Latest / Expertly Said / Dan Gaynor on Using AI to Question Your Own Narrative

Transcript

Dan Gaynor on Using AI to Question Your Own Narrative · Expertly Said

• Kristen Sweeney: My guest today is Dan Gaynor. Dan is currently serving as Executive Fellow at Harvard Business School, which is a program to engage the expertise of outstanding practitioners to enhance teaching and learning at the school. Previously, Dan was the founder of KelpData, the first AI-powered platform for corporate reputation. Kelp was acquired by Signal AI, which serves 50 % of the Fortune 500 in July 2022. The company's data analytics platform delivers industry reputation, insights, and benchmarks hundreds of companies. enabling communications and marketing executives to shift their strategies, capitalize on emerging opportunities or mitigate new risks. Signal AI's technology and team are enabling enterprises to better understand where their reputation stands and how to strengthen it. Dan, welcome to the show. Dan Gaynor: thrilled to be here. Thanks, Kristen. Kristen Sweeney: ⁓ Absolutely. I am really looking forward to this conversation. You have such an interesting career journey. I'm wondering if we can kind of go back to the beginning, wherever that starts for you. ⁓ Your background is in communication. So how did you get your start there? What was it that first kind of drew you to that space? Dan Gaynor: Yeah, well, I went to college at Washington University in St. Louis and majored in political science with a minor in writing. So I was always super passionate about trying to merge the worlds of communications and politics. Little did I know that was probably one of the most dangerous and least lucrative economic paths that one can pick in the U.S. ⁓ But nonetheless, after Kristen Sweeney: Yeah. Oh, well you're talking to a musical theater major here, so you're in good company. Dan Gaynor: There should be some sort of like unification mechanism of these guys together creating their own companies. But ⁓ yeah, after the London School of Economics, where I went for grad school, I moved to DC and I was obsessed with the idea of working in the foreign policy arena. So I actually as a first job ⁓ was the digital strategist, quote unquote, at a small nonprofit advocating for a mix of foreign policy solutions. This was like shortly after the war in Afghanistan and Iraq. And then I got picked up by the Obama administration as a national security appointee for ⁓ USAID, which no longer exists. It got doged out of existence, but was the premier foreign aid agency in the world, about $70 billion to the hundred most troubled countries in the world working on alleviating extreme poverty. And my job was to be the chief speechwriter for the administrator, which was a really, really ⁓ intense and fast paced job because on one hand it could be the earthquake in Haiti and Nepal and how we're marshaling efforts to save people under the rubble. On the other hand, it could be forming public private partnerships with Coca-Cola to deliver vaccines through a refrigerated supply chain because if there's one product that has 100 % market penetration in the world, it is bottles of Coca-Cola. And then my big project was actually the Ebola epidemic, which compared to COVID was like 10 times scarier. So, you my job was to script the words that would advocate for a controlled and calm response to that pandemic, let alone advocate to Congress why we needed to educate girls in rural Afghanistan or vaccinate vulnerable populations at a time when we could invest that money in say American infrastructure. It was a really fascinating job. ⁓ From there, I ended up going to presidential campaign for Martin O'Malley for president, which unfortunately got less than 1 % in Iowa. And then I finally came to my senses and went private sector. So I went to Nike corporate headquarters out in Beaverton, Oregon, where the parking garage has a slide that you can take down to work every day. And you can go rock climbing at 3pm. And I almost played Serena Williams in Ping Pong. I had breakfast with LeBron one time. It was a real, really fun job. Kristen Sweeney: Please. Great. Dan Gaynor: We helped start a team called the Narrative Center of Excellence because at Nike, every sport is its own business unit. Soccer and football and baseball and tennis, they're all business units. There's a team now that sits in the middle of them that tries to think about what do we stand for at a global scale? What is Nike's, say, core vision of empowering every athlete mean when it comes to an innovation launch of the back to the future self-tying shoe, let alone ⁓ earnings call where you have to sit with the CFO in the boardroom and talk about managing SG &A and CapEx. Really fascinating job. I'll skip ahead, which was from there, I moved back to Boston where I'm from, took a job at Weber Shamwick, which at the time was the world's largest publicly traded marketing firm, helped start a practice called the narrative practice, if you will, where I worked with the data team and the strategist team to architect what companies stood for. So whether it was Johnson and Johnson wanting to pivot into being a tech company, or it was IBM wanting to pivot into an AI company, everyone is sort of undergoing their own transformation. And we would work directly with the executives to help figure out what is that master story, the core pillars of it, the key themes that compose it, and how to orchestrate it out to the world. And that led me to starting my company, which I'll pause right now, was about automating that process to quantify corporate reputation. Kristen Sweeney: Yes, and I have many, many more questions about ⁓ Kelp data, and we're definitely going to spend some time there. ⁓ I'm curious, in that transition from government agency to one of the biggest consumer brands in the world, what lessons did you take away or apply from your time in government and subsequently from your time at Nike? Dan Gaynor: I would say there are two big things and both relate to the idea of narrative, right? A narrative is a North Star for all communications, meaning that everything should theoretically flow from it and reference it. And the reason, point two, is that consistency is the only way for a narrative to penetrate the minds of the audiences that you're focused