704
Daniel Nestle: I give one input, says, look, I need to publish my podcast. You know, go grab the files, put them in here, do this, get that, make this happen. You know, do the. Do the corrections and give me the drafts, like, done. You know, that's. Because I'm going to have to do that now with AI, of course, that's so much less time than it used to be, but sure. You know, taking that three hours down to 30 minutes is very attractive to Me.

129
00:53:58,872 --> 00:54:01,648
Scot Westwater: But what if you could take it down to about three minutes?

130
00:54:01,744 --> 00:54:02,288
Daniel Nestle: Even better.

131
00:54:02,344 --> 00:54:37,430
Scot Westwater: And that's. And I think that's where the agents really come in. So everyone talks about agents like they're this far off thing. But I've been working with agents for the past year. I won't mention the platform, but there's a platform that's built that you can use. Claude OpenAI different language models. Perplexity. And effectively what you do is you create individual virtual employees almost, or AI employees, and you give them, like I give mine, specific personalities. So I've got one that does use cases, one that does web research for me, whatever. And then you basically just give it a series of steps.

132
00:54:37,510 --> 00:55:12,616
Scot Westwater: So if you have SOPs already established for how you do something, especially like when you're talking about the podcast, you could literally give those instructions to this assistant and literally give it the input and it'll then do its job and then hand it off to the next one. So maybe you've got one that takes the transcript and creates the show notes, and then you got one that actually will go through and upload the files and whatever the different things are. Like, those are all things you can actually do and start to stitch some things together. So is it true? Agentic? Not yet, but we're close enough where you can almost fake it and see a lot of those efficiencies.

133
00:55:12,728 --> 00:55:13,400
Daniel Nestle: Yeah.

134
00:55:13,560 --> 00:55:54,946
Scot Westwater: Right now. And so, like, there are platforms that exist and if anyone wants more information about it, reach out to me and I'm happy to talk about it. But the idea being that you can actually start to take advantage of it now and start to see a lot of benefit in your own work. I mean, I've deployed at this point probably 50 different agents that are doing various things, from editorial to research to outreach. I mean, there's all kinds of things that I'm doing, but I started small. I saw some success with it. I went, oh, what else can I do with it? And just kind of looked at adding features. So I wasn't trying to boil the ocean all at once. It was literally like, you know, let's triage our email. Let's start there and see how that works.

135
00:55:54,978 --> 00:55:57,356
Scot Westwater: Oh, that worked pretty well. Now let me see what else I can do.

136
00:55:57,458 --> 00:56:08,312
Susan Westwater: Yeah, I think the other thing that we're seeing is right now, and this was a conversation we had earlier today, is there's two things. One, the value of our work, the value of content.

137
00:56:08,416 --> 00:56:08,872
Scot Westwater: Yeah.

138
00:56:08,936 --> 00:56:49,470
Susan Westwater: And basically the value of brand are going to change. There's analogy of like when I moved from that agency of four to. Which was. Wasn't for when I left, but then to an agency of 977 people. Suddenly all a bunch of the hats I used to have to wear because I was in a small organization went away and I had this freak out of imposter syndrome of holy hell, what is my job again? Am I over my head? Wait, I'm supposed to focus on these things? And then of course, I got, you know, I got over it and I realized, no, I can dive this much deeper into the things that were the talent and things I wanted to bring to my role. That is what's going to happen as well is like from an agency and B2B perspective.

139
00:56:49,970 --> 00:57:13,348
Susan Westwater: Yeah, we're not going to bill by how many hours it takes us to do something. We're going to bill by like more of that old adage of the story of the, what is it, the submarine. And the guy comes and shows up with a wrench and in two minutes figures it out, tightens this one screw and they say, and he says, that'll be $200,000. Well, it only took you two minutes. Yeah, but it took me 20 years to learn. That's the screwdom essay.

140
00:57:13,364 --> 00:57:13,924
Daniel Nestle: Exactly.

