Episode 107 - Ryan Shenefelt

Erik Martinez: Something I keep running into with business owners and leaders I talk to; they're using AI. They're getting some value out of it, but mostly they're using it to get things done faster, not to think differently. Today's guest changes how I think about that. Ryan Shenefelt is a digital marketing and AI innovation lead at de Novo Marketing.

Erik Martinez: He's been in this work for 13 years and one of the things he does that I find genuinely useful, is that he uses AI to argue against himself before he makes a decision. Not to confirm what he already thinks. To find holes in it.

Erik Martinez: Ryan, [00:01:00] welcome to the show.

Ryan Shennefelt: Erik, thank you for having me. It's great to be on and I love talking about AI and love, an AI focused podcast where I can really geek out and have some fun with this.

Erik Martinez: Yeah. I remember when we met and we geeked out for a couple hours on everything, AI. So,I've been looking forward to this conversation. Ryan, before we dive into that topic, can you just provide the listening audience, a brief synopsis of who you are and what you do?

Ryan Shennefelt: Of course,I've been at de Novo, in a variety of different capacities over the last 13 years. I started as a website developer, transitioned into paid, digital as well as organic social. Really the full digital landscape at the time, 10 years ago. Migrated into some account management and then, moved into a new role overseeing the digital team.So, overseeing both of those efforts and then when I shifted into this role, one of the additional titles I previously had of, AI and innovation.

Ryan Shennefelt: It didn't go away, it just got absorbed. So I also have the opportunity to look at all AI and innovation efforts [00:02:00] and lead our task force on that at de Novo too. So even outside of the digital space, our task force is in charge of AI and innovation at the agency as a whole.

Erik Martinez: So what you're saying is like the rest of us, you wear about five different hats.

Ryan Shennefelt: Yes.

Erik Martinez: On any given day, there could be a different set of hats tomorrow.

Ryan Shennefelt: Usually each hat is on every single day. It's just a matter of quickly pivoting between those. I also still, have some account management roles as well. So, that has been something that I never want to get rid of. That's my favorite part of agency life is the, interacting with clients, all of that.

Ryan Shennefelt: It also gives us, a direct link into the needs that our clients have and this industry is quickly pivoting. We know that because we're, on the pulse of all things AI and all things digital, right? But. you can't always force your clients into things until they're ready. So it's, it has been great keeping this account management side of things as well so that we know the pulse of our clients and our clients will be very different.

Ryan Shennefelt: We've got [00:03:00] hundreds of clients across different industries and it's just important to be able to tell or to determine those signs of when the client is looking for more AI or when they're still very skeptical. And also navigating the different pieces of, AI compliance at all of these different businesses, right?

Ryan Shennefelt: It's not just our AI policy, it's theirs now as well in helping them form those as well as working within their policies and trying to make it mesh with ours.

Erik Martinez: Yeah. And that's assuming they even have a policy.

Ryan Shennefelt: Right.

Erik Martinez: You know, it is funny, I was talking with an attorney friend of mine and we were talking about AI and he and his business partner areon the extreme conservative side of this topic, right?

Erik Martinez: They're not really using AI. I mean, he's kind of dabbling with it for little things, but I was talking to with him about like, Hey, you know, there's ways to use AI that doesn't actually impact your actual work, but can help you. It can help you prep for your trials and it can help you do this stuff.

Erik Martinez: But then we started talking about governance andgot onto a totally different topic [00:04:00] that was more HR related and realized, hey, you don't even have a policy on that.

Erik Martinez: Let's tackle that problem. But, we're not gonna talk about lawyers today. We're gonna talk about, AI and marketing and getting started, because I think, one of the hardest things is like we're all playing with these tools.

Erik Martinez: Even the people who tell me we're not doing much. They're using chatGPT or Gemini, or they're using AI overviews when they go and search. And, they're finding interesting little use cases both in their personal and professional lives. But I think when you first get started, I actually taught this,group of newbies yesterday. They're all familiar with chat GPT, but they're not really using it. And what was really interesting is I really started super basic. Well, I felt like I started super basic and I quickly realized oh, I'm six months down the capability curve. And they were right at the beginning.

Erik Martinez: So, the first question for you is, when you first got into AI, what did you try that [00:05:00] you thought was too complicated and how did you simplify it down so that it actually worked?

