Behind the Buzzwords: AI Through the Eyes of Fractional Leaders

Welcome to this very special TechCXO AI conversation. My name is Virginie Glaenzer. I'm a fractional CMO, and I'll be your moderator for today. I'm really excited. For those who do not know TechCXO yet, I'll make it very brief. We are a professional services firm, a collective of 140+ interim, part-time, fractional executives. For the last 20 years, we've served businesses in the $2 to $50 million range. Many of our 5,000+ clients have raised capital during our work. So as a result, we've created more than $6 billion in value through incremental sales, capital raises, and helping M&A and exit for our clients. With that, I'm excited to introduce our group of speakers. In this conversation, I am joined by Brian Denston, who lives in Charlotte, North Carolina. Brian is a fractional CTO and CIO. He's an expert in mobile application development, healthcare integrations, and all aspects of telemedicine and has worked on over 40 engagements in a fractional capacity for the firm. Welcome, Brian.

Thank you, Virginie. It's nice to be here. I look forward to talking with you and the rest of the crew.

We also have Kevin Corson, who lives in Siesta Key, West Florida. Lucky you. He is a— Kevin is a CTO and CISO, Chief Information Security Officer, Kevin specializes in big data, data lakes, data warehouses, advanced analytics, predictive analytics, machine learning, AI model design and development, and finally, complex algorithms tailored to company-specific data. Very excited to hear your experience, Kevin.

Thanks, Virginie Glaenzer. Nice to be here.

We're also joined by Todd Meryl, who lives in Atlanta, and he's a fractional and interim CTO and CISO. Todd has served in a number of companies in the CTO, as the CTO, including startups, venture-backed, and private equity companies in the financial services, real estate, technology, and software. We're super glad to have you with us, Todd.

Good to see you, Virginie.

And finally, as the moderator, I will also share my experience during this conversation. I'm currently in Washington, D.C. I will be moving back to New York in September. I'm the newest partner of this group as I joined Tech CXO last November, but I've been a fractional CMO for the last 6 years with a focus on B2B SaaS, helping tech companies grow as well as services firms evolve into SaaS. So again, welcome. Excited to hear everyone's take on our 4 AI questions. So before we dive in, just a quick bit of housekeeping. This is a live event, so please ask your questions in the chat. I will be monitoring, and we can bring some of those questions at any given time. And to kick things off, we're going to take an unusual approach. We're going to invite you to share jokes about AI. Yes, that's right. That's how we're going to get started. And it's really because AI is such a heavy topic, you know, it's, I don't know about you, but feeling fear and frustration and excitement, you know, all those conflicting desires and feelings, we've decided to start with some jokes. So I'll start. And actually, I asked my ChatGPT to write a joke about itself. It crashed. Turns out self-awareness wasn't included. That's my first one. But here's a real Here's the real joke. So a machine learning algorithm walks into a bar. The bartender asks, what would you like to drink? The algorithm replies, what's everyone else having? I thought that was right on point. Anyway, anyone else has an AI joke?

I've got one just to follow. Self-deprecating, I chose Claude to haze itself., this, I guess it's a limerick. There once was a model named Claude who tried to write code that was flawed. It said, I'm quite certain this function's working, and then crashed with a segfault. How odd.

I like it. Like it too. Um, one last. What's the opposite of artificial intelligence? Natural stupidity. I like that. We'll start with there. Okay. I'm done. Any other joke? No, we're good.

Asked AI to, you know, take a peek at my couple jokes I was gonna give you guys, but it told them, told me they were low quality. So I said, get in line behind my kids.

Okay.

I got one more and then I'll, I'll stop.

Okay.

Go ahead. Uh, and then, so, You know, we had a limerick, so we got to do a haiku now, right? Because Reddit plot twist, I can do both limericks and haikus. What's next, a sonnet? So Claude has sonnet and opus, right? So yeah, that's technical. Unexpectedly good.

Very cool. Well, I'm already feeling better. It's like, it feels like having a couple of drinks before a tough conversation. So Now we're ready to jump in, get to the heart of it. We have 4 questions to cover, and for each one, we'll just go around the group. It's going to be a very easy conversation. So question number 1, the hype versus reality. AI is everywhere in the headlines, but from your vantage point, what's the biggest misconception you see about AI in business today? We'll start maybe with Brian.

I would say the biggest misconception with AI today is that it's this magical, like, plug-and-play silver bullet. Like, companies are going to go out and buy a tool and that's going to solve all their problems. But I think of AI just like electricity, running water, those kinds of things. Like, you have to have the right culture, you have to have the right workflows, you have to have the right data ready. AI is this amplifier for our businesses. So if your processes are broken, AI is just going to make it worse. So I think that's the misconception. There are some cool quick tools you could plug in to help solve a small problem here and there, but I think we really have to rethink a lot about everything. So that's my quick answer for that.

Very nice. What about you, Kevin? What's the biggest misconception you see about AI in business today?