on. Right? There's gotta be basically a red thread where you are consistently hammering home Kristen Sweeney: ⁓ speaking my language here. Absolutely. Yeah. Dan Gaynor: the same core themes that compose your view of the world. And so just to give you a quick example, at USAID, where I was part of the Obama administration, we would consistently say the words, and if my colleagues heard this, they would roll their eyes, the new model of development. That was the master frame. The new model of development means insert initiative, right? It could be public-private partnership to create a startup incubator in Afghanistan. and it could be rolling out the Ebola vaccine faster than normal pharmaceuticals. When it came to Nike, it was about envisioning a world where everybody is an athlete, asterisk, and if you have a body, you are an athlete. Okay, well, what does that mean? That means we're gonna be expanding our sales in China so that we can focus on growing our business on global markets. It also means expanding the definition of what a top performing athlete is. It's no longer about LeBron and Serena, it's about you seeing yourself as someone who has great human potential. So regardless of where I worked, it was, okay, what's our master story headline? How does that then translate itself to multiple themes like innovation or sustainability or inclusivity? And then how do we consistently refresh that narrative with propagating proof points that make it newsworthy and relevant and broadly accessible? Interestingly, that's actually what led me to the AI world, because if we could consistently track key storylines that everyone across every industry is campaigning on, we could theoretically unlock areas for differentiation when it came to corporate comms. So I actually found the Obama world very, very relevant to the Nike world. And I'm surprised there isn't more intermingling. Kristen Sweeney: Yeah. Yeah, yeah, I love that. It's interesting, and I wonder if you dealt with this or if those organizations were ⁓ well-disciplined. But sometimes I find that there's this very human thing where you kind of get sick of your own message, and you think like, we need to change it up because you perceive it constantly, and you live it day in and day out without really having that. empathy of putting yourself in the consumer, customer, audience's perspective and saying, are, you know, a touch point every now and then. actually that repetition is exactly what is needed to drive the message home. Dan Gaynor: You know, I'm a big fan of comedians in cars getting coffee, which was Jerry Seinfeld's web series where he would rent a fancy car and pick up a comedian and they would go get coffee. And he will talk all the time about how at some point the joke is no longer funny to the comedian, right? The comedian has worked on that joke for 40 shows in a row, but the audience is hearing it for the first time. And it is so important, especially in such a fragmented, Kristen Sweeney: Yeah. Yes! Dan Gaynor: Metaphotophonous media environment for you to remember that it is all about hammering home one message five different ways every single day in and out and obviously data can help you do that more consistently and efficiently But it is almost what I had to learn when I was in the high school drama club Right when you're practicing the play over and over again and in high school drama club You know, there's like kind of one weekend you do it Friday and Saturday night You have to remember that the audience is hearing it for the first time and you need, in fact, you owe the material that energy, otherwise you're not doing service to all the great preparation and work that you put into it. Kristen Sweeney: 100 % that commitment of keeping it fresh. And as someone who's done, you know, dozens if not hundreds of performances of the single show, I could tell you there's some real shenanigans that go on to keep things interesting on stage. Dan Gaynor: I am a little embarrassed that I even mentioned High School Drama Club in front of somebody who's a real professional, but nonetheless, we're here. Kristen Sweeney: ⁓ please, that's, ⁓ you know, all my favorite people ⁓ usually have some kind of connection to the arts in some way. Dan Gaynor: Yeah, yeah, I'm about as amateur as it gets, but ⁓ nonetheless. Kristen Sweeney: Love it. Love it. ⁓ So I want to move on and talk a little bit about ⁓ Kelp Data. You know, I had to double check my dates with your bio. You built and sold, I believe, an AI company before the launch of ChatGPT. Is that correct? Dan Gaynor: That's correct. ⁓ was definitely getting a lot, I was getting a lot of raised eyebrows when we started. The idea was, what if you could quantify a company's reputation? How would you do that in an automated way? Now, keep in mind, surveys and focus groups and panels have been around since Babylon. We're always asking people their opinion. But what if you could harness unstructured mass quantities of data, for example, like media data on a global scale? Kristen Sweeney: Talk about foresight. Yeah. Dan Gaynor: billions and billions of articles, let alone tweets and LinkedIn posts and Reddit threads, you name it, segment out what the topics are in those pieces of content, grade those topics and apply them to individual companies. You could create, in other words, an index of how you are perceived on dozens and dozens of key storylines that every company is campaigning on. As an example, in the world of healthcare, Every organization from a hospital to a pharmaceutical company wants to be patient first or patient centric or putting patients at the center of everything we do. Easy Google search. Another example is innovation. What company isn't focused on innovation and adopting new technologies, whether it's Johnson and Johnson inventing better artificial knees or it's a tech company like Salesforce saying that we are going fully agentic. Everyone is campaigning on that thread. So Kristen Sweeney: ⁓ yes. Dan Gaynor: Kelp data was about automating that process, sucking in a huge amount of data, segmenting it into storylines, grading companies of those, and giving our clients a chance to understand how they might differentiate from their peers, both within their industry and adjacent to their industry. So when we started pitching it to folks, they were