141
00:57:14,052 --> 00:57:39,192
Susan Westwater: That's where that value is going to come in. And we can't discount that as from our humanity perspective, truly human things. But the other side with agents that I think is also is understanding. Do I really need to launch something to space to answer a question? Yes, it's super cool to have super intelligence, but I don't need super intelligence to do some of the things I need to do.

142
00:57:39,216 --> 00:58:19,464
Susan Westwater: And as a matter of fact, superintelligence might actually be a challenge if I'm trying to keep them on script, if I'm trying to tell it, don't optimize this, you have to say these words that I'm able to use these things within the potential of what it is, as opposed to having, you know, having the Formula one Ferrari that I can never drive down Lakeshore Drive because A, it's not fit for it, and B, I can only go 40 miles an hour. Those are the things that I think we're going to see in the future. The other thing too is this big gap though, between knowing when to use those things and familiarity of business to understand, okay, I don't have to build it, but how do I use it?

143
00:58:19,632 --> 00:58:55,540
Susan Westwater: And I think that's something like we laughingly said today, oh my gosh, it's another seven years of this of. We were teaching people all about voice and how to use it and the education and kind of the evangelism. We're now looking at that. Of how we use AI and evangelism as well. Of, here's how I can help you so that you're feeling comfortable. Here's how to do the start small and test. And it's not start small because it's stupid and small. It's. That's only the amount of impact I need today to convince me to continue building. And that's okay, because then that's something I can do in three months, as opposed to 18 months.

144
00:58:55,580 --> 00:59:44,430
Daniel Nestle: Launching a giant thing that's so important to just keep. I think that's a great way to think about it. And, you know, I have been having discussions about, okay, great, we're going to have super intelligence or we'll have agents, but at the end of the process or whatever they produce, they still have to be managed, and you still have to have somebody to look at what it's creating. If it's on the creator side, like, if you're putting agents in charge of running our power grid, hey, that's out of my pay range. I don't know how to do that, but I would hope that there's humans watching. I really would. But there's. Then we're all using all kinds of algorithms to optimize different parts of our. Of our infrastructure. And that's. That's fine.

145
00:59:44,770 --> 01:00:10,364
Daniel Nestle: But I just, you know, I, I struggle a little bit with the idea that, oh, once we have super intelligence, then we can all sit back and, you know, it'll just take care of it. I don't. I just simply don't think. I simply don't think that's true. I feel like, you know, the. The advent of more knowledge at our fingertips has been the story for all of humanity.

146
01:00:10,492 --> 01:00:11,068
Scot Westwater: Yep.

147
01:00:11,164 --> 01:00:49,150
Daniel Nestle: You know, we had camp, we had. We were hunter gatherers. Suddenly figured somebody figured out how to. How to pick up a piece of bone with a little bit of meat on it. And just like, oh, there's, like. I can, you know, I can put this. Use this and draw something on the wall. Maybe not with meat, but, you know, I'm talking about this plant has a particular color. Oh, it looks good on the wall. Good. Goodness gracious. I can put the whole story of our travel here. So you suddenly knowledge grows. So then you know, you get to the Library of Alexandria, everybody's got this knowledge. The printing press happens. Everybody's got more knowledge now. I mean, it didn't happen as it didn't have it as immediately and in, in as accelerated a way as it's happening now.

148
01:00:49,270 --> 01:01:22,172
Daniel Nestle: But in every single instance, not everybody did that. Like, right. You get to the point, I think as people, we all have this place, this limited point where we get to and we're like, okay, that's enough knowledge. I'm fine, you know, I don't need any more knowledge, you know, and I think it's just like you said, it's like, okay, I have the super intelligence to do things. Will I really need it to do all these things? Like I've got things that I can get this far and I don't need the rest of this stuff.

149
01:01:22,316 --> 01:02:08,182
Scot Westwater: Well, in practical business you probably don't need AGI, quite frankly. A lot of what we're talking about is automating basic business processes, whether you're using AI or other tools. But at the end of the day it's trying to be more efficient with the time that you have and not spending time on very time intensive manual processes. That's one of the components. So that would be the internal efficiency story. Or are you trying to increase sales, or are you trying to find new business? What are your objectives and kind of start there and look at how can technology in general, it doesn't have to be AI, but how does that help enable that or solve some of those issues? Yeah, but I'm not convinced that we need AGI or supercomputing or super intelligence for business. Now.

