Show 2: The Authenticity Engine – Mining Human Truth

The "Authenticity Engine" is the RedLine methodology for fusing AI speed with human soul.

Transcript

Copy

SPEAKER_01 0:04

Welcome to the Signal Podcast, brought to you by Redline Advisors. I'm Tyler Marshall, and I'm here with Redline CEO and founder, Mick Hollison, and Redline partner, Calvin Shu. So let's talk about the antidote. You've built authenticity engine. Mick, is this just a branding exercise or is it an operational shift?

SPEAKER_02 0:23

Wow, this is uh this is all the above, uh Tyler. This is both a mindset shift, uh a bit of a bit of a tooling, a retooling kind of exercise, and it's certainly a methodology uh as well. So when I think about the authenticity engine, I think about a an inherently and a fundamentally different way for marketers to create content. I think for the past 10 or 15 years, marketers have been slaves to the SEO world. Everybody content is king. And and and by the way, that still holds true, but in an entirely different way than it did before. And everybody spent all of their time optimizing around headlines and meta tagging and ensuring that they would show up in the right way in a Google search and and paying Google, you know, ungodly sums of money to make sure that they showed up in both paid and organic search at the highest possible levels. And and and by the way, a lot of B2B tech companies got really good at it. But guess what? The whole world has been turned on its head by this idea of generative engine optimization or GEO. And it requires a completely different kind of content. And now here's like the warning in advance for CEOs that might be listening or or watching this. I hate to tell you, folks, but uh you know, you've got to get in the game now. So if you are a C-level executive and you are not already beginning to produce some authentic content, you're not doing what I'm doing right now. You're taking time out of your busy schedule to connect directly with your customers and buyers in an authentic way, then you are already missing the plot. And the reason that you're missing the plot is that the first audience for everything that you do now is more than likely a machine. It's the LLMs themselves. And they value authentic ironically, right? Ironically, the machines value the authentic human content more than they do their own machine-delivered content, which is sort of a mind-bender if you step back and and think about it a little bit. But when I think about the authenticity engine, I think about that first step, and it's creating authentic content generated by the executives that happen to run your company. So whoever you are out there, that that's a big role that you need to play. And it means marketers need to step it up. It means they can't just, you know, use an LLM, pen out, you know, spit out some 20-page white paper that sounds exactly like your competitor's 20-page white paper. They've got to generate original content that involves as many of the real live human beings in their company as they possibly can. And that's that's the first thing that they need to do in this authenticity engine. Once they've begun to do that, they can begin to nuance that content and did all the different deliverables and vehicles that they've that they've got to build out. But that that becomes core to content generation in the AI era in a way that it never was before. It was a nice to have. Videos were like a nice to have. Now they are the the critical sort of cornerstone of your content uh strategy at any B2B company, particularly B2B tech, which is so crowded right now.

SPEAKER_00 4:06

Yeah, I I think that that that's uh exactly right. I think the the if you boil it down, the name of the game is we're playing a big game of how beat the summarization, right? How do you beat the summarization? And we we say defy the algorithm, right? Defy the algorithm. Um and uh that that's why video works so well, is because it's extremely hard for an LLM to quote unquote watch a video and then just summarize it. They would rather just say, hey, here's a great video from CEO of XYZ company um that makes this point, right? And and they will pass it along verbatim. Um I think that's true also of your tech leaders and of your sales leaders. And when you make a pitch or you're trying to explain technology, um you know, if you hand it a white paper or you hand an LLM a slide deck uh with a sales pitch, it can very easily say, oh yeah, this is a slide that tries to sell the benefits of this, it costs this much, and boom. Right. And all that carefully crafted messaging and you know, the pictures and the brand identity and uh everything that you put into the slide goes away because the it was too easy to summarize. And so the the more of these sort of um, you know, I I I I'm I'm representing a human or I can't summarize it, and therefore I need to just represent it how it is, otherwise, because I don't know what to do with it, you know, from the LLM's perspective. Um, those are all sort of uh you know great ways of breaking out of that that AI trap.