like, well, wait a second, I've got media monitoring or I've got surveys in the market. What's this about? Kristen Sweeney: Yeah. Dan Gaynor: It was really a leap of faith, if you will, to help folks understand there's this new emergent technology around discriminative AI, which is all about segmenting and labeling, that then with generative AI captured the zeitgeist. So I wish I could tell you that I knew ChatGPT and Claude and Jim and I were going to take off like they did. But I was ⁓ very, very gratified to see that the world ⁓ took These technologies in a place that I could have never envisioned and certainly from a communicator's lens. They've unlocked so much potential Kristen Sweeney: Absolutely. And we're going to talk about some of that potential ⁓ a little bit later. I'm curious, what came first for you? Were you exposed to technology and said, wow, what could I do with this? Or were you trying to solve a comms problem specifically from that perspective? Dan Gaynor: The answer was what I'd seen at Nike, I realized was a consistent problem only once I got to the world of consulting at Weber-Shanwick. So at Nike was my first real big corporation. And I just assumed that some of the challenges we had there with regard to keeping a consistent narrative or outflanking at that time Adidas and Under Armour and New Balance and Puma were sort of just unique to the footwear industry. But then I started working with executives at IBM and Johnson and Johnson. and Salesforce and Takeda Pharmaceuticals, I was like, wait a second, across every single industry, everyone is asking themselves how they can, A, campaign on the same storylines, and B, be different on those storylines. And so the way that we would attack it from a consulting standpoint is we would download a bunch of data, which was like a one-time, big expense. We would apply a bunch of analysts to that project. Okay, guys, go in and manually label different. pieces of content as a topic, which of course was super inconsistent because if Danny was labeling something as innovation, Kristen might be labeling something as sustainability. What if it intersected? It was about the innovation in clean energy. That would be a really tough, difficult analysis. So ⁓ what I came to the conclusion was one, everyone has the same challenges and two, everyone has the same aims. Everyone wants to win on innovation. That's the challenge. Kristen Sweeney: Pattern matching, yeah. Dan Gaynor: And everyone has the same aim. They want to be differentiated. And that's when I teamed up with my really brilliant co-founder, Shann Biglione, who at that time was chief strategist at Publisys, which is another gigantic agency. And when I first brought him the idea, I think he said, famously, this is impossible. And I remember thinking like, ⁓ great, you know, I'm not going to work with that guy. It's kind of an a-hole. It's too negative. And let alone he built it and has been like a world-class partner and a great friend. Kristen Sweeney: Hahaha Dan Gaynor: sins. So part of the AI journey that perhaps I'm reflecting to your listeners here is you got to work with it and it is more than possible with a talented team to ⁓ create something that is never created before versus optimizing a process you're stuck with. Kristen Sweeney: Yeah, 100%. Wow, I love that. And yeah, it sounds like it's a bit of both, right? There's the coms problem piece, but also seeing this manual and saying there's got to be a better way to do this and get this done. Dan Gaynor: Yeah, having one guy comb through a spreadsheet and saying, think this is about innovation is pretty, I don't know, stone tablet. And we're trying to move things towards MacBook Pro. And you know, there's so many concerns around, ⁓ you know, what would my job look like if I weren't doing this process? And what I've been able to see firsthand, having worked now with like 100 plus organizations on quantifying their reputation through the solution I helped build, though certainly didn't individually build, is that It unlocks their ability to be more strategic. gives them validation for their gut instinct. It allows them to have more sway, if not give a sense of direction to other functions. I've seen that if you can quantify corporate reputation, what you have the ability to do as a communicator is walk down the hallway and say to the operations team, the finance team, the marketing team, the HR team, you own a piece of this too. Here is my direction to you in terms of the the you're putting out, the initiatives you're pushing forward, the investments you're making, the folks that you're hiring. You, in other words, have the ability to quantify a company's most valuable asset, which according to Harvard Business Review, reputation composes like 85 % of a company's market cap, not necessarily the value of their products or what's in their warehouse. So for me, it's been really, really gratifying to see actually as you decommission less efficient processes. and you are able to better quantify the impact that you are making on the company, if not the industry and the world, you then have the ability to operate at a way higher strategic altitude. So that's been super gratifying and fun to help build. And now there's so many new applications of AI ranging from risk monitoring to synthetic audiences for message testing that I'm interested to explore next. Kristen Sweeney: Yeah. Yeah, yeah, I definitely want to chat a bit about that. What I really like about what you've described here is that it's validation and it's data, but it's also including the qualitative, right? It's not just measuring clicks and impressions and these things that are signals. It's actually pulling kind of the language itself ⁓ and evaluating that way. Because I think that qualitative piece is so important. We can't always just base on the, the actions we take online. ⁓ Those are representations of how we feel, but words and language can often capture that in a different way. Dan Gaynor: That's exactly right. Since this is a podcast, I guess I should go into hot take mode. So here's where we go. I am convinced that we have now gotten all the vertical data that we need in comms, right? So I think any communicator understands, okay, I get it. I've got to have data for my job, the channels I own, let alone the organization as a