Ryan Shennefelt: That's a great question because when it comes down to it, you hear all of these things about AI solving everyone's problems, right? But I just keep going back to the hallucinations. I started my AI journey with Mid Journey. So, I was all about image creation. A little bit of daVinci back when that was big.

Ryan Shennefelt: So I started where you can see the hallucinations. Physically see them. And I brought that with me into the rest of my AI journey. So, I'm very AI forward, but also very AI skeptical, right? Because with the written word hallucinations are a little harder to find. So, some of the first things that I did, I was trying to consolidate information.

Ryan Shennefelt: I work in digital marketing. It's numbers. But I also am a marketer and an advertiser. I love the creative side of advertising, so I was really looking for something in order to help me with the numbers side of it to quickly analyze this information. And when I first started, this in 2023, 2022, I just [00:06:00] put in spreadsheets, I put in information, I put it all in there.

Ryan Shennefelt: And this was before Notebook LM, so I was leveraging chat GPT, and it was not doing a great job. It was too much. I was putting in too much information andI didn't have any questions that I used to validate the data. Now I've learned you need to ask it very basic questions to ensure that it understands the data.

Ryan Shennefelt: You might upload a PDF and it couldn't read the PDF at the time. It's still gonna try its best, but without those points of validation, you fall out. So, to answer the question, if I had to simplify it, it was start small, start with one piece of data or one file.

Erik Martinez: One simple process.

Ryan Shennefelt: Exactly, And then add into it. It's an iterative process. This is not like Google search anymore. You were asking it questions, you were making it better. And that was the piece. I was going too deep to start. And I think that some of my cautious tendencies were holding me back, but I did have to just trust the process, because it wasn't gonna be perfect right away.

Erik Martinez: Yeah. The statistics [00:07:00] say, something like, 93% of people never ask more than one question. And one of the things I've learned since I first started is, hey, if I ask a question and I don't get the right answer, I'll just ask it.

Erik Martinez: Why? this is the answer I was thinking I was gonna get. Why did you give me this answer? Oh, I didn't know that you were looking for X, Y, Z. And I think the, the probably the most important thing, for me now is give it more information. Be more specific. Be really as precise as you can be to get the best possible answer.

Erik Martinez: If there's one thing you would tell someone to stop doing today, like you're a beginner or even if you're, an intermediate user. Regardless, what's one thing you're like, don't do this.

Ryan Shennefelt: When I worked primarily in social media, one of my favorite things I could tell my clients was to stop posting on every single platform. If you don't have the capability and ability to post on every single platform, don't do it. [00:08:00] And I would say it is very similar feedback and a recommendation for the AI space.

Ryan Shennefelt: Don't keep vetting programs. Don't keep vetting models. Don't keep vetting AI partners. Work with what you have. At least for a few months. Whether that just be a standard text-based platform like Gemini, Chat GPT. Start on that, work on that, and then you will slowly figure out how you wanna leverage it in your company.

Ryan Shennefelt: Slowly figure out how you can leverage it at your business. Get your team on board, right? It's not just a matter of, oh, we've, we vetted this platform. We finally figured it out. Now go. Then you run into the same problem that all of these fortune five hundreds are running into where they're saying so and so percentage.

Ryan Shennefelt: It's not working for them. They didn't do it right. There will never be the silver bullet. Work with what you currently have. I would assume, almost every business listening to this podcast, they're either a Microsoft office or a Google office.

Ryan Shennefelt: And if you're paying for that, you therefore already have a very good version of copilot and a very good version of Gemini that has your [00:09:00] data locked down and you're not training the public model on it. So, start with those. There's always going to be something else that you wanna look into, and figure out.

Ryan Shennefelt: And once you figure something out, you're gonna feel behind, right? But incremental progress, all about building those slow wins and truly figuring out what works for you and iterating on it, you're not gonna find perfect right off the bat. Don't keep vetting new platforms, new projects. just finish the ones you already have.

Ryan Shennefelt: You will feel better about it if you just finish it. and keep going.

Erik Martinez: Thank you for saying that because I've been preaching that, but maybe the problem is I've been preaching. I really haven't experienced this and I see lots and lots of examples of companies and individuals going after a tool. Because, we've somewhat been taught over the last 20, 25 years that a tool will solve your problem.

Erik Martinez: And you know what's really interesting I find about Chat GPT, Gemini, Co-Pilot, Claude. Is that, right within that there's a lot of tools, that do [00:10:00] amazing things if youtake the time to learn how to use it. And it's really gotta start with you starting ask questions.