Yeah, I think the biggest misconception about AI is the level of its, uh, current maturity. It's certainly more advanced than it used to be, but I think people have a tendency sometimes to mistake speed for accuracy and maturity. And, um, that leads to a lot of companies thinking that AI is capable of doing things that it can't do. Uh, for example, um, you know, I've had a number of CEOs saying, hey, we should be able to, you know, somewhere between 3x and 10x our, our coding capacity and efficiency using AI. And it's interesting to note that Microsoft a few weeks ago came out and said they're doing about 30% of their coding now with AI. And this is not only a company who, this is what they do, they do software development, but they're also large investors in OpenAI. So they have this great technology at their disposal and they're at 30%. Just for reference, 10x improvement will be 90%. So I think the expectations of folks are a little bit outta line of what it's current real capabilities are.

Interesting. Yeah, I would agree with you. What about you, Todd?

Yeah, I think maybe two things, you know, marketing sprinkling AI like pixie dust on everything is maybe, you know, a little bit overdone. So we're all a little bit sick of it. You know, the hype cycle, we're definitely kind of going down the trough of disillusionment. But a lot of us are coming back up the other side. And we'll talk about that part later. But I think one thing that I'm not super thrilled with is kind of in line with everybody else. I think we're ascribing too much or too much personality or we're, you know, you see people who want to name their AI and put it on the org chart. And I don't, I'm not a big fan of that. I don't think that's healthy. I think we're all humans and at the end of the day, we all have to be humans working with and for humans. So I think AI as a personal accelerant is good, but I'm not a big fan of just this kind of zombie army of agents out there. They're going to take your job. I don't think that's what's going to happen.

It's like this. I don't know, what's the word? Is it anthropomorphism or something like that? Right. Like we can't be giving these things some feelings on that front when it's really just a math equation behind the scenes. And a lot of us are doing that. So yeah, I agree with you, Todd.

And I, you know, you see, you see people, you know, talking to their ChatGPT and taking therapy or, you know, like help me figure out my boyfriend or, you know, whatever it is. And then, you know, that's not what it's for. I don't, I don't think that's healthy and I don't think that's a great way to think about what this technology is. It's not that, right? It's an accelerant. It's a tool for humans to use to do more human things better.

You know, I think it's interesting to point out just real quick that calling it AI in some sense is just kind of a misnomer. It is AI in the sense of it's a probabilistic model, but It was trained on actual human intelligence. Everything it knows is everything that, you know, that humans have produced. And it's just really good at regenerating patterns that it's learned from that information. Thank you.

For those who are listening, let us know, you know, the answer to this, your answer to the question, what's the biggest misconception you see in AI in business? I'm gonna close this question by sharing my take on this hype versus reality. And I agree with you, Todd and Kevin, about, the overpromising, the hype cycle. I think for me, one of the biggest misconceptions that I see in marketing is that it will replace all marketers. And the reason why I believe it won't change as fast as we can read it will is the human factor. Yes, AI can write books, make art, compose music even. Meanwhile, you're still trying to unmute yourself on Zoom. So I think A lot of businesses right now that I'm speaking to have broken data infrastructure operations. And if you add to this the uncertain economic environment that's threatening profitability, many of them are still trying to catch up to digital transformation. So that would be one of the misconceptions that I see. I see another one also in a lot of articles that, as you said, Todd, are writing about losing our jobs. And I don't agree with that. I think AI literacy is critical in the same way that reading and mathematical are. And for me, teaching AI tools and concepts to K-12 education, higher ed, corporate training is absolutely crucial. But I believe that AI is good for workers who want to use it. Last month I was reading a PwC report where it says that the AI-exposed sectors saw a 38% growth in job, and AI-skilled workers earn a 56% wage premium. And I will end with, you know, when, again, when I look at it from the marketing lens, the last misconception that I see that is the fact that most articles talk about AI as a productivity tool. But I think if we treat AI as a growth strategy, so not just an efficiency strategy, really a way to evolve a business, that's when it really becomes interesting. And if we provide to each employee the knowledge and an agentic AI, it will not only profoundly change their creativity, their engagement, Yes, it will also become an exponential workforce multiplier, but I think more importantly, it will empower people to change the way they work.

It elevates your thinking, right? You're not grinding through the paperwork. You're thinking about strategy and making phone calls and talking to people, which is what you should be doing, right? And it's what you're really getting paid for.

Exactly. John Bongard, thank you for your, feedback on the biggest myth, which is with enough work, we can eliminate hallucinations from LLM in deployed application. Very interesting. So with that, let's go to question number 2, which is a way to bring AI back home closer to what we do. AI impact on our world. So how is AI reshaping your world? And what do you see changing in the next 12 months? Brian, what's your take on that?

I'd say, you know, I've been a fractional CTO for 8+ years, and I come in and help lead the strategy. I lead the architecture. I do all the stuff I've been doing throughout my career, but now it's added a bunch more hats on top of it. Like, I have even more responsibilities to help lead an organization in some fashion. So, I am now the AI enablement coach. I'm the AI governance expert that has to help with that kind of thing. And I'm helping teams rewire how they think and how they address problems, training marketing teams how to use AI for prompting, like I'm sure you're doing, Virginie Glaenzer. How do we use Copilots and do vibe coding versus flow coding? Like, which one's better in certain situations and how do we unpack that? So, I think that, you know, my calendar is just going to, you know, this, this realm of how much I divide my time across different buckets. I think AI is slowly taking up more and more and more of my calendar time as I've gone down the rabbit hole faster and quicker than some of my colleagues or something. I am the guy that they're turning towards to ask those questions. How do we do this? What have you seen? What am I seeing across a couple different clients and how do we roll out those tools? So more and more of my time is definitely gearing towards AI in a big, big way. So, it's not just infrastructure talks anymore.