SPEAKER_02 5:38

So Cal, Calvin, you know, in the previous episode, you talked a little bit about how you stumbled upon this idea of signal whisper ratio and checksum statements and things like that uh through your through your music uh and the fact that you know without a little bit of human input, you couldn't really get an original thought out of the out of out of the LLMs. I'm I'm using a thought there as a metaphor for the uh for the riff you were trying to create or whatever it happened to happened to be. So in marketing terms, uh you know, how do you how do you think of that? Like how do how do we translate that idea of original human thought in something as creative of a pursuit as music into something that might be as boring as a, you know, we are a big data platform that services large enterprises and makes their cost of storage less money.

SPEAKER_00 6:45

Like how do you how do you how do you just use that, print it? Um yeah, I uh it it's it's a great point. I mean it it takes um, first of all, that that huge amount of introspection. And then, you know, in the other episode we talked about the golden circle and and really soul searching about what it is you stand for and what it is you believe, and then finding a way to express that in terms of um whether it's language, um, sometimes uh the appropriate analogy or metaphor can work to your advantage, right? So an L LLM is trying to interpret what you say. If you give it a metaphor that has a whole bunch of associations with it, as long as it's the right metaphor, um, it will start drawing from that base of knowledge and associating your brand, your company, your solution with those other thoughts. So um, so I think that's what it is. It's one fully understanding what it is you believe in and what you're doing and why you're different. Um, and that part is really important too, because a lot of people think they're different from other solutions uh and and express it in a way that sort of unbeknownst to them or ignored by them. Every one of their competitors is expressing that differentiation in the same exact way. So you really really have to stress test it. Um, and then you have to use the LLMs to your advantage and think about what associations it's making, do some testing and and investigation to how the LLMs are quote unquote thinking. And I think also bringing a creative lever of your own to those especially important differentiators, right? Like what do we call it? The secret ingredient, the secret sauce, your you know, your your special purpose, whatever you want to call it.

SPEAKER_02 8:33

That may be something that you want to brand and come up with something more creatively named or unique for it, because again, that makes it hard for an LLM to summarize it away and just dismiss it as the like acronyms work great, or even you know, made-up words for features, because the LLM will, if you nail it and you get it right, it'll forever associate your company, your business, your brand with that particular phraseology because it's it's somewhat ownable content. Whereas, you know, saying I'm the biggest, bestest, fastest, you know, widget in the world, it is really, really hard to differentiate on that basis. So so yeah, I hear where you're where you're coming from, loud and loud and clear.

SPEAKER_00 9:25

Think about what's happened in in sort of um music lawsuits over the past couple of decades, right? Like we we've they've shown you can't copyright a chord progression because that that's sort of sort of free domain, and you know, that people are making music, even if it's exactly the same as somebody else. But you can copyright lyrics. You can copyright a direct experience that you express in uh you know, in in those words. And so that's sort of the same thing. It's what makes that chord progression yours is in at least in a pop song, is the lyric and the meaning that you put on top of it. Um and that becomes a differentiation. Got it, got it. Interesting, interesting.

SPEAKER_02 10:03

You know, one of the things that I I've found fascinating as we've worked on this with clients, Calvin, has been that the way an LLM interprets a specific word and how much impact that can have, or a phrase, or m really maybe most powerfully a metaphor is radically different than the way a human being might interpret the same set of things. And one of the things I really want to emphasize for everybody out there is uh let's say uh you know you hired Redline, you hired the the the best in the biz to help you out with this, that that's great, but you still need to do the regular marketing good hygiene stuff of testing. Yeah, meaning that if you don't actually test the message with the prospects, the target buyers, and even your existing customers, maybe in some cases most importantly with your existing customers, you might not ever get that kind of ground truth feedback. Because you can you you and I can figure out the tooling to make sure we get a good answer from the LLM. But if you just invent a bus and messaging in-house and never run it by some real live human beings, you're you're probably not going to be too happy about that on the uh on the other end of the experience, right? So um what have you found about how them how the machines interpret you know a word or a phrase or a metaphor versus how humans do?