whole, which could be, you know, my business unit or my team, et cetera. Kristen Sweeney: ⁓ yeah! Dan Gaynor: But I think we've actually reached peak data saturation in the world of comms. You referenced a lot of them. Impressions, which will tell you that two billion people read that story about you in the New York Times. No, Two billion people get access to the New York Times writ large. The actual readership of that coverage is tiny. We might think about likes, retweets, comments. Well, we actually know that on social channels, it's only 10 % of the users ever produce content. Most of us are passive readers. Certainly I'm guilty of that. Like I go on Twitter. I haven't tweeted since I took a Euro trip in college. And you can just go down the list. The bigger issue in fact is that as we've verticalized these data sets and comms, the data sets don't talk to each other, right? The social team is like, look at the amount of likes and retweets and comments we got. The own team is like, look at our Google analytics. People are spending four minutes reading our blog posts. And then the media team is saying, look, we secured Kristen Sweeney: Yeah. Dan Gaynor: all this incredible top tier coverage, we got that piece in the financial times or we've had great message penetration, none of them connect to each other. And so where I get really excited in using AI for comms is going from vertical to horizontal, creating horizontal layers that everyone can use. As an example, a synthetic audience might be someone that the media team targets, the social team targets, the thought leadership and own channel targets. That could be somebody like in the world of pharma. patient advocacy groups. You'll have different ways of referencing our earlier point, transmitting a narrative to them, and those different ways will of course be different depending on the channel, right? A tweet is gonna be really short, a long form video is gonna be a big YouTube play, thought leadership from your CEO at the JPMorgan Healthcare Conference, et cetera, et cetera. We should be looking at technology through the lens of how can I create a horizontal system that everyone can use versus everyone gets their own sort of golden honeypot of data that justifies their jobs, but ultimately doesn't aggregate up to the overall impact. Kristen Sweeney: Right. 100%. Yeah, yeah, I love that. And that's a really, it's a really exciting vision. Dan Gaynor: Yeah, well, we're working on it. Kristen Sweeney: I want to go now to what I imagine was a pretty big kind of maybe professional, maybe personal transition point for you, selling kelp data. ⁓ Can you just talk a little bit about that experience? And I believe you were involved. You stayed on with the company for a little while, at least. ⁓ But I'm curious. How that felt, we talk with folks on here who talk about different milestones and transition points in their career. I imagine that was a pretty big one. What was it like to kind of say, what's next? Dan Gaynor: Absolutely, yeah, first time founder, ⁓ very intimidating to go through the whole ⁓ &A process. ⁓ It was really exciting when we had a few bytes and Signal AI ended up being the folks that acquired us. Signal's a now over decade old AI company that was founded in London and was interested in working with us, not only for our technology, but really to move up within enterprises, right? Media monitoring is kind of seen as a commodity and what we were doing was working with executives at a more strategic level. Kristen Sweeney: Sure. Dan Gaynor: bringing those two forces together, their side on the tech stack and the data and our side in terms of transforming that data into insights, we thought was a good team up. So we were their first acquisition, Signal's now gone on to acquire a few other companies and actually they got acquired themselves by a big PE firm. So fascinating to be part of that journey from first acquisition to the holding company then getting acquired itself. You know, I'll tell you, I think the tricky thing is maintaining the special culture that you build and a culture can be Kristen Sweeney: Sure. Yeah. Dan Gaynor: just you and your co-founder or in our case, I think we had like 10 employees when we came on and sort of had this special unit that by definition is supposed to, and I mean this in a good way, like propagate through the bigger company. At the same time, the bigger company has their ways of working, their sales processes, their communications processes, their own toolkit even that we had, for example, moved from PowerPoint to Google Slides. All of these things are really difficult. And the main lesson I learned is that once you get acquired and you're operating in an earn out or some sort of incentive structure, you need to really equip your team to be emissaries for you in the organization. Because as the founder, you're gonna get pulled in all these different directions. They're gonna ask you to present to their investors. They're gonna ask you to be on a ton of sales pitches. They're gonna be asking you to integrate your product with theirs. And I had this incredible core team, shout out to Michael and Kara and Abby and. Liz and Biagina and Jillian and Britt, had this incredible team that was actually in the work every day working with the rest of the bigger company to ensure that there was a smooth transition. Make no mistake, it took about a year, but in the almost four years I was at Signal AI, we went from a small, fledgling, interesting unit to almost like half the business. And I'm incredibly proud of sort of the internal cultural dynamic that ⁓ probably was as to our integration as the actual technology and product and quality of what we were providing itself. And I think, by the way, that's a lesson that's true in big companies as well. If you're listening to this and you're at like a big corporation and you're starting a new team, please do not underestimate the power of like buying people green tea every single day, asking what are you focused on? How is your career trajectory unfolding? What are the incentives you're held to? What does your boss want to hear? And figuring out how to retrofit what you are doing. into their job versus saying like, is the thing I'm doing, I'm busting through walls, you can't stop me. And that was a big lesson that I've learned where if