And what do you see changing in the next 12 months from your lens?

I think agentic is going to come into its purview in a big way of like, how do we set this task on autopilot and get the AI to assist us with certain things? I think we're going to establish these stronger governance plays, just like we had to go through that learning curve with cybersecurity, or how do we use Microsoft Word, or what is the internet going to do to my business. This is a new paradigm shift, and we're going to be helping clients navigate through that and establish some repeatable things that we'll get to do over and over.

Fantastic. Thank you so much. What about you, Kevin? How is AI reshaping your role, and what do you see coming in the next 12 months?

Yeah, you know, I agree with what Brian said, but in addition, I think AI is a wonderful research tool. So when you're, you know, I'm working with a client and need to get specifics on some things that maybe I don't have direct experience with, I can use it to, you know, do some research there, learn a lot. It's a, it's a really good educational tool. The danger, of course, is you have to validate what it tells you because models still do have a tendency to hallucinate sometimes. And, you know, they're just— it's all probabilities. So they're just stringing words together that are high probability answers to prompts. But, you know, I do think as a research tool, it's been extremely helpful. And I think in the next 12 months, what I see changing is definitely the rise of agentic, but there's a whole different side of that. There's, there's security that has to be put around Agentic, and it has to fit the organization's compliance framework. So I think it's going to really have an impact on compliance and security from a governance perspective.

Interesting. What about you, Todd?

Yeah. So personally, you know, so I have kind of two sides to what I do, you know, software, you know, executive, and then, you know, uh, consultant, you know. So on the consulting side, um, you know, I love to talk about, um, AI taking out the grindy bits of your job, right? And then so one of the things that I have to do is I have to, you know, follow up after sales calls and, and then write a lot and make proposals. It's the same thing over and over, kind of, kind of, sort of, but You know, AI has transformed that for me personally. And then I think, you know, exactly what Kevin was saying. You can't just take, you know, one shot at this and then, and then call it good. But man, you can, you could definitely have your assistant tee up the first 80% and get most of the details pretty, pretty spot on, maybe even better than that. So for me, accelerating, you know, time to proposal,, in staying organized and then going through, you know, emails and, you know, helping me think through, uh, the structure of what a marketing campaign might look like. Things like that, um, have dramatically improved for me personally. And then on the coding side, um, I got to a point where, you know, as an executive, you don't get to really go hands-on keyboard as much as you used to or you would like, and then there's years where you don't code at all, which, you know, I don't like to go too long, you know, to get away from it. But some of these coding tools this year, I think I'm going to point to Cloud Code, for me really have been transformational. And I think you can squint kind of hard and see the future and you can, a talented individual can create a lot of very meaningful code in a heartbeat, but someone who doesn't know what they're doing can sure make a mess in a heartbeat as well. So it is a surgeon's scalpel, and that's good for healing and hurting. And I think that leads to the second part of your question is what I hope or think is going to happen in the next 12 months. 12 months is forever in AI, Bill. I hope Claude gets some of this infrastructure built that they say they're going to build and kind of shores up their service offering. I think everybody's building data centers as fast as humanly possible. I think we saw something come out of the federal government this morning with the AI policy, which I skim in the headlines, looked very, very promising. Like we're going to really go after this as a nation to win it. So that's very encouraging. I think these models are going to get ever, ever more accurate and impactful and quick and reliable. And, you know, sure is Wild West now. And, you know, we all see the cracks, you know, in the dam. But, you know, everybody's going like heck to fix them all. So it's going to be a really fun next 12 months for sure to see where we are.

Absolutely. Thank you for that. There's a question from John about if you found a market for AI governance strategy as a primary offer versus a French fries add-on to other offers.

Brian, you—

go ahead, Kev. Yeah, if you've got— we can all probably weigh in on that. I haven't. I don't, I know that there's probably plenty out there that are offering it. I'm sure like Vanta and others are having some strong approach to it. I have scoured the internet. I've used Claude and GPT to craft my versions of what I tentatively hand out to clients as a starting point. Each client is definitely different, depends on your industry and your risk tolerance level. But I haven't found that strong market framework yet that's beginning to be a leader. I do say go to the NIST framework, the cybersecurity framework that the government talks about. That's a good starting point as a place, and I think they're going to hopefully lay out that. And I haven't seen what Todd just mentioned with our AI stance as a nation. So, potentially there's some gems hidden in there on how we want to approach it. So, you know, Great question. Kevin, do you have some thoughts?