SPEAKER_00 11:41

Yeah. Um, well, first of all, uh I think one thing that I learned was that shorter is better. Um, you know, when when humans communicate with each other and they're trying to explain a concept, right? They feel like, okay, let me explain it more and put more detail around it, right? We were taught that as kids. Put give me more detail, give me an explanation, give me the justification. And um what actually happens in in LLMs is every additional word is another jumping off point for it to go start another summarization, right? It's it's another potential node for it to create a branch that goes somewhere else. So you need to think very carefully about those words. Shorter is is better. It reduces the amount of drift that you might see. And then uh yeah, like I think even when we're applying this stuff to our own red line positioning, we found that the difference between the word build and create um could put you in a totally different market segment. If you put in the if we had said, you know, we we build messaging, um, it it tends to interpret that as, oh, they're a software platform or something like that, right? That's something that helps build something literally. If we say create or um something like that, that that becomes more of a service, which we are, and that puts us into a cat.

SPEAKER_02 12:55

We craft Calvin. We don't just create.

SPEAKER_00 13:03

It's a very interesting uh um nuance and and and you you you'd be very surprised at how much of a single word uh can throw an LOM in a in a large game of telephone going down the wrong path.

SPEAKER_02 13:17

Yes.

SPEAKER_01 13:18

Uh Calvin, you mentioned earlier that uh LLMs have a really hard time summarizing video and and capturing what what is in video. Is there is that just another thing that will get better and progress as as AI matures, or is there a hurdle to that that will continue to be a hurdle as we move forward?

SPEAKER_00 13:37

Um I think it'll get better, certainly. Like uh, you know, with all the things like um you know AI transcription, right? So that that's probably one of the tools that it uses is the AI transcription. It will search the text and try to pull that stuff out. Um but there's still going to be, I think, um both on the side of the LLM of um knowing that it can't convey everything. And and there there is a uh it's weird to say this about a machine, but there is a laziness factor to the LLM too. It will calculate and figure out what is the easiest way for me to get this information to that person. And the link to the video is probably it. And they put a little summary on it and say, hey, watch this video at minute marker or whatever, it talks about this thing and helps guide you to the right part. But it it it also feels like, well, if I'm going to recreate this entire video experience, I've got to generate video, I've got to generate text, I may have to generate a voice. That's too much. Let me just you know send the link. So that's it. There's definitely uh a laziness factor built into the LLM. And then I think on the other side of it, from the human perspective, um, you know, they uh there's I believe in an inherent desire to say, well, you're telling me that something's in the video, just give me the video. Let me let me just see it, right? Or maybe help me find the right spot in that video so I don't waste my time scrolling through the entire thing. Um, but we want to see it if we're presented with the fact that it came from a video. I think there's an inherent let me let me see it with my own eyes kind of um experience point.

SPEAKER_02 15:11

Well, Calvin, you've uh convinced me yet again that LLMs are kind of like bad behaving teenager um with your uh with your commentary on it. Now I know they're lazy too.

SPEAKER_00 15:21

So I've been spending way too much. I need to go out and spend some time with some humans today, I think.

SPEAKER_01 15:30

Calvin, that the engine has a human-powered story capture phase. Why is it crucial to get engineers and sales reps on record rather than just polishing a slide deck?

