Curtis Sparrer

Episode 40: Curtis Sparrer

If AI Thinks Your Company Is Dead, You Have a Problem

[00:00:00] AI Realized, the podcast about everything that is new, now, and next for enterprise executives deploying AI. Hosted by myself, Christina Elwood, and my collaborator, David Yakobovitch. Our guests will share what's driving AI adoption, use cases, and business models for the data and AI economy. AI. Where the

[00:00:37] future is bright. AI.

[00:00:42] Welcome to the AI Realized, a podcast for enterprise executives leading AI adoption. From tackling security, data, and operational challenges to navigating organizational transformation, AI deployment offers a unique opportunity to redesign our organizations from the inside out. [00:01:00] I'm Christina Elwood, your host for today's episode, and we're talking today with Curtis Sparrer, principal and co-founder of BoSpar PR and a member of the AI Realized leadership team.

[00:01:11] Welcome, Curtis. Oh my God, that is the most official thing I've ever heard. I am-- Not only feel welcomed, I feel a little bit like, "Oh my God, am I in the right place? Will they have someone like me?" But so I, I feel welcome, so thanks. Oh, wonderful. Well, just to provide a little bit of background, besides being just a charming and amusing fellow, Curtis has represented brands from PayPal to Tetris to the alien hunters at SETI Institute, and he's the bestselling author of "Game Face: Becoming a PR Detective," my favorite gift book for young adults.

[00:01:47] A former newsroom journalist and current president of the San Francisco Press Club, Cur- Curtis recently led our AI Realized executive roundtable on the discoverability shift, how GEO, AEO, and SEO are [00:02:00] shaping the way brands get found in the AI-first world. That's exactly where we're headed today. So let's dive right in.

[00:02:08] Curtis, you've recently facilitated the roundtable on discoverability. It was a room full of senior leaders working through these issues in real time. Let's set the table for our listeners. Give us the thesis. What is the discoverability shift, and why should a Fortune 500 CEO be treated, be treating this as a board-level issue and not something that just belongs in marketing operations?

[00:02:30] Back in the olden days, uh, there was Google, and Google was great. And eventually, there, there was AI, a- and AI was great, and as people started using AI, uh, they started thinking, "Well, do I really need Google?" Because you could be a lot more specific with your search. And this is especially true in B2B, where you're not just wanting to have something, you know, like a broad category, like maybe books or [00:03:00] clothes or flights to places, but very specific solutions.

[00:03:03] And as we have seen, generative engine optimization and answer engine optimization have become critical for brands that are wanting to be discovered. And the flip of that is that if you're not showing up in people's AI queries, you can be good as dead. We had one company come to us because in part, ChatGPT and the other engines were saying that they were no longer operational, and we had to figure out how to correct that narrative because there's no 1-800 number for Sam Altman.

[00:03:41] And we had to convince all the aid engines that no, no, it's very much alive. And that's not an easy fix because a lot of times marketers especially think, "Well, I'll just change something on the website," and that's not enough. And you can't just, you know, send a note to one of the, uh, [00:04:00] AI engines and say, "Nope, we're alive," or, "You need to correct this."

[00:04:03] You need to do something totally different. And so now more than ever, executives have come to realize how important it is to be found on generative engines and how PR plays a critical role in that. And so that's pretty much, uh, where we are. And so during this, uh, roundtable of executives, we were all talking about where we were, where we saw, you know, success, where we saw challenges, and where people were really double-downing an investment.

[00:04:41] Well, I think it is really interesting, Curtis, that we are living in a transition era where we still have Google, and there are still many people who use Google, so your website and your SEO and other work that you do to be found in a search engine still matters. But what it takes to [00:05:00] be, uh, represented, let's say, um, in a way that you want to be represented by an LLM is a very different strategy.

[00:05:08] When an LLM restates your brand in its own words with no visual identity, no website, no tone of voice, no brand guidelines, what does brand actually mean anymore? And where should enterprise brand investment go so that, um, the, the brand is being shaped but not prescribed, I guess is how I would put it.

[00:05:31] It's super self-interested of me to say PR, but-

[00:05:38] earned media or, you know, in layman's term, journalistic news reporting is how AI understands the world at large. And when you think about it, it does make sense because any marketer could put up anything about what they want to say, and AI is not really [00:06:00] doing its job if it just goes with what Curtis Sparrer writes about Bosbar, for example.