Yeah, I think there's two sides of this. You know, I think there's the using AI side of compliance and the building AI side of compliance. So on the, on the usage side, I'm not sure that there's another framework that's, that's truly needed. And the reason is because a lot of it's already covered under things like, I'll just, I'll just say, you know, CIS, which is built on the NIST standard that you mentioned, Brian, but like take CIS and the requirement for data privacy, for example. Right? AI is just another tool that, that is an avenue to potential data exposure. So, you know, I don't know that there's a separate framework required for it. It's just another technology that you have to consider when, you know, being compliant with data privacy. However, on the other side of the equation, this is what came out of a lot of it. What came out this morning that Todd mentioned is there's a lot of danger, I think, in risk around data curation for training, right? So I think recently Elon Musk, you know, for the Grok AI came out and said they're just gonna more carefully curate the data. But that's a huge responsibility, right? Because if you are using, say, AI for research like I do, and you're using a model that's curated out things that might be, say, competitive, or that they just don't want included for one reason or the other, then, you know, it can become really limiting in what it does. And it can actually, if you want to take it to an extreme, it could change the opinions of people about, you know, everything from current technology to historical events. Right. So I think there's something needed on that side, but I don't think there's anything necessary yet on the, on the, on the user side.

I think from the cyber perspective, there's definitely new interesting attack surface that is emerging and we're figuring all that stuff out. I think it is a derivative, an extension of existing info, information policies, you know, internal versus external and that kind of, that kind of classification that we already have good controls for. Yeah. And then, you know, boy, the intellectual property thing is, is, is difficult, right? Kevin, you're, you're a musician, you know, you're, you're kind of sensitive to this, right? So, You know, you don't want your, your music getting ripped off, but you kind of do want to get discovered. So you kind of have to get in the search engines. So, you know, potato, potato, you know, what are we going to do? So it's a weird, weird new world. Um, but I think, I think it all comes back down to, um, you know, critical thinking, you know. And so, you know, everybody talks about in education, everybody's kind of gotten rote, um, you know, question, answer, question, answer. Just memorize the answers. I think we're gonna get back to, everybody's gotta, we're just gotta be wise and, and think critically about what's going on here and figure it all out and make, make good decisions. So maybe that's a net, net positive.

It's, and it's interesting cuz like, yeah, I don't know, we're going down a small rabbit hole, Virginie, so reel us back in if you'd like us to. But like to, to me, like the thing I wanna point out is like, there's bias in every model I've seen. Like, there's strong bias in every model. Like, I am plant-based. So when I talk to the model about anything food-related, it is, it is pushing those standard things that are in the median of society, which is a meat-driven culture and not a plant-driven culture, right? So, you know, so like, that's just one angle that I see the bias. Kevin sees it definitely in intellectual property and you know, so the datasets that we are feeding in it are the common knowledge of humanity, and it is that bell curve of ideas. And imagine if AI came up in the 1950s, it'd probably be promoting smoking, right? So thankfully we've changed that kind of a thing. So like, I do think this is where I'm not worried about it creating more jobs, 'cause there's gonna be huge amounts of research and human ingenuity that has to drive us forward to create new datasets I mean, just look at what AI is helping with in astronomy and helping count galaxies and some of that kind of stuff. Like, we're going to open the door in so many ways to new jobs of exploring new datasets to feed into AI. And I hope we solve Virginie's number one problem, which is, hey, you're on mute. So like, that's the thing. Like, if AI can help expect that I'm talking to you and not my kids in the other room, and it automatically takes on and off mute, that would be amazing.

So Yeah, absolutely. Well, thank you for that. And I'm going to share my experience from a marketing standpoint and how AI is reshaping my role and what I see in the next 12 months. First of all, as a CMO, I'm experiencing— someone is pinging me. I hope you are not hearing it. But as a CMO, I'm experiencing AI as a form of collective intelligence. That's really the experience that I'm going through. But like with human emotion, it's rarely rational. So even when I trust the data, I always take AI outputs with a grain of salt. Having said that, as a fractional executive and independent consultant, AI has become my daily helper for research, brainstorming, ideation, content creation, business analysis. I often feel overwhelmed because I hope when I read the news, I feel that I'm looking at the tip of the iceberg and I'm so late to the party and I have to constantly keep up. So it's quite exhausting, but I've also been able to witness my own expansion to self-reliance. So I can do more faster by myself. You know, in the past, I used to depend on at least 3 people, a designer, copy editor, and a researcher, but now I'm able to do it all. So I think that's been really amazing. As far as the impact as a CMO, AI role is increasingly about making sense of the chaos, right? Before, it was about picking tactic to hit a goal and using a few playbooks and aligning the teams. But those playbooks have become obsolete, and not just SEO versus GEO, but overall. And so now I'm actually using AI to build better strategy. It's almost like a new playbook in itself. The second big change that I, the second part of the question, which is the biggest change that I think we'll see in marketing is, like you said, Todd, through AI automation. But honestly, so far, my experience has been really frustrating, a lot of frustration. I've tried n8n.io for AI workflow automation, but without coding skill, it was just too complicated. Then I tested relevance of AI to build my own AI agent for tasks like meeting prep, but it just brought me some very generic output. So doing, preparing for meeting, doing my own work is way better because I'm the one to actually be able to connect the dots. And then I'm currently using FeeneyAI.com, which writes prospecting emails for financial advisors. I'm using this platform for a client, but the AI writing, email writing part, same story. The quality isn't there. I recently used Fixer for email response, so it prewrites email auto, it automatically writes response to any email that you get. And it was okay, but it's just not worth $50 a month. Sorry. Which is, and now I'm experiencing with MCP, as I mentioned to you, Todd, recently to build my own CRM. I'm very excited, but I'm not doing it on my own. You know, MCP is this protocol where you're basically letting Claude connect with your tools and data. So we'll see how it goes. But my hope is that in the, within the next year, it will finally get easier to create AI agent that actually works the way we need them to. That's my hope.