Ryan Shennefelt: And getting comfortable with it.

Erik Martinez: Yeah, and then playing around with the settings, and then playing around with deep research and then playing around with Notebook LM. You mentioned that a couple times. I'm a huge fan. They just dropped like five new features yesterday. I was just like, oh my gosh, this thing's getting so good. But that's a whole different conversation.

Ryan Shennefelt: These platforms are like the new Microsoft Word, Photoshop, etcetera, right? Previously you used to type things in Microsoft Word and it didn't come built in with spell check and grammar checker. Now it does. Now it's there. You would use Photoshop to go in and I'm sure you would edit the lighting in your photos, but you wouldn't use it to cut people outta the background.

Ryan Shennefelt: You'd use a different program. Well, guess what? Now they have all of those pieces. These programs and these early entry points are adding so many features. It makes me think back to digital advertising, or just organic digital when Facebook all of a sudden had reels, right?

Ryan Shennefelt: This was a few years ago. [00:11:00] It will become what you want it to be, I truly believe. Or you will figure out a way in order to do it with your platforms. Once you figured out how to automate or leverage AI for that one process.

Ryan Shennefelt: Now you can use AI to make that AI process even better. And maybe that is investing in a very specific tool or platform. But I think that you had mentioned people aren't leveraging the full capabilities of these entry points that people, have right at their fingertips because they just keep looking.

Ryan Shennefelt: We want one that specifically does digital advertising. Great. Perfect. You can do that with some of the opportunities that you already have.

Erik Martinez: Yeah, I totally agree. So let's pivot onto, a new question. 'cause I'm really curious. If you had to delete 80% of your AI usage and keep only one habit or thing that still improves your work every week, what would you keep and what does that look like?

Ryan Shennefelt: Yes, this is coming from an account management standpoint. So, I'll give two answers. [00:12:00] One, account management and two digital. I would say that from account management, we do a lot of summarizing of information. We have to take a lot of information that clients give us at kickoff points.

Ryan Shennefelt: So, all of their company data, a lot of information on projects, master plans for communities, etcetera. And the one thing that I would keep in AI as an account manager would be Notebook LM. Being able to put in all of that information. It does come down to trust. It comes down to trust because I worry about those hallucinations, right?

Ryan Shennefelt: I worry about it going off into Google and leveraging information that might not be true about my specific city or client. I wanna control that knowledge 100%. And that's why I've really, fallen in love with Notebook LM, because I feel like I'm in more control. As Gemini and chat GPT work to add, their sources.

Ryan Shennefelt: I still don't love it. Even copilot, it's adding sources and I'm like well, you're just sourcing A PDF, but you're not telling me the page of the PDF. And quite frankly, this PDF is 300 pages, so I need some assistance [00:13:00] there. But with Notebook, lm, I've really fallen in love with that. I'm an audio guy, so I love podcast.

Ryan Shennefelt: And that's something that I've been using more and more. The podcast feature that they then give you in order to dig into that information. So if I could only keep one tool when it comes to, AI and account management, it would be Notebook LM. From a digital standpoint, one of the big pieces of working at an agency is keeping your clients in the loop, right?

Ryan Shennefelt: Making sure your clients know what you're doing. And the big time suck for us as an agency is reporting. So, digital advertising, reporting is a piece of it. You'll work to build all these dashboards. You'll evaluate all these systems, but at the end of the day, clients don't wanna see dashboards.

Ryan Shennefelt: They pay you and they pay an agency in order to explain things to them. And that's something that, we have been leveraging AI for. Yes, everything is still vetted by humans, still approved by humans, but AI has a really great way of summarizing information. We still add our de Novo [00:14:00] flare and our de Novo spin to it.

Ryan Shennefelt: But, just having AI look at our dashboards. Look at our increases. It was difficult to get the dashboard in a way that the client can see it with their business name, but then still anonymize it in a way that we can feel comfortable uploading it into AI. But that has been a piece of it as well.

Ryan Shennefelt: A separate section of our dashboard, if you will. That talks about the updates that we make as we're going in and updating our accounts throughout the months. We report one time a month. However, we make updates two to three times a week. And, what's the fastest way to summarize that?

Ryan Shennefelt: Well, AI. And that would be the one thing that I keep when it comes to digital advertising. Reporting at the end of the day, we identified it as a whole, right? When it comes to our timekeeping software and time tracking software, reporting ended up being a big piece of how we were spending our time for our clients.