SPEAKER_00 15:42

Well, I I think there's two purposes for it. One is for us as Redline, in order to build the testing harness and and all the potential options for messaging and word choices, uh, we need really specific truth from the client on what it is we're trying to achieve, what truly makes it differentiated, how it works, what what what do we need to explain and how it works that makes it something you know compelling for the buyer at the end of the day. And and hearing that directly from engineers, from you know, field salespeople who have been out in front of customers, um, from you know, from the the from the of course from the marketing team as well, what works, what hasn't worked, um, that that's extremely important for us to capture. And we build that into a rubric that then helps gauge and measure and test and and you know come out at a number that at the end of the day, a single number that can can help encapsulate how well your message survived this whole process, excuse me, of uh of AI summarization. And then the secondary effect of that is that when we go through all that and we do it all, then when we hand it back to the client, it looks like you. You you recognize what it was um that put in and and you know, maybe maybe the words are a little um, you know, maybe convoluted or drier uh for the LLM to be able to understand and interpret and summarize um correctly. Um but there is that element of truth that the the customer will recognize and that that gets carried through um into the message. Um, you know, it it's too easy, and because believe me, I've tried this to to go and just pull out the marketing material from uh from a client's website and say, okay, you know, build me a testing harness that that tests whether you know our our LLM targeted message is is accurate. And what you end up getting out at the other end of that is um, you know, it won't be wrong per se, because it it's taking the the existing marketing message, but it doesn't have the the the germs of that that human truth, the the why they did it. Um and more importantly, it won't have the strategic direction that that organization wants to go in, right? So we can only measure um from existing materials where you are today. If you're trying to shift your message and if you're trying to break into a market, um, you know, that's stuff that can only come from the leadership and from the engineers, and and uh it's not going to be found in in sort of the the common marketing materials that you would get today.

SPEAKER_02 18:20

So, Cal, I've got a quick commentary for you there. So I my my oldest son is extremely literal. And I I I I both love that about him and it drives me crazy at the same time, especially for somebody as uh talkative uh as as I am. So he's almost like a human LLM. You know, he just like just give me the packs. Kind of a thing. But I as a result of all of that, I've I've ended up coming to the conclusion that to your point earlier, that the the shorter is better. There's there's no doubt about that for the LLMs. But you also you really do have to hit that authentic truth. And I'm bringing it back to that because you talked about like engineers on video. And if you want to get the sort truth, just you know, let an engineer talk. Um, as my very literal uh son once told me, Dad, so let me understand this. If I get if I process this correctly, marketing is lying. And I was like, I'll you know, no, hopefully, hopefully it's not that bad.

SPEAKER_00 19:37

My wife is a marketing professor, and uh you know, a lot of times in her intro marketing quest classes, she'll ask the class, like, you know, so what do you think marketing is? And and one of her favorite answers from a student was, oh, marketing is getting people to buy stuff that they don't want or need.

unknown 19:54

Yeah.

SPEAKER_02 19:54

Well, there you have it, right? So the but the reason I bring all this up is because even in that exploratory work that you and I do, yeah, it's great to get a company's their standard pitch deck, here's their marketing or messaging source documents, here's all their sort of standard fair stuff. But but even with our clients, when I find that jewel of a video that was recorded by their CTO that the marketing person wishes never saw the light of day, that's often where I actually find what that company's secret ingredient really is. Because even though that engineer may not know how to articulate it to the market, it's in there. Like they do understand it, they do get it. They just might not know how to put a pretty wrapper and a bow around it and and tell a story around it in a compelling way. But it's in there, right? So that getting that authentic content, especially from your technologist, is is a critical part of the authenticity engine recipe.

SPEAKER_01 21:01

So, Mick, walk us through the synthesis phase. How do you take that raw human emotion and scale it to without losing the magic?