[00:06:06] However, when there are other known respected entities reporting on me or Bosbar, for example, then the understanding of what either entity is is a lot more easy to corroborate. And what we've discovered is that AI, depending on the engine, has preferences on different media outlets and how it understands things.

[00:06:33] Now, it's not to say it only goes to journalists. There are other sources it cites. It's big on Reddit, for example. Some engines really love Wikipedia. And the real challenge is that back in the good old days of, you know, SEO, you basically had a monotheistic religion where people were going to Google and the Oracle Google would come down and FTO [00:07:00] e-experts would chase themselves silly after a new edict from Google happened.

[00:07:03] But now we have these eight different companies with eight different priorities, and that means that there are a new rules that like a polytheistic religion would require. Some will cr- require a virgin sacrifice, others are going to just ask that you be nice to people, and some will prioritize Reuters and Axios, and others will be excited about Wikipedia.

[00:07:28] And so for marketers, this is critical because they have to keep abreast of all the different engines. For PR organizations, this adds yet a new angle of relevance to what they're doing because if, uh, a journalist is reporting it, that's going to help shape a company's narrative. And as someone who is at the very center of this, like me, I can tell you, uh, different [00:08:00] examples where new business prospects came into me saying, "You were the only thing that came up when I typed in my laundry list to ChatGPT about what I was looking for."

[00:08:12] And so that's really what a lot of brands are looking for, is they're looking for them being, you know, top of the list at complicated queries and any requests. And the question is for any, you know, brand leader is how do they make that happen? How do they appear? And some of it, of course, is what I've just described.

[00:08:34] The other part, of course, is the backend stuff. A lot of people, for example, don't prioritize- Their home or about page, thinking that, you know, that's kind of as basic. But for AI, that's profound. And so that's getting a fresh look that had never been looked at before. And so I think that we are still in the early days.

[00:08:58] One journalist described me as a [00:09:00] GEO expert, which I laughed at because, like, this has just been a few years coined, and it's now just coming into parlance right now. And to use the word, to bandy the word expert, that's a little bit fresh. And so I think we're still all getting our sea legs, and that was really what I thought of when I was at this panel, is everyone was talking about things, about what they were doing and experimenting with.

[00:09:25] I didn't feel that anyone was displaying a lot of confidence about what they were seeing and what they were doing, and I think that when it comes to the tools, yeah, people are using certain tools, but I don't think there's a love affair with any tool. I think that people are being pretty polyamorous with what they use.

[00:09:48] And so I think that there is still a lot of opportunity to figure this out and to see what really is going to work. But to your central question about what does a brand [00:10:00] mean in the age of AI, it's a yes and. Yes, it's everything that a brand has always been before, but it's also what it is to AI and how AI is perceiving it.

[00:10:15] And so there is a gold rush of activity to win the AI battle, and there are a lot of different ways that people are doing this. Well, so I, I feel like there's a, um, advantage that brands that have a lot of established history and a lot of earned media under their belt, as opposed to, say, a challenger brand or a startup who has little history and little content in the world, auth- authoritative or otherwise.

[00:10:48] What do you advise for the companies that are not, uh, ubiquitous brands when trying to establish themselves in an authoritative and accurate way in [00:11:00] the world of the LLMs? I, I, I feel like I'll be super disappointing, and I'm okay with that. I've made my peace with that. But the fundamentals of a good PR program are still true today, and I think that you need to start with the right story, and you need to start with the story that is going to interest journalists.

[00:11:33] Now, let me add, though, that there is a wrinkle to that- You might want to add different sort of priorities where you might be prioritizing, for example, Reuters. And one of the reasons you would consider Reuters is they are international, but they're also not behind a paywall. And so that's [00:12:00] easier for a lot of different AI engines.

[00:12:03] And so having that conversation of is the paywall going to help, does this publication have a relationship with, uh, Anthropic or OpenAI is now an added wrinkle if you're going to be doing this. The other thing that has really kind of changed is how people are looking at press releases. Before, it was kind of quaint.