Ryan Shennefelt: It was important, so we knew we had to keep doing it. But, also it was the least favorite part of our digital advertisers [00:15:00] jobs. Our digital strategists, they hated doing the reporting. They love going in and optimizing, making changes, making recommendations. They hate the reporting. And that's where it's like, okay, AI, we don't want anyone thinking it's replacing their jobs.

Ryan Shennefelt: It's about doing the things you don't like to do. And reporting was one of those. So we're not using full AI reporting, we are using the summaries that AI is giving us based on the information that we're giving it still. And that's the big piece. So if I could keep one thing, reporting for digital advertising and Notebook LM for account management.

Erik Martinez: Yeah, I am so with you on both of those things. 'Cause I've felt the benefit of using Notebook LM. By the way, just so you know, Gemini now has a little feature where you can go in and access your Notebook LM notebook to feed your Gemini prompts. That's brand new. That is in last two weeks.

Ryan Shennefelt: Okay.

Erik Martinez: So, just for the listening audience, the reason Ryan's so up on Notebook LM, [00:16:00] he feels like he has control because it's grounded in your data.

Ryan Shennefelt: Only your data, right?

Erik Martinez: Only your data. You can ask it to go out and do research and bring external data in. But, you can keep it 100% grounded in your data, which is really, really powerful. I'm thinking, in terms of capabilities, one of the biggest capabilities for any marketing business, any department within a company, is having knowledge base of. Who, what, when, where, why, right? What decisions did we make? What processes do we have? All that stuff. All of that can now be brought into Notebook LM, and it can be summarized in a slide deck. It can be summarized in an audio file. It can generate a little video explainer. It is really powerful tool and I'm with Ryan on it. It is one of my favorites . If I had to keep one thing, I'd keep Notebook LM as well.

Erik Martinez: Uh, moving on to assumptions.

Erik Martinez: What's the one assumption marketers [00:17:00] make that AI is uniquely good at challenging? And how do you Prompt AI to stress test your plan before you spend money?

Ryan Shennefelt: When it comes to marketing, they always say you are not your target audience. That is repeated all the time, but it's still hard to get out of your head when it comes to that. Sure, you can go in and look at statista polls, target market analysis, you can do all of that. But, what I like to do or what we've done a few times, and we've seen some good success with on our smaller budgets. Where we don't have the budgets in order to test things as much as we normally like to. Running a plan through, chat GPT or

Ryan Shennefelt: running it through Gemini and saying, this is who your audience is. This is who you are.

Ryan Shennefelt: Would this plan reach them? That is a very easy thing to do.I do sometimes worry about, the training dates on some of these things, right? Because things progress quickly.

Ryan Shennefelt: But, One of the things that I think from a digital marketing standpoint, leveraging it to do some focus groups. Not only on your ad creative, but on your strategy. [00:18:00] On the strategy piece of things. Creative is one whole side. And at de Novo, we do have our creative team as well as our digital team.They work very closely, but they're two different groups of people.I'm talking mainly from our strategy side of things. Leveraging AI in order to see if you have a strong budget allocation, could be a really good opportunity. Saying

Ryan Shennefelt: this is our current plan. This is our target audience. Would you make any changes? Would you say, oh, let's move some of this from LinkedIn and move this down into just purely Google programmatic. And we've had some good success with that. But truly going in and, vetting your plans, not just the creative side of things, not just vetting the copy. Vetting your plans has been very helpful.

Erik Martinez: Yeah. You know, it's interesting that you say that because I have gotten into building synthetic panels.

Ryan Shennefelt: Yeah, exactly.

Erik Martinez: Yeah, I built synthetic panels with very specific focuses. And I have a cross section of different persona types that I use to evaluate what I'm writing, what my content looks like, [00:19:00] the tone, the point of view, originality, all of those things.

Erik Martinez: I bake all those things into it and then I get feedback. I use that feedback loop to help improve what I'm doing. So I think that's a brilliant strategy for using AI . And here's the best part, if you're AI skeptic, building these little personas, there's no risk.

Ryan Shennefelt: Right.

Erik Martinez: But I think you'll find if you do it and you do it well.

Erik Martinez: That you will get tips and ideas that you hadn't quite thought about. Or, you were thinking about and just oh, you know what, that's a nice spin on that. I'm gonna take that a little bit further. And it's a really low risk way of using AI to improve something that you do every day. So I think that's fantastic.