SPEAKER_02 21:09

Yeah, it's a great question. And, you know, it takes a fair amount of analysis, is the is the truth. So uh, you know, I just got done telling a little bit of a story about working with engineers and making sure you you you kind of find those core nuggets inside what it is they have to say that might very well unveil what really makes a company's products or services unique and differentiated in the marketplace. So, you know, a big part of it is taking in information in both structured and unstructured ways. So it it redline our methodology with every client. I have kind of a really simple, I just use Google Forms intake form where I ask every member of the executive team and anybody else in an extended team they'd like to participate, a core set of questions. And they range from things like, you know, tell me what makes your company special or different and why you go to work there every day, all the way through to personality type questions, like what makes you tick, what makes you great in the workplace, those kinds of things and everything in between. I try not to make them overly onerous. It usually takes about 45 minutes to an hour to complete one of these intake forms, but that's a combination of structured data where I'm asking literally yes, no, or grade this on a scale of one to five kind of questions. Or in some instances, they're much, much more subjective. And I just want them to write a whole paragraph on what I think what they happen to think is the most valuable feature in their product offering or who their competitors are. And what once I've gone through that intake process, I've done a bunch of interviews, usually just over Zoom or Google Meet or whatever it happens to be. We take all of that information, we combine it with information that has maybe been submitted to us in a more formal fashion, marketing and sales decks and so on. We try wherever possible to talk to real live customers and business partners. And in effect, we feed that all into one big data stew. And then we use a collection of LLMs. This is part of our secret sauce. We don't just use one, we use several very intentionally. And then we have one favorite. I won't name it on the podcast here, that we use to summarize the summaries. And that is a really powerful way for us to do that initial intake process for the authenticity engine. So it's a combination of structured and unstructured data, data from customers, partners, and of course from key employees in the business as well. So hopefully that that kind of gets at the answer. I miss anything there, Calvin. You uh, you know, you you you know this process as well as I do, if not better.

SPEAKER_00 24:08

Yeah, no, not at all. I I think um uh as we what we just described, just to kind of wrap it all back up, right, is we we said when you work with AI, you work with LLMs, you have to have that source of human truth. And and and that's exactly what this is. This intake process is getting at that human truth. And um, and yeah, from there, yeah, you you can use the LLMs, but you know that you've started from a place where there is that authenticity, and then you can scale it using the guidance and the guidelines that you you put and the rules and and all that that you you put into the LLMs. And uh, and then that becomes a much more quickly iterative process as long as you've got that correct starting spot.

SPEAKER_02 24:51

What's what's fascinating too, Calvin, and I know you and I get to get to see this, and it's what's one of the reasons it's so critical that we have C-level, ideally CEO level participation in our process, is that we always learn a lot more from the differences in the opinions than we do from the things that are the same. And oftentimes we walk into a company and the CEO is like, I've got everybody on the same page. They're most definitely singing from the same hymnal. And I'm like, okay, well, here's the verbatim. And here's your head of product, your head of sales, and your head of marketing describing your three most differentiated features. And you got nine different answers there, cowboy. Um, so you know, it it is not the same. You're not singing from the same hymnal yet. And if you're not singing from the same hymnal internally, as my friend Stan Slap would like to say, if you can't sell it inside, you definitely can't sell it outside. And so that part of that process of taking in all that insight is to expose to companies that while they might think they're aligned around their core storyline and narrative, in actual fact, more often than not, they're they're not quite as aligned as they think they are.

SPEAKER_00 26:11

I I have a favorite story from one of our clients and going through this process, which was when uh so you you collected all that information in the survey, you did the summary of the summaries and you you read it back. And um it's at that point that the CEO heard something and said, Oh yeah, in a very vulnerable moment, which I loved, right? He's like, that's probably my fault. I I beat it into these guys that I didn't want that phrase theology, or like I was you're going against that. And um, and so that's why everybody answered that way. But now that I hear this back, like that may not be the right direction, right? So it was sort of that having that that little bit of uh the mirror reflected back at him, you know, helped him um also define what he truly thought was authentic, or or maybe the over influence that he had on how everyone was thinking and and how it was maybe distorting stuff.

SPEAKER_02 27:00

Yeah, that's a great point, Calvin. I uh I I remember that one vividly as well.