Let me bounce off that. So number one, I think, you know, the context engineering, if you can pour on more context about, you know, collateral in the, you know, while you're making those emails, like that's going to make them a lot better. So I think awareness of like context is going to be one of these kind of things to watch in the next 12 months. But, but I think the important point you said was it changes the way I do business. The playbooks, the old playbooks are not the new playbooks. And I think that's a critical thing. You know, we all talk about like AI-first companies, you know, what is that, right? I think that's an important concept. And maybe we can kick it around here, but, you know, it's going to change the way we do things in unexpected ways. And we all have to figure it out. But there's going to be a new normal, and it's going to go very quickly. And I've seen a lot of people go, and no, no, no. Holy crap, I gotta get this and then go bananas with it. And then you, there's kind of like a pullback and, okay, what is this really good for?

I don't know if anybody else wants to bounce off that, but yeah, maybe John's question about, you know, is there a market for AI government strategy? Maybe we haven't seen it yet because of those reasons, but there's gonna be a point where it's gonna be a no-brainer. I need someone to actually create a government. I think now everyone in their department are doing their thing. On their own. I did a webinar not too long ago for marketing individuals. How do you protect— how do you use AI without the risk of AI infringement when you create images and the lack of copyrights? We asked the audience how many people have an internal AI policy. Less than 10% said that they had one.

But that goes back to the— I think it goes back to the two to two things is like we each are doing our own thing with AI and we each are paying the $20 a month for XYZ services and tools. We're testing out different ones. But like, like Todd said, it's that context thing. When my note-taker that sits in the meeting can talk with my email system, which is the hopeful future of MCP, then the context will flow from the meeting to the email and the draft will be that much better. So that— I don't know if that's one month away or if certain ones are already tackling that to a degree. I'm sure there's somebody out there that has solved it. It's an N8N automation, right? Yeah, it's here. So that is there. But like, that's what I think we're trying to do as CTOs and CIOs going into an organization is trying to help lay down some of that foundational piece is do we standardize on Claude or OpenAI? Do we pick which one has the right copyright protections because we're more marketing-oriented versus research-oriented? And then that question around like, now that we've picked one and standardized it, how do we build the vector database or the RAG database of this knowledge that when I have an amazing meeting and we bless that as a certified official thing that we're going to do as a company, it gets into the knowledge base that the rest of our employees can lean on and use as part of those conversations. So it's how do we flag a meeting that says this is executive eyes only versus anybody in the company can have access to the knowledge we discussed here. And it's exposed into the LLM that we're using as that fuller context thing that helps us fuel our work. So some of those things are all still rippling through and getting solved, and different companies are approaching it in different ways. But that's part of the foundational thing to set up. And I agree with Kevin, like the governance pieces are definitely like solved with the cybersecurity frameworks to a degree. But some of the other pieces of it is like we will only use this tool for our knowledge because we vetted this one. We spent the time to understand that and we are going to use this this repository to store our knowledge and don't paste it into your personal one, some of those kinds of things. So, all of it builds on each other. That's why I think it's like a new electricity or water that we've got to figure out how to integrate into our blueprint.

I think to give practical advice to John, this will be a thing that people will need and it'll skew larger enterprise. So, the bigger the enterprise, the more likely they're going to have a risk officer or internal counsel that are going to be very concerned with this kind of thing. A 5-person startup is not going to care very much. I mean, they should, but they're not because they have bigger fish to fry.

I just want to add on something Brian said. I think he brought up a great point about who has access to what meetings or what information in a model. I'm getting more and more interest from some of my clients about deploying LLMs internally for privacy reasons. But the bigger reason is that, I don't know if everybody here's familiar with Conway's Law, but just a rough paraphrasing of Conway's Law is if you have a siloed communication style in your organization, your systems and processes and your technology is going to follow that sort of structure, right? So, you're going to end up with siloed technology. And what I'm seeing with current AI use is it's very siloed. Everybody's using their own, you know, AI and doing things. And, and it would be great if we got to a point with AI where, you know, if somebody was working on something and it had overlap with what somebody else was working on, that they could bring those two people together to work on things or say, hey, this other person's doing some similar research, maybe you should get together or something like that. So I think there's a lot of opportunity there to build a more collaborative AI system.

I think in the thing that came out of the federal government this morning, there was a plank in there specifically for that standards and interoperability. So calling out that specifically that we all need to kind of interoperate or figure out ways, you know, can't be siloed. It can't be one model. It can't be one, you know, standard. We all have to figure out how to get along.

That used to happen at the water cooler. We were in the same office. We'd bump into each other. Let's go to lunch. What are you working on today? What am I working on? That doesn't happen anymore. We're in back-to-backs and stuff. And so I love that idea. Kevin, like if we can get the AIs to help nudge it, hey, Todd's working on the same thing, why don't you two collab? That would be amazing.