[00:12:25] We would think, "Okay, so what journalist would cover this? What outlet would cover this?" And frequently there was a thought of, "If we can't think of a journalist to cover this, should we even do this?" And if we couldn't think of a journalist who would cover it, sometimes we wouldn't issue the press release at all.

[00:12:48] And so that was something that was a discussion that a lot of us had. Nowadays, that's not the case because we know AI is reading [00:13:00] everything. And because AI is reading everything, the idea about why not put a press release becomes obvious. Mm-hmm. And then the question then becomes do we pay for the press release to be syndicated on the wire services or not?

[00:13:15] Now again, back in the old day, you know, we'd have a kind of a nuanced point of view and we're like, "Eh, why do we need to do that? We can just put it on the website. That's fine." You know, we would only put press releases out there if they actually, if there's some sort of government requirement, like it was an M&A issue.

[00:13:33] But now with AI, if you put a press release and put it on the wire service, that's authoritative. That will actually track in certain results. And the thing is is that AI values consistency, it values consensus, and press releases do that. And so what was considered perhaps a tactic of a [00:14:00] bygone era is now hot again.

[00:14:02] And so that's what we're looking at now. Mm-hmm. Well, if you were looking to move the needle and get cited on LLM, how would you rank things like structured data or authoritative backlinks, original research, Wikipedia, Reddit, any, any of the others? What are your top-rated items for, and what are the overrated places for having your, your content and your message show up?

[00:14:26] Let's go for the most controversial. Let's go for Reddit. Go for it, man. We love controversy. I think there are plenty of articles about how Reddit is really kind of having its moment in this new AI Wild West, and of course it should. There are lots of fascinating rabbit holes you can go down. The problem with Reddit from a purely a comms point of view is that it's an up or down vote, and a lot of people don't have any tolerance for [00:15:00] marketing malarkey, for lack of a better term.

[00:15:03] And so if marketers are gonna go all cocksure when they go into Reddit and put out content that doesn't read the room, that could be a real problem. And so I recommend when you go into Reddit, you do it with the Kierkegaardian-level fear and trembling, because you are entering a space where everyone has a vote, and if you are not, uh, really reading the room, it could end up very poorly for you.

[00:15:34] Mm. That's not to say I think you shouldn't do it, but you should do it in certain ways, and you should experiment first so you know what you're getting into. And I don't think I'm breaking any news by making this. I think that certain people, though, get very emboldened by this, and, uh, some people, through just sheer exuberance, don't really read the room and then end regretting it later.

[00:15:57] I would say that Wikipedia [00:16:00] has its pluses and minuses, of course. You had the wiki editors. You had the fact that certain things are instantly updated on Wikipedia, like if there is a, uh, celebrity, for example, who does anything, that gets immediately updated. If you have a product announcement or something like that, it's a different story.

[00:16:22] It w- could take forever. It may not happen because the wiki ed- editor may not deem it, you know, sufficiently notable. And there is a whole process and a science, uh, with updating Wikipedia as well, and that's why Rockipedia is now there to take advantage of, you know, the frustration in the market for that.

[00:16:42] Mm-hmm. I would say that the media appetite for news is still probably your best bet. I think that, to my point earlier about the fact that AI prioritizes [00:17:00] consensus, is a huge thing to consider. And so if you have a press release and you've timed it out with journalists you've pitched, and journalists write a story and your press release goes out and it corroborates a lot of the details, I think you're gonna be in a lot stronger position from an AI point of view than you would otherwise.

[00:17:22] But I think that it-- the West is still being won here, and I, I think that- What we are saying right now might be deemed quaint, not just in five or 10 years, but maybe even five months, maybe five weeks with the way things are going. Well, do you, do you have the same... Let me ask it, a- ask this question again 'cause I, I brought it up a minute ago about startups versus established brands.

[00:17:45] Do you feel as the, the approaches are the same, and it's a somewhat level playing field because a lot of the information is temporal, or do you feel that established brands have a real leg up, and that startups or younger brands or less well-known brands [00:18:00] have to do more or do better or do differently?

[00:18:04] The reason why brands who are either startups or established have kind of a equal playing field is that there's always been that sort of equal playing field in the arena of journalists. Journalists are either looking for interesting companies that are startups that are gonna change the world, or they are looking at the big behemoths.