I were to ever build anything again, which I'm certainly thinking about in any type of organization, big company, all the way down to two guys and a laptop, it's really about the human element when it comes to integrating two cultures, two products and ⁓ building technology together. Kristen Sweeney: Yeah. 100%. This is a theme that's been so ⁓ prominent for us and our client work lately. A couple of engineering firms we're working with and you know, the work could not be more focused on digital and AI and this kind of transformation. And the theme that we keep hearing from the execs we're working with is just like, it is all about people, people being willing to change, thinking about people when you build these processes. It is about so much more than. you know, the technology itself. Dan Gaynor: Right, and there's a reason that OpenAI and ⁓ Claude, it's Anthropic, are investing a ton of money in building out consulting teams. Like think it is a sort of fascinating dichotomy that as we're so worried about automation and job loss, there are real issues with regard to enterprises really adopting AI in their workflows. And what that means is a lot of communicators are like using ChatUpt and Claude and ⁓ Microsoft tools to like summarize emails. They're certainly not using them at the zenith of what these tools can be used for truly like automating low value time intensive tasks that never should have been on their plate in the first place. Where I get really excited with the potential of AI is that communicators as much as any corporate function have the ability to put a tangible footprint on AI to shape AI in their workflows from crafting media messaging. to simulating audience reaction, to sharing insights in a more collaborative and aggregate impact style. ⁓ So I get really excited with the idea of AI being a way to automate processes that make you a more strategic communicator. And I think that communications more than marketing, finance, operations, HR, or any other corporate function will be sort of like at the vanguard of shaping these technologies in its image. You can tell I get pretty excited and that means... Kristen Sweeney: ⁓ yeah, no, that's quite a vision. And that's, think, a perfect segue for us to dig a little bit deeper and spend some time there ⁓ in our one big idea. What do you think it takes to, you know, I think hear a lot about, yeah, like someone's on an LLM and they're kind of using it, but there's this feeling that it is surface level ⁓ or certainly not leaning into the full potential of the technology. What's a communicator to do who maybe doesn't consider themselves a technologist? And they say, how do I really get the most out of this technology? ⁓ One of the things I reflect on is that sometimes the things that it like I find its best-ed are so unsexy, right? It's like they're not the glamorous things. You can call them transformational if you want, but it really is like, I didn't have to copy paste that four different places. ⁓ So yeah, I just want to put to you, what's the maybe frame of mind as someone who obviously has a track record of being very future-focused and kind of thinking beyond what's the kind of current boundaries of my scope and my capabilities? How can communicators think in that direction? Dan Gaynor: Yeah, so as you noted at the beginning, I'm now an executive fellow somewhat preposterously at Harvard Business School. And one of the things I always pick up on in working with academics who think I am nowhere near detailed enough to be doing academic research, I should just note that, is you've got to break a process apart into its core components. And the reason I think AI is so applicable to being tailored by communicators is that they... natively by the way that they work have the ability to think about their workflow in terms of core components. So what do I mean by that? I mean that think about you're in any communications job. First and foremost, you've got to think about, okay, component one, who's my audience? So for any narrative pillar that I'm going after, let's say it's innovation, well, I'm probably aiming the themes of innovation. Let's call them quality, automation, technological revolution, whatever. I'm aiming those probably at more technical audiences or more executive audiences. Okay, so I need to understand that audience really well. AI can certainly help with that, but I'm going to skip ahead. Then, okay, what channels do those live on? Well, the executive audiences, it's probably more of a conference play. The technical audience, if we're recruiting, say, like developers and engineers, we might be looking at different HR sites than like Glassdoor. We might be, say, oriented towards going after colleges like MIT. We may also be putting out more ⁓ short form content on the TikToks and YouTube shorts of the world to get the idea. Okay, then from there, it's like, well, what's the right message to translate through those channels towards those executive or technical audiences? Well, AI can be really helpful in that. It would allow you to co-create content with synthetic versions of those executives, of those developers, those engineers that you want to communicate to. So we're... I think about AI being relevant for communicators is break apart the core components of how you think, that you sort of just take for granted now as a communicator. What's the right audience? Where do they live? What's the right channel? And what's the right message? And AI is as equipped as ⁓ to help, it's equipped to help communicators in as tangible way as possible. That's not the same thing for finance or operations or HR. They don't have a workflow that lends itself as well to AI. Kristen Sweeney: Yeah. Dan Gaynor: So the first step I would say is map out your workflow into the core components of audience, channel, and message, orchestrate of course the narrative pillar that you're working within. And that's step one. Kristen Sweeney: Yeah, I love that. You've ⁓ mentioned a couple of times synthetic audiences. ⁓ And I am lightly familiar with this concept, but I'd like to go a little bit deeper and learn more about it. Because my sense is this is, I think early in the kind of prompting days of LLMs, was like, act like a so and so. And then it was just kind of baked into this prompt. But what we're seeing more and