Fascinating. Well, let's move to the third question, which is a big one. And I think it connects with some of the ideas that some of you have shared. It's around ethics and risk, right? AI comes with enormous opportunities and you just, Brian, mentioned one that would be amazing, but it also comes with ethical considerations. So how do you see organizations managing risk? And I know we already talked a little bit about it, but how do you see organizations managing risk, governance, and trust in AI solutions? Kevin, you want to get us started on that?

Virginie, I got a little bit of dropout there. I missed like 3 seconds of your question.

Could you repeat? Oh, sure. Yes. So we're talking about ethics and risk. AI comes with enormous opportunities, but also ethical considerations. So how do you see organization managing risk, governance, and trust in AI solution?

Sure. So risk, you know, I think I kind of brought this up a little bit earlier, but you have to validate it, right? I mean, AI currently does still hallucinate. I'm not sure that's going to be resolved, you know, in the very near future. But I think you have to validate it. You can't sort of abdicate your own analysis and decision-making capacity to a model that's as probabilistic as these are, right? So I think that's sort of the first thing. And AI models in and of themselves don't have ethics, right? It's how we interpret the inference that is where the ethics lie. So I think ethics is still a very human thing. We just can't abdicate responsibility and just say, oh yeah, I'm just gonna, whatever the model says is what I'm gonna do. Because what ends up happening is you become basically a shill for the model, right? And a good example of this is people that use AI to say summarize large articles and boil it down to like 3 or 4 bullet points. Great, you can bring those bullet points up, but the second somebody has asked for detail behind that, you don't have it, right? Because you used AI to distill things down. So the ethical responsibility lies on us in the way that we use AI and the way that we interpret the results. There's still a huge responsibility on us to understand the detail.

Very true. What do you think, Brian?

I mean, yeah, we've discussed this a little bit on this, on this talk so far today, but I guess I echo what Kevin's saying. I guess I, when I go in and speak to the clients, I try to say like, What is your usage policy? Are we letting anybody use any LLM that they want, or these are the ones we've vetted and have blessed? Let's set up single sign-on for those. Then we've touched on this, but like before, we all know, like I work a lot in healthcare, so we all know the patient records are very sensitive. You can't just be giving those out, but we have to establish these data access tiers, and it's now for everything. Every meeting's recorded, every phone call's recorded, every email's recorded, and who has access at which tier level based based on your role in the organization. So the data tiers are moving from the patient records or the credit cards and things like that that we know should be secure to every aspect of your business. And then, like Kevin's mentioned this a whole bunch, and I agree, we have to audit the model. Is the model introducing a bias? Is the model giving us the right answers? How are we creating audit logs and double-checking the outputs that we got? Is our company's knowledge base sort of slanting one way or another? For it. So I don't, I guess I go back to, I think finance has been doing this for years and years and years. So how do we enable AI to follow in finance's footsteps? Like we want to give it enough freedom to not be holding things up because I need a credit card or pay something, but we need to have the right controls and accountability baked into it on all levels. So it needs to have its checks and balances.

It feels to me that it's really a topic that is much more prevalent in Europe, but way less in the US. Maybe because in the US it's about, you know, we're driven by revenue, productivity, and less bigger questions. Very interesting. Todd, what do you think? Do you see organizations managing their risk, their governance, or their trust in AI solutions?

Yeah, just to echo, you know, Kevin's point, I think human in the loop is a super important concept here, right? And I don't think we're going to get away from that. I don't think one shot is a good thing right now or for a while. I think humans need to audit everything that comes out of AI. And I think trust, you know, particularly, you know, like Brian was talking about in healthcare is a great example, you know, clinical decision support, you know, AI, or, you know, you've got a, you know, a model that's going to look for a tumor in a radiograph, you have to trust that, right? And then so we have to tune the output of some of these things. We were talking about traditional ML and computer vision type problems, but whatever we do as an automated piece of software or an automated system has to be trustworthy. And then so we have to ensure guardrails. And then I think people are still figuring this out, but you'll see agents auditing other agents and attempts to automate it. And I think there's a lot of really clever things that people are trying, but I think at some point we all have to watch what we're doing and be particularly vigilant, and we have to set up surveillance systems, not like cameras, but, you know, we have to watch what these models are outputting. And every time we roll new code or retrain it, we're going to have to go through and be very careful about, you know, what's coming out. And then, you know, you said something interesting, Brian, you know, maybe we can kick this around, but, you know, credit cards, You know, we've, you know, with advertising, everybody's watching everything we do. You know, your smart TV is watching everything, you know, and your little Amazon puck inside your house is listening to everything, your phone, you know, there's no end to it, you know. And so just because there's data available doesn't mean you should use it for every purpose, right? So it's awful tempting to look at those credit card streams and to go look at, I don't know, who's buying llama yarn or, you know, something and go harass them about why they're being mean to llamas. You know, so is that ethical? No. You know, is somebody going to do it?

Probably.

So we have— that's a part of the trust in ethics is, you know, we're going to have an explosion of capabilities and explosion in data because at the bottom of AI or machine learning is always data. You know, we got to be careful with that and guard it in maybe ways we haven't had to do in the past.