[00:18:29] The big behemoths, the big enterprises have a lot of different rules that make it harder for them to react as fast as startups. And so I don't think that the playing field has fundamentally changed when it comes to how you can be perceived on LLMs. I think what has happened is that LLMs have created yet another, uh, arena for that battle [00:19:00] to play out, and I think that battle of how it plays out is actually more of a scoreboard on how public perception is going, rather than just something that can be easily gamed.

[00:19:15] Because based on the conversations we were having, Christina, it didn't feel like anyone thought this could be easily won. It felt like there were certain improvements that they could make, but everyone is still trying their best, and they are still discovering new things. I think my favorite thing in that discussion we had is one woman who revealed that she put out a press release and got 20 leads from it, and there was nothing revolutionary in putting out a press release.

[00:19:49] They've been doing that since the 1800s. And so I think that a moment where, yeah, in the AI era a press release can lead 20 [00:20:00] business leads is a bit where everything old is new again. Mm-hmm, for sure. Well, give me a play that you actually ha- uh, ran for a Boz Bar client in the last six months that simply didn't exist as a tactic two years ago.

[00:20:17] I think that when we're looking at tactics, I think that most of the- Work is, I hate to say it, principally the same. And I know someone's gonna come up after me and say, "Oh, I have a tactic," and maybe I'm just not thinking this through. But the tactics are the same. I think what's really net new is the measurement and the consideration.

[00:20:44] I think that when we are looking to see, well, how well did a campaign go, well, we now have a new way of looking at it that we never had before, which is, well, what is AI, AI saying about it? How are we standing or stacking up against our [00:21:00] competitors? Mm. Uh, there are times where we are optimizing our content for AI, and that's been a work in progress since we really started talking about ChatGPT, where we've been looking at press releases and making sure they're readable, that they're scrapable, and that comes also with other content, and that's really made FAQs more important than ever before because FAQs are a way that AI understands things.

[00:21:34] But it's not like I'm in a meeting saying, "You know what we need? We need a new FAQ." Mm-hmm. It's more of a sense of, you know, these are the best practices, and we're always iterating and improving upon them. Mm-hmm. Okay, okay. Well, in this next segment, I think, um, our audience needs to hear a bit about a scenario that is maybe keeping, uh, general counsels and chief comm officers up at night, and that's LLM [00:22:00] hallucinations.

[00:22:00] LLMs hallucinate sometimes materially false representations. Can you take us through a playbook for when that happens, what actually works to correct the record and what doesn't, and how fast, uh, the needle can be moved? So the example that we've talked a lot about is RealSense because RealSense had a very real moment where ChatGPT, Claude, and others said it was dead.

[00:22:27] And there's something wonderfully binary about that. You're alive or you're dead. It's a one or a zero. And so this is something that is correctable insomuch as if the AI is saying it's alive, that's not a zero anymore. Mm-hmm. So- And so, uh, you know, just to, just to make sure the audience is, is grounded here.

[00:22:48] Cl- RealSense is a, a company that was spun out of Intel, um, and so it, it existed as a brand before the spin-out. Is that right? It existed as a brand before the spin-out, [00:23:00] and what happened was that there was a report that Intel was discontinuing some of the, you know, various product lines, and for whatever reason, that was conflated to RealSense, which does robotic eyesight, being shut down.

[00:23:22] And so this became a b- Big problem for them because all of a sudden everyone was not seeing them in searches, because you're not, like, looking if a company's alive or dead. That's kind of like, I don't know, macabre. What you're looking for is, "What are the companies that can help me with this problem?" And if AI thinks you're dead, you're not coming up in that list of results.

[00:23:47] Now, that's a huge hallucination problem, and it's not, like, by, you know, varying degrees. That is an actual binary thing. And so we worked [00:24:00] through the steps to stop or to change the hallucination, and when we challenged ChatGPT and Claude on that and said, "Hey, why did you do this?" Both of them were defensive and said, "The reason why we did this was based on the existing information.

[00:24:17] The company, however, is very much alive. We got it wrong." And the thing that bothers me the most about this is that algorithmically all these engines know how confident they are in the information they're giving you. So I don't understand why these engines don't just say, "I have 50 degrees, percent confidence here," or, "I have 80% of confidence here."