more is deeper specialization in terms of. ⁓ different platforms or agents, I might be using the wrong term, there's sort of a specialized context built in. So when you talk about synthetic audience, ⁓ it's probably a lot more than saying, pretend like you are a healthcare provider, ⁓ give me a response. Dan Gaynor: Yes. I am so glad you framed it that way because I talk to folks, I probably talk to a chief comms officer every day of the week from the Fortune 1000. And I can't tell you the amount of times I've heard, well, I just go into chat GPT and I say, from a regulatory perspective, what do you think of Pfizer? And it is like great that you are experimenting, not to dunk on that too much, but that is not a consistent, benchmarkable, trustworthy, synthetic audience. So what do we mean by that? A synthetic audience is a virtual version of a real person, but it could also be other things. It could be the virtual version of an organization like a company. It could also be the virtual version of a cohort of individuals like conservative politicians or patient advocacy groups in oncology. So once you start thinking of it from those three lenses as an organization, as a aggregate cohort, or as an individual, you can start breaking apart and training synthetic audiences to serve your communications workflow. So what do I mean by that? That means that you could theoretically automatically capture content from sell side analysts who are covering your stock on a quarterly basis or key journalists from CNBC or the Financial Times or Stat News and Fierce Biotech. You could also start combing through huge reams of data that might represent, as I mentioned, conservative politicians or liberal politicians that might take together like 60 different politicians views and combine them together. Certainly that's true for things like patient advocacy groups, climate activists, or other key stakeholders that there's of myriad organizations within. So the way that you do it is you think first about the content. What is the content that this synthetic audience could be best represented by? When it comes to say politicians, that could be their floor speeches. their white papers, the legislation they've introduced, certainly their Twitter and LinkedIn and anything else that's publicly available out there. When it comes to organizations like patient advocacy groups or activist organizations, think about their own content, Their newsroom, their blog posts, their op-eds, their executives appearing on podcasts like this. And then of course, when it comes to even internal audiences like employee groups, Kristen Sweeney: Mm-hmm. Dan Gaynor: you might want to look at your ENPS surveys or your Gallup Strains Finder or your town hall commentary and Q &A that's submitted to executives and use those to train the audiences. In other words, it's no longer, I think, good enough to say from a regulatory perspective or acting as a journalist without feeding the right data into an LLM. And the real tricky thing is the ability to orchestrate feedback from all of these audiences consistently. Kristen Sweeney: Mmm. Dan Gaynor: whether they're an organization, a cohort, or an individual, and get consistent feedback from them. In other words, like creating a grading system that gives you an A through an F, or a zero to 100, or individual feedback on messaging. So first and foremost, think about the data sources to inform those audiences. And second, think deeper about how you would define those audiences so you know the right level of fidelity to engage with, all the way from an organization down to an individual. Kristen Sweeney: Got it. And then once you have built a strong sort of avatar, synthetic audience, how do you use it effectively? Are you testing, here's this press release, how will this be perceived? ⁓ Or are there other specific applications? There's like the sort of intuitive, yes, I understand how I would use that, but there may be other use cases I'm not aware of. Dan Gaynor: Fabulous question. I would say the spectrum ranges from like content all the way up to strategy. So on content I think you nailed it, right? I'm working on an earnings script. I'm crafting a press release I want to run it by this audience of synthetic analysts that cover my stock because it's really about ⁓ &A But I also want to understand at the edge of the target how say I don't know perspective employees might react to it say we're acquiring a company and moving into a different vertical, would that perhaps trip up folks that we want to hire and talent that we want to retain? So content's definitely a play there. I also think getting strategic recommendations is totally possible. So you could be asking synthetic audiences, what conferences should I appear at? What channel should I be focused on to really appeal to you? You could also be asking, what other audiences ⁓ Kristen Sweeney: Yeah. Dan Gaynor: are you interested in collaborating with? You can really actually chat with these folks in an agentic manner as if they're real individuals. So as an example, if you wanted to focus on key influencers in the pharma space, you might want to target particular doctors or executives. Those synthetic audiences could tell you sort of their network around them that they care about. So they care about certain media outlets, certain influencers. let alone like certain conferences or other peer execs or doctors, you name it. So yes, it can definitely edit press releases, but it can get all the way through giving you advice that should shape how you deploy your narrative to the world. Importantly, and I think this is where it gets really exciting, it's like, we're so prone to navel gazing in the world of comms, right? The executive says we should campaign on sustainability and innovation, you just say, So while there's the content play, and there's the calm strategy play, I think the zenith of using synthetic audiences is to ask yourself the toughest question of all, which is, is the narrative that I've been campaigning on and all the components of the narrative pillars, like my innovation story and my quality assurance story and my sustainability story, are those even the right things? Do you have the courage, in other words, to let the synthetic audience question your hard-fought judgment, your... basic plan that you've been