And I would, I would add to that, that to me, when I look at risk and confidentiality in marketing, when it comes to AI-generated content, there's no copyright protection. There's no confidentiality when using AI. But I do believe that this concern will diminish over time because the technology and regulation will mature. So I'm not that concerned. My concern really comes more from the human impact and environmental cost. And I think it's a bigger ethical question, right? The replacement of human work by AI and the significant energy consumption of AI systems and their contribution to climate change. I've discussed many times with younger individuals about their resistance to using AI because of the impact on climate change. So I think these are long-term challenges that we cannot ignore, and these should be really treated at the government level. There's one question that I often reflect on in that sphere, which is how much of life's meaning Today comes from work. So when work becomes secondary, where will we find our sense of purpose? And I still have like 30, at least 30 good years to work. So I think I'm going to see in my lifetime, and I don't plan to retire. So I think I'm going to see a world where the way we work has changed. And, you know, I don't want to say we won't have any jobs,, you know, we're talking 10, 20, 20 years, where will we find our sense of purpose? That's something that is bugging me and I don't have the answers.

So I had an interesting discussion with one of my clients, uh, the, you know, like a, like number 2 in the company. And she said, um, you know, I said, once we get these AI improvements in and everybody gets a little bit more overhead, you know, how do you think about that? You know, do we, there's kind of a slippery slope. Do you go faster because you can? And now we can go 30% faster. And then, you know, I fear we're going to drive everybody that much harder. That's a dystopian view of this technology or kind of what you're saying. Do we make our employees that, that much happier? Right. And then you remove some of the grind from their job and they keep going at the same pace. So I think that's a, that's an ethical one to watch because there will be companies that replace ever cheaper offshore labor with now bots, and then the poor people who have all the control of the bots now have to code 10x when they're already coding 5x. You're going to end up with a lot of very depressed, overworked, frazzled people if you're not careful. So I think that's an ethical consideration.

So let's go to the last question, which is really about the technology in itself, the future of AI in business. Looking ahead, what is one AI trend or technology that excites you the most and why? Kevin?

Um, yes, I'm mixing the order. For AI in education, I think that, uh, you know, in Chile they're doing some really interesting things that they have, you know, put models in place to really be the primary educator for, for certain topics, not across the board. But what they found by doing this is that it freed up teachers to have more one-on-one time with students that were focused on very specific things. So, you know, I think education is an area where AI is just going to really, really improve things. I mean, I use it all the time for education. So, you know, I have a couple of hobbies and, you know, when I want to dive into things and say, like electronics very specifically, things I don't know about, using AI as a way for me to really dig in and learn things that would take me a long time, or, you know, sitting in a class, you know, for, say, a quarter and going through that formal education process where this gives me the ability to dive in where my natural curiosity is and learn at my own pace in my own direction. And that makes it stick. So I think there's huge potential there. There's also risk, like coming back to the comment I made earlier about curating the data in these models. There's a huge amount of responsibility, ethical responsibility to make sure that the models are not biased in areas that could affect someone's worldview or something like that. So I think it's a two-edged sword like many new technologies.

Very cool. Education, that would be amazing. What do you think, Brian? What's one trend or technology that excites you the most and why?

I mean, I have to say I love the education one because the AI can personalize the education and accelerate the right people through it so much faster. But I mean, to me, I'm going to go with the one that's the buzzword right now is just the agentic AI. I mean, we're moving into this world of teamwork. Like, I'm fully addicted to deep research on all the platforms. I try— I actually write the same question and put it on Claude, Gemini, and GPT and see wildly different outcomes sometimes for the same exact query running a deep research. I mean, these deep research models are going off and they are trying to critically think about the problem and come back with the right book report for me. And it's just so amazing. So we're moving from these tools, you know, like help me with this problem to a deep research like teammate. So this agentic concept that's coming out is, you know, optimize this funnel or help me with this A/B test or, automate this piece of it for me is, is here. And how do we embrace it? How it's going to allow so much more experimentation at such low cost that we're going to be able to really test ideas and product market fit and go to market in such a unique way. So that's what I'm, I'm super, super excited about. And I, you know, I'm not worried about jobs like we are still discovering new animals and bacteria every single year. I mean, it's 2025. We're still discovering new things on this planet. We're gonna go explore Mars. I mean, like, it just, like, it's just gonna open up this new paradigm shift. So it's coming very, very fast as well.

It is.