[00:24:43] Even the town gossip knows to say, "This is what I heard. This is what I think. I'm not so sure it's reliable because so-and-so is always tipsy when she's doing blah, blah, blah." But AI is never that forthcoming. It's always like, "Here's the results." And so the [00:25:00] hallucinations are a corporate vulnerability that has to be solved.

[00:25:08] And that's the- Yeah, well, LLMs certainly have no corrections desk, so there's no one to call. There's, there's no way to just say, "This is wrong, fix it." So how do enterprises build reputation resilience under those circumstances?

[00:25:23] I think the most important thing is an audit, and the audit's where you start, and you get your baseline there. Mm-hmm. And from that audit, you then have to determine, okay, what are the things that we need to change? And, you know, where are they getting that information from? And usually a good audit you can find out where AI is drawing that information from.

[00:25:50] And some of the corrections are easy, some of them are a lot more difficult. Oftentimes, what you'll find is AI is drawing from pretty [00:26:00] outdated information that for some reason is algorithmically decided is important. And so some of those corrections could be handled simply and tactically with a press release.

[00:26:12] Some of that can be handled updating your About page or your home page, and part of that can be handled also with schema. Now- You know, the thing is, is that whenever we get into the situation, I'm not the guy who's handling the back-end programming. There are women and men who are much better suited at doing that and, you know, can talk far more eloquently about the details that are required for that.

[00:26:43] But those people who are doing that can help at the back end, but the real challenge is if you're looking for a perception shift, that's when you have to go up market, and that's when you need [00:27:00] analysts or academics or experts who really are able to shape the gravity when it comes to a brand's perception.

[00:27:11] And I think that is a place that a lot of people don't think about enough, and I think that is really the biggest opportunity in this sort of AI shootout. I love it. That's a great, uh, suggestion and, and not one that came up in our roundtable conversation. What's the consensus view right now that you think is gonna look flat-out wrong 18 months from now?

[00:27:36] Boy, we should put some music here where it shows that I'm thinking. Um- Well, I'm not gonna hum for you. Uh, I mean, you know, can you hum the Jeopardy! tune? That might be- I can't. I can't. I'm tone deaf. I'm sorry, I can't do that for you. I'm tone deaf, too. But un- unfortunately, this voguing hasn't really created a eureka moment for me.

[00:27:56] I would say that there... [00:28:00] Let me back up here. There are a couple things that I have seen. Uh, one is that outside of the Bay Area, I have not seen a lot of concentrated thinking about how brands appear on AI, in, in AI. And when I say that, I- I'm sure certain people will push back, but I... when I go to New York or Chicago, it's kind of seen as this small thing or this thing that will have to be addressed.

[00:28:35] But I don't feel the kind of white-hot urgency that I feel in the Bay Area. Maybe I'm just not being invited to the right places. But I think that if there's any kind of consensus that will need to be changed is how this is everyone's problem now, and that this for brands might be the only problem. [00:29:00] And I think that if there's anything that needs to be changed, it's perhaps the urgency of thinking about this.

[00:29:08] And so I think that when I am brought into boardrooms and such about things like this- People will confess that they haven't been, that it's been something they've been hitting this news alarm on, or it's something that they're gonna get to eventually, or they're just not sure if they need to do it, and I find that astonishing.

[00:29:31] Mm. Because this is not even the future. This is the white-hot present, and we see that, uh, an overwhelming percentage, C-levels have started an AI-first approach to major, uh, B2B purchase decisions, and we see that most Americans are using AI too for the same thing. And so the fact that a lot of [00:30:00] people would say that it's something that they're going to get to when the field is a lot more mature is something that just seems reckless to me.

[00:30:09] And so- When your customers are moving, you gotta move with them, and we're seeing that in B2B, but we're also seeing it in B2C. It was one of the things reported at the round table, that in the B2C world with, uh, the opportunity to be able to both shop and soon to buy within the LLM, it's completely changing the landscape of commerce.