running with for years and years and years. And if you can get to that point where you trust the content that goes into the synthetic audiences and therefore trust the output of the synthetic audiences, I think they can do a good job optimizing or calling into question your corporate narrative itself, which of course would have huge knock-on effects to the way that your entire company as an enterprise communicates. Kristen Sweeney: Yeah, I love that you talked about the courage to do that. ⁓ There is a truth teller prompt that I have run ⁓ for my business and it obviously it only has the data set that you are, you know, your chat history with within the LLM and there are other certainly other decisions and conversations I have outside of that many of them, but ⁓ it's brutal. It's it's ⁓ Dan Gaynor: Okay. Kristen Sweeney: You know, I think that that kind of feedback can be really challenging, especially, you know, I know myself, I'm like very much a problem solver and it's really daunting to see something like that and be like, ⁓ my God, I have to fix that yesterday. You know, that like, what now? What do I do with this information? And sometimes it can feel easier just not to have it. ⁓ Dan Gaynor: Right, I mean, we're so focused as communicators in massaging the message. Are we saying the exact right words? And we're making this somewhat naive assumption that those words are ever gonna be heard or read. The more important question is to say, am I campaigning on the right storyline? Am I buffeting that storyline with the right proof points? And that is as much an internal exercise as it is an external exercise. Kristen Sweeney: Yeah. Dan Gaynor: So referencing something I said earlier, for communicators, AI can provide this validation for a shifting of course that your gut instinct or opinion would have never facilitated. You know, just to give you an example, I was talking to a CCO the other day and she had a strategic plan that she was presenting to her CEO. And she turned her laptop around in the meeting and said, this is what my agent says about what our strategic plan should be. and the CEO said, aha, and took his computer and turned around his agent and said, my agent disagrees. It was this weird Wild West standoff. And that is AI done wrong, right? If you individually are training your agent with your content and your construct of how it should make decisions, and everyone else does the same thing, you're gonna have a bunch of Wild West standoffs where everybody's pointing their agent and their guns, so to speak, each other. Kristen Sweeney: Wow! Yes. Yes. Dan Gaynor: So the key thing to your point is, all right, let's have a real collaborative discussion around what data sets and what applications of AI we want to really focus on. And then it can become a convening force. It can become, as I said earlier, this horizontal mechanism, regardless of the channels you own, the audiences that you own, whether you're in media relations or executive comms or social media and content, it can become a unifying force. Kristen Sweeney: Yeah. Dan Gaynor: And that comes down to a communicator breaking down their workflow, getting feedback on what other folks' workflow are, and then creating AI as effectively like a joint project that unifies the culture instead of fragments the culture. Kristen Sweeney: Yeah, that is so cool. And this is fascinating. We could go on about this for much, much longer. And perhaps we will offline. But I do want to make sure we have a chance to, before we wrap up, hit up our lightning round segment, The Last Word. So I'm going to transition this out. Thank you for this conversation. It's been really amazing and a really different and nuanced take on the applications of AI that I would say I appreciate it. doesn't, it's not leaning into fear mongering ⁓ or some of the, you know, kind of overblown hype. And certainly there's a lot of that in the ether right now. So thank you for a really thoughtful conversation around it. Dan Gaynor: No problem, I'm happy to, anybody's listening, happy to chat. I'm Daniel Gaynor on LinkedIn, and I'm always happy to co-invent stuff with people. I was sitting on my couch on Sunday night, and I made a healthy breakfast app. So I would not say I'm always operating at the peak of corporate strategy. Sometimes it's just the best way to get chia seeds. Yeah, chia and flax in your oatmeal is a key thing, I'm warning. So I made it on quad design. Kristen Sweeney: okay that's okay. There you go, gotta balance that blood sugar. Yep. Alright, let's do it. Question number one. What's your favorite word to say out loud? Dan Gaynor: Okay, but hit me with the lightning round. stellar. Kristen Sweeney: Do you use it often? Or is it like held in reserve? Dan Gaynor: I find it is ⁓ my way of saying fantastic or sounds great. And I'm not sure that it's something that I consciously would pick as my favorite word, but I find myself saying stellar a lot. Kristen Sweeney: Alright, question two. What's a word you may know from reading but you are not quite sure how to pronounce it? Dan Gaynor: There's one word, especially in the corporate context, that strikes great fear in me when you sort of feel it coming down the manufacturing ⁓ pipeline and you're about to say it, is specificity. Specificity. I can say, let's get specific, but to say with specificity, I always trip up on that one. Kristen Sweeney: ⁓ Gotta slow down. Yeah, yeah, yeah. That's a good one. Good one. Question three. What's a phrase or saying you wish could be retired for good? Dan Gaynor: ⁓ I have a corporate answer and I have a personal answer. So the corporate answer is let's stay in touch. I don't think there's any version of let's stay in touch that means let's stay in touch, right? When you say let's stay in touch, it's effectively saying we will not talk soon or again. And I think you should just be like, great talking to you. I've learned a lot. It's been stellar. Goodbye. You say let's stay in No one on a first date has said let's stay in touch and means to go on a second date. Kristen Sweeney: Great, hit me. Ha Hahaha It's best stellar. Dan Gaynor: You know what I mean? And then. Kristen Sweeney: Yeah, I sometimes say we, we don't do anything. We never does anything, right? So us, let us, you know, it's a sign that no one's really in charge. Dan Gaynor: That's a one. That's a good one. So that's. Okay, so that's my corporate one. Let's stay in touch. My personal one, I have a three and a half year old and a one and a half year old. And I think there is, and you know, it's very difficult to get the kids anywhere, arrange any play dates, go out to dinner. I don't think there's anything that is ⁓ more selfish than another parent saying, we're happy to host. We're happy to host means that you have to truck your kids in the car, bring them to our playroom, you have to bring your snacks. You know, I'm happy to host. You come to me. Don't get me wrong, in other parts of life it's great, but when you have little kids and you're trying to arrange a collaborative exercise, a dinner, a play date, we're happy to host. That is selfish of you. Kristen Sweeney: That's hilarious. You know, I think there are people who feel differently, but I agree with you. We are big hosters and I'm like, I would rather clean up the mess than have to go anywhere. I think there are people who are like, I'll go anywhere as long as you don't come into my home. So maybe it takes all kinds. Dan Gaynor: That's what I'm saying. That, I'm happy to, it's much easier to vacuum than it is to get them in the car seats and get them dressed and deal with the meltdown. So I'm sorry if anybody out there feels attacked. It is my lived experience. It is difficult to transport. Okay, I'm happy to host. I'm happy to host too, but you know, okay. Kristen Sweeney: Agreed. Agreed. No, I'm happy to host. What's a communication practice or principle that's non-negotiable for you? Dan Gaynor: Michael Bloomberg has a good line, in God we trust, everyone else bring data. So what I mean by that is there is ⁓ no world where your gut instinct alone is sufficient to shaping strategy as a communicator. Everyone should at least have some data buffeting their strategic direction. And to my point earlier, ideally your data sings part of the overall song. harmoniously with other pieces of data. So step one is to yes, have data. Step two is to make sure your data is connected to or tells part of the story with other data sources. So I mentioned like different channel sources. And then step three is to turn that data into insight, right? I'm not just coming up with an idea and cherry picking data that buffets the idea I already came up with. I'm coming up with an idea as guided by the data. And that's really where AI can come in. AI can help. present, translate, streamline, pick the verb, data sources that allow you to narrow your option set. But ultimately, if you're really data-driven, that means you are trusting the data to give you option one, two, and three, and not try to just come up with four, five, and six, because you don't really trust it in the first place. So in other words, my key takeaway would be bring data. Kristen Sweeney: Yeah. Yeah, great. And last one, what's the last thing you read, heard, or watched that really stuck with you? Dan Gaynor: little out there, but my YouTube algorithm has served me up a bunch of stuff around how geography shapes history. So for example, even going back to prehistoric times, ⁓ like how fertile agricultural land is, can connect to certain demographics and where folks live or certain industries popping up. So for example, in like Alabama, there was this ancient sea. Kristen Sweeney: Yes. Dan Gaynor: that's led to a certain strip of land being this like incredible breadbasket. And that led to more agricultural businesses. And then that led to like certain political lines being drawn. And I'm super fascinated by that concept in the professional world as well, right? Like as somebody who went from the Obama administration in politics to selling shoes at Nike, to consulting big businesses, to being an entrepreneur with like two guys and a laptop and then getting acquired. Like I've seen how much I have changed. dependent upon the geography of my career. And the proudest thing I had, other than like having a great wife, when I got married, I had somebody from every job stop in my career there. Like every single job had been there. And for me, it was like this like visual panoply of learning that I've been able to get exposed to. So while it's true in going back to prehistoric time, how geography shapes history, I also think like, Kristen Sweeney: ⁓ yeah. Dan Gaynor: your professional geography really shapes the way that you're able to develop intellectually. And there's that line that you're just an average of your 10 best friends. Well, I also think as your career goes, you're kind of an average of every job you've had. So I think a lot about taking conscious steps to put myself in rooms where I'm by far the least smart person. Certainly that's true at Harvard Business School and building companies. And I would just encourage folks. Kristen Sweeney: Mm-hmm. Dan Gaynor: as they're thinking about building new tools, particularly with AI, to try to expose those tools to as diverse an array of jobs and intellects and backgrounds as possible. Kristen Sweeney: Yeah, I love that. ⁓ I'll be thinking about that for a while. Thank you again so much for, yeah, send them my way, absolutely. Really appreciate the conversation today. You already shared your LinkedIn. Anywhere else you want to direct our listeners if they want to connect or learn more about you. Dan Gaynor: I got some cool maps to show you and happy to chat. No, just look me up Daniel Gainor on LinkedIn and happy to chat anytime or ⁓ even if you have an inkling of an idea and you're saying, okay, how would I go about bringing this to life? I'm happy to think about content, metrics, audiences, and any potential application out there. I don't think I've met one person in the world of AI, let alone in comms AI who feels like they've totally figured it out. So if you feel behind the curve, you are simply in the same bucket as me and many other people. You're totally average. Kristen Sweeney: in good company, yeah. Dan Gaynor: and I would be happy to collaborate. In the meantime, thank you so much, Kristen. It's been really fun to be on here. Love the show, and I will be sure to like, subscribe, share. I can't wait to see where this goes. Kristen Sweeney: ⁓ you're doing my whole spiel for me. Thank you so much. Thank you very much. All right, talk soon. Dan Gaynor: I know how it is to host.