150
01:02:08,206 --> 01:02:13,990
Scot Westwater: There are going to be a lot of things, you know, in the physics realm, in the chemistry realm, things like that.

151
01:02:14,110 --> 01:02:14,534
Daniel Nestle: Yep.

152
01:02:14,582 --> 01:02:57,872
Scot Westwater: You know, urban planning, that's where a lot of that stuff's going to start to benefit. But the big challenge that I see is there's a lot of noise in the AI space about AGI and superintelligence and it's almost paralyzed the business space because there's plenty of things they can do right now to get started, start to experiment, have some of those learnings, have some of the successes. But because we keep churning out, today there are four more models released and then tomorrow there's going to be another seven and then there's this robot coming. It's like this never ending funnel of stuff. But part of what we do is help people filter out what's relevant to them versus what's just kind of general noise or cool stuff, sci fi stuff, but might not be directly related to your business.

153
01:02:57,976 --> 01:03:41,068
Susan Westwater: Or let's be honest, some of the demos we see that are demos are very wizard of Oz demos of like, it'll behave like this. But actually, before you even get touch this, it's gonna be 12 months because we got this one POC running. And that's okay. Cause we need that work and we need to see where we're going. But there's stuff we can do in the here and now that honestly is not rocket science anymore. And that's okay because that's awesome, because it's familiar, it's controllable, and it'll actually help us build our comfort zone so that when the big stuff's scalable and doable, we'll actually have a comfort zone as opposed to. I can't imagine anyone who's jumping into the space right now.

154
01:03:41,204 --> 01:03:49,484
Susan Westwater: My hat's off to you because I feel like we've had all this time to kind of look at things and filter it and every. Like, it was not the breakneck speed.

155
01:03:49,532 --> 01:03:50,172
Daniel Nestle: No.

156
01:03:50,356 --> 01:04:14,810
Susan Westwater: And as of yesterday's presidential announcement, it's going to get even worse of we've got, you know, while he's talking about Stargate, a whole nother model released from a completely different not open AI with Deep Seek. And it's like, oh, crap. All right, well, now I got to know all about Deep Seek or something about it. And that's kind of also where we try to offer ourselves up of. You're already trying to run a business.

157
01:04:14,930 --> 01:04:16,618
Daniel Nestle: Yeah. Let us figure all these other things out.

158
01:04:16,674 --> 01:04:45,204
Susan Westwater: Let us. Let us figure this out. We. We promise and hand on our heart, you know, we're doing this for good. But it's also. That's bring in those folks so that way then you don't have to some degree, because. And that's because there's so much volume coming through of what makes sense and what applies to your business today as opposed to. That's freaking cool and amazing. I have no idea how I'm going to apply it, but freaking keep at it. You've got the funding. Thank you.

159
01:04:45,292 --> 01:05:28,530
Daniel Nestle: You know, I'll tell you, I can't think of a better way even to sort of, you know, to. To wrap that up than just to say, listen, you don't need everything. Right? You just. We. We have people like us crazy nutcases out here who are looking at everything for you. And even I don't look at everything for sure. Like, I'm just like, okay, this is not really right. I can't deal with this right now. I'm just gonna focus on these 44 different things I don't need. 73, you know, and just focus on what you can focus on to make those incremental changes. Look, as we. As we kind of wrap this up for real, any last words, guys? Is there any final last, little bit of wisdom that you'd like to drop on the crowd here?

160
01:05:28,910 --> 01:06:10,430
Scot Westwater: I mean, I feel like the biggest thing I can say is do what you did and just get started now, whether it's learning or experimenting or use case identification, like, just start. There is so much fear out there. There's so much, quite frankly, misinformation related to AI and it's just a matter of getting in there, getting comfortable, getting those quick wins, seeing how it can actually help your process. I'm also a big proponent of, like, shortcutting process by knowing how to do it manually and then going, oh, I wonder if I could try this little bit right here. Get those wins under your belt, get comfortable with it. It's not scary, but unless you start, you're never going to know.