[00:30:28] So if you could get every executive listening to stop doing one thing immediately, what is it? If I could get every executive listening to stop doing one thing, it would probably be to avoid my calls. Stop doing that. Take my calls. Uh, I, I think that if I could make the case to every executive, it would be to really value the human relationships with media, and I know that sounds pretty [00:31:00] Goldilocks and saccharine, but right now, what I am seeing from my journalist friends is that they're getting overwhelmed by the sheer amount of AI-created content.

[00:31:15] And every PR person in existence seems to think that if they do a deep read or analysis using AI, and then use AI to create that pitch, that that's gonna convert into a great story. And here's the reason why that doesn't work. It's because all those pitches look alike, so every journalist who is already getting hit by 100 pitches is now getting hit by 1,000 pitches, but they all seem the same.

[00:31:49] And so your reporter's like, "Delete, delete, delete." If these brands invested in human relationships with [00:32:00] media where you're actually talking to a journalist, talking, not like just emailing slop, you're gonna much more likely have that story picked up about your company. And- I've tried to explain this to a variety of different companies saying, "Hey, you should p- you should support the San Francisco Press Club or the Society of Professional Journalists," which I have a connection to both.

[00:32:25] But hey, let's broaden this out. There is a press club in New York. There is a press club in Chicago. You should support your local journalists, and in a way that's ma- you know, meaningful to them. And for those who just think it's, you know, just a transactional inter- interchange, uh, with emails and stuff like that, it's probably not gonna work because a lot of those journalists are just getting spammed right now with AI-written content that has a pretty obvious tell.

[00:32:54] And so that's what I would tell every, you know, executive. Mm-hmm. I think that's really sound [00:33:00] advice. Curtis, you've been a leader for a long time. What is the leadership skill that you find most useful in this AI era, and why? Curiosity. I think that curiosity is the best thing that can keep you au courant, as the French would say, with what's going on.

[00:33:23] And I think that executives who are not curious, who are not willing to play with the new technology are going to get fossilized. And I think for me, playing with AI has given me an appreciation of what it can do and what it cannot do, but I'm not coming at it with someone who's never done it. I think that one of the more coarse analogies I made to you during our panel was how people who really don't play with [00:34:00] AI sound like, you know, young queenagers describing sex, where they're, they're willing to bluff it, but it's clear that they have no idea.

[00:34:11] And I feel that people who don't play with AI, who don't get their fingers, hands dirty with it, who have not tried to vibe code, I think that they're missing out on the experience that helps them be a credible player in the space, a credible storyteller, or understand what's valuable and what's not. And I think that's the real thing that I would put out there, is that you need to be curious about how this technology works, and you need to test yourself every day to see if you can learn something new.

[00:34:45] And I know that sounds all saccharine and stuff, but I have seen a lot of people talk about AI by being on the sidelines, and then they can bluff it for a little bit until they can't, and then it's a huge [00:35:00] tell, and then it becomes pretty silly. Yeah, for sure. For sure. You can't be on the sidelines. And where should our audience go to follow your work, to follow Bospar, and to keep up on all of the wonderful content that you're publishing on this topic?

[00:35:16] My favorite place to live is definitely LinkedIn. Uh, if I have a moment of any kind, I wanna share it there first. And I wanna say a little something about LinkedIn. I find that the best posts are the ones that treat LinkedIn more like a fun social network as opposed to a braggy, uptight, look at me, look at me with my white paper, you know, network.

[00:35:43] And I find the more human you are and the more approachable you are, the better your engagement goes. And so the people at LinkedIn who just huff and puff with kind of the traditional front of the class, A+ [00:36:00] student sort of content bore me. I find that I'm a lot more engaged with the people who are either adding some lightness and fun to my day, 'cause we're all gonna die, or the people who, uh, have something interesting to say but aren't gonna be necessarily pedantic about saying it.

[00:36:18] And so LinkedIn is where I like to live, but we curate all our content on bospar.com. And when I say curate, that almost sounds like this, uh, East Coast waiter talk where I'm like, "That's the curated menu of spirits and bl-..." That's where all our stuff goes to live. That's where it's like the receipts are found and the record is.

[00:36:40] Okay. Well, thank you very much, Curtis Barr, principal of Bospar PR and leadership team member here at AI Realized. Thank you for making us smarter today, and we appreciate you being with us on the AI Realized podcast. Thanks for putting up with me.

[00:36:58] Where the [00:37:00] future is bright.