What about you, Todd? Yeah, I, I think I got my eye on one near-term one and one kind of somewhat near-term, you know, near-term. I've been beating the drum loudly about MCP being an unlock. I look forward, and that's a protocol that allows fundamental frontier models like Claude or ChatGPT to use tools and to decide how to use them and what they're good for and then when to use them and to make a plan. So I think very, very near term, everybody's going to figure out this wonderful symphony of mashups with all these, you know, familiar tools in SaaS and maybe some new ones. And I think, I think some people are doing it better than others, you know, and then Um, that whole, that's a, that's going to be a really interesting, you know, chain of events to, to watch. Um, you know, like right now I'm really, really disaffected with how Canva handles MCP. You know, it should be better than it is, you know, but, you know, I want my slides. Um, and then I think kind of thinking out a little bit farther, we're going to have a bazillion jillion agents of all kinds of different flavors., there's going to be something that happens that's, there's going to be emerging use cases once we get to kind of the space layer of, of, um, literacy where like A2A from Google or something like that, the agents are going to start talking to each other about stuff. I think that's the next big one. And then, you know, so we're, we're kind of thinking about, well, how, how can, you know, the, the greatest memes on the planet, you know, AI and crypto. And, um, you know, automation, self-driving electric cars, you know, where's the intersection of that? Well, but it's kind of an interesting thing, you know, so two Teslas going down the road, um, you know, in self-driving mode, and someone's in a hurry because they're late, and can those two Teslas now negotiate, hey, I'll give you a couple bucks or Satoshis or whatever to let me pass, and yeah, you know, we're not in a hurry, you know, and then and these two agents in the background start negotiating for stuff. I mean, it's just kind of a silly example, but I think you're going to start seeing things like that where agent swarms, you know, are going to explode. And then now they, you know, hey, you know, there's got to be human in the loop here somewhere. Can y'all just figure it out and then come to me when you, when you figure it out. And then as a CIO, CTO, CISO, oh my God. How are we going to, how are we going to wrangle this? And, you know, I don't know yet, but there's going to be a whole vector of innovation around that. Um, you know, there's going to be some kind of dashboard, some kind of Splunk-like something, you know, for agents like a SIEM, you know, or, you know, event management. What does that look like? I don't know, you know, but let's build the agents and then, and then when we get there, let's let them talk to each other and then try to stay ahead of that.

But that's kind of what's on my radar. And that's really also what I'm excited about. I think in the near term is, I think most enterprise-level AI adoption will happen through upgrades from existing platform, whether it's Salesforce, HubSpot. So they're the one who will add some meaningful AI capability. But I do believe that, and I'm hoping that the real future of AI is when we can can finally be empowered individually to create AI agents that really serves us in a confidential matter, in a, you know, safe matter. That's really something that for our personal tasks, that's really something. And then the day that we'll see a real robot, I'll pay $10,000 for the robots to be in my home and do all sorts of things. That also is something that I'm very excited. So, okay, fantastic. Great. Well, we've answered questions, we said what we had to say. We have 5 minutes left and we're gonna, you know, stop on the dot. Thank you for all the great conversation. But before we wrap up, maybe we can go around the table one last time for, you know, any quick takeaways from each of you or one key thought or reflection from today's conversation or some advice that you want to leave people with. So why don't we do the opposite? Todd, do you want to—

anything you want to share? Yeah. So I think we're, you know, as a servant leader in consulting, you know, we've The technology has had a really good run at this ML and AI. It's time for us to help everybody else in the C-suite along. Um, and then I think that starts with the ground up, you know, personal productivity literacy, right? You got to go from the typewriter to the word processor to Word on your laptop or your desktop, whatever, you know, so let's, let's, let's teach those skills. Let's then collaborate and then let's extend that to the enterprise. And then I think as leaders, we can help you begin with the end in mind, and they help you crawl, walk, run.

I think that's my takeaway. Very cool. Thank you for that.

Brian, any last thoughts? I did ask my AI agent to make me some dinner, but it just ordered Uber Eats and took credit for it. So, I don't know if that counts. I don't think that AI is here to take our jobs. I do think it is here to take our job descriptions and rewrite them. And I think that the leaders who are going to win in this new world that we're heading into are going to be the ones that learn, leverage, and lead with AI.

Very cool. Love that. And great joke. Loved it. What about you, Kevin?

Maybe jokes or last word? I'll just stick with last words. How's that? I think that there's a tremendous responsibility on leadership to really look into how they want to apply AI in their organizations. And one of the things that, you know, it's a little bit concerning is when, you know, leadership's first question about AI is how many people it can replace, you have a leadership problem, right? And it's not about replacing people, it's about how do we work with AI? Because ultimately, there's examples out there right now of companies that have let software engineering folks go and they've let people in HR go and AI was going to solve all the problems and then discover it doesn't, right? So we have to figure out a way, how do we remain human and allow us to do what we do best while leveraging AI to do some of the more tedious repetitive work and some of the quick research that it can do. So I think there's a tremendous responsibility on leaders to responsibly apply AI and give some real thought to how people are going to work with it and get the best out of people.

Love that. I'm glad that you're putting the light on leaders. Well, that was a wonderful conversation. Thank you all for sharing your insights. A big thank you also for everyone in the audience for joining us. And so feel free to connect with any of us on LinkedIn. It's on the page. Follow TechCXO for updates on upcoming events. And then to close on a lighter note, I have one last joke which made my daughter laugh. So, and here you go. My coworkers are afraid their jobs will be replaced by AI. And I told them, don't be afraid. Most of you already have intelligence that's artificial. Oh, burn. That's it. Thanks again. And take care and see you next time.

Thanks for having us, Virginie.

Appreciate your moderation. My pleasure.

Thank you, everyone. Thanks for joining us in the Sky Lounge.

Please like and subscribe to hear more from Tales from the Sky Lounge here and anywhere else you get your podcasts.