161
01:06:10,890 --> 01:06:12,898
Daniel Nestle: Yep. Susan, anything to add to that?

162
01:06:13,034 --> 01:06:52,912
Susan Westwater: I mean, I would also say there's a lot of urgency about if you don't start now, your competitor is going to steal everything. It's going to steal your job. That's really not the way to think about it. That's just going to terrify yourself. It's also going to keep you up at night. We've got plenty of other things about AI to keep us up at night. What, what I. My advice is to start thinking about and are you ready? But also look at what you've got now. Don't treat it like this precious, fragile skunk work that can never touch anything else. If you don't make it part of your channels or part of how else you're doing things, it's always going to feel like this otherness, and it needs to.

163
01:06:52,936 --> 01:07:30,968
Susan Westwater: If you're going to be putting yourself out there as a comms team and you need to be cohesive, you have to, because if you don't have an AI policy and you push it out and you don't let people touch it, that people are going to use it because they will, because curious nature is your best star. And then you're going to lose out on all that information. You're going to lose out on their wins, their losses, and they're not going to socialize it. So it's still going to be. We're still at a time where it's very much at the individual and business unit, team, business team level. Enterprises are doing it but it's not necessarily scaled. This is the time to start kind of getting your collective team together and looking at it from that perspective.

164
01:07:31,144 --> 01:08:21,474
Daniel Nestle: You're preaching the gospel and that is exactly what I like to hear and what I will be telling people moving forward. Susan and Scott Westwater, I really appreciate your time. Thank you so much for coming. If anybody is interested in learning more about Susan and Scott and about Pragmatic Digital, go to PragmaticDigital on the web. Check them out, their bios are there. You can find out as much information as you'd like to on there. But if that's not Enough, go to LinkedIn, find Susan and Scott with 1T Westwater. It's very simple and it will be spelled properly in the episode title. You can also find Susan on Twixter JW75. Don't read into that folks. SJW75 on Twixter and then, you know, on the conference circuit we'll see them speak here and there a lot.

165
01:08:21,642 --> 01:08:44,569
Daniel Nestle: And I highly urge any of you who are hitting any of the big, you know, like SEM related or social media marketing stuff or what's the one we go to CEX or Macon, any of these big conferences in our world, just look out for Susan and Scott because I'm sure you'll find them. So thank you both so much for coming.

166
01:08:45,270 --> 01:09:05,645
Susan Westwater: And then one last thing is if you do locally Chicago, we do head up a meetup called AI and Chai. Thank you Paul and Kathy for letting us steal AI and CLE and making an AI and Chai. And that also is a quarterly place where we're going to be getting together with some information to help folks understand how do you bring AI to your organization?

167
01:09:05,837 --> 01:09:10,158
Daniel Nestle: Awesome. Oh, one last thing. I should have, I should have said, I said buy the book guys. Everybody out there.

168
01:09:10,214 --> 01:09:10,904
Susan Westwater: Yeah.

169
01:09:11,062 --> 01:09:21,060
Daniel Nestle: Voice marketing. Harnessing the power of conversational AI to drive customer engagement. Go to Amazon. The link will be in the show notes. And thank you all very much. Thanks guys. Thank you for joining us.

170
01:09:21,100 --> 01:09:21,332
Susan Westwater: Thank you.

171
01:09:21,356 --> 01:09:22,160
Scot Westwater: Thank you.

172
01:09:29,100 --> 01:09:53,470
Daniel Nestle: Thanks for taking the time to listen in on today's conversation. If you enjoyed it, please be sure to subscribe through the podcast player of your choice. Share with your friends and colleagues and leave me a review. Five stars would be preferred, but it's up to you. Do you have ideas for future guests or you want to be on the show? Let me know@dantrendingcommunicator.com thanks again for listening to the trending Communicator.