Ryan Shennefelt: And there's so many platforms out there that offer the AI focus groups. They're doing a glorified version of what you can do witha one paragraph summary.So again, just use what you've got. You'll make it work. And if you realize you don't like the time that's going into that, then look at some of those third [00:20:00] party offerings.

Ryan Shennefelt: But what you said, Erik, I want to zoom in on that. You don't have to do it. That's the benefit of still having that human control over things. We've evaluated platforms that will do agentic planning for you and automatically make those placements. We have seen, with some of our just very basic, like a Facebook performance or a, performance max plan, right?

Ryan Shennefelt: It will go in, it'll optimize your copy, and then all of a sudden the copy isn't right. All of a sudden they're cutting out part of your client's name that has to be in there. You still need that human control and leveraging AI in this, it still gives you 100% control. You have the ability to say, you know what, that's a very good point.

Ryan Shennefelt: Like you said. Or, maybe not. And back to your episode with Pat Berry, the last time he was on you had said that when people say, oh, I've got it to act just like me. You are not your target market. So, you still need to kind of get outside of that and use AI as a thought partner to poke holes in things.

Ryan Shennefelt: But give it the target market. Because you don't want it [00:21:00] to act just like you. You don't want it to act just like what you've previously done. You need it to challenge your thoughts. And that's where that's great. The employees say yes, consultants say no and why. Don't treat AI like an employee.

Ryan Shennefelt: Treat AI like your consultant on these things. You don't want it just to say yes and be glossing you all the time. You want it to challenge you and that's where it's a really good partner.

Erik Martinez: Yeah, I think if there's one thing I've learned over the last year for myself is using AI as a thought partner. Because you know what? Let's be honest, any idea that we have, if we could dump the contents of our brain into a chat window and have it summarized, that would be amazing.

Erik Martinez: Oh, by the way, we do have those capabilities now, but we're human and probably 50% of our thoughts actually make it out. Maybe less. I don't know. But I know for whatever I'm working on, I know that when I give the initial context, it's not fully baked.

Erik Martinez: So, I use AI to pull more [00:22:00] context out of me. So we can work together to come up with whatever solution. That's the way I've started viewing it. I view AI as, my digital doppelganger, who is also a contrarian,Erik I don't want to sound like you.

Ryan Shennefelt: Right, exactly.

Ryan Shennefelt: Just out of curiosity for your panel, if you will, did you create that in a gem or as a custom GPT? Because what I'm curious about is how you, how you stop it. Once you've had a conversation with a, with a chat for too long, it gets a little wonky, right?

Ryan Shennefelt: And you've got some limits on there. So do you have it as a gem or a custom GPT where it spins up a new instance of this panel? Of this group each time?

Erik Martinez: I do. When I build these things, I start with a project.

Ryan Shennefelt: Mm-hmm.

Erik Martinez: I start with a project and then I build it, and then I put it into a custom GPT. And then, about once a month I go in and tweak the panel.

Ryan Shennefelt: Okay.

Erik Martinez: I go in and tweak the panel because I've learned things in the previous month of using it that I think will make it better and give me better [00:23:00] feedback. Honestly, I've built this system. I built what I call a content accelerator, and it's a combination a briefing agent, it's a ghost writer and the evaluation panel, and I now have them working together.

Ryan Shennefelt: Yes. That's the ultimate goal. You've gotten to the goal of so many, but just start with one, right? Start with one.

Erik Martinez: Yeah. and it'll make your life easier. It will over time make your life easier. It doesn't at the start. It will take some more time. And I think that one hurdle right there, when you first start building these things, you're going to spend more time than you would have if you just did the thing.

Ryan Shennefelt: Yeah.

Erik Martinez: But, you'll start saving that time and being able to apply that time to either improving the quality of whatever you're working on or devote that time to something else that is high value and high impact to your organization. And then that's the way that I'm thinking about AI these days is, it's not gonna solve every problem.

Erik Martinez: It's not perfect. Every time they update the model something weird happens.

Ryan Shennefelt: Just like with [00:24:00] search engines, right? Search engines have models.

Erik Martinez: That's exactly right. There is a learning curve here. And if you have the opportunity to spend a little bit of time to experiment with something that you're struggling with, use AI to help you solve that problem.

Erik Martinez: You'll start building those muscles that you need to do more complicated work down the road, and it takes time. It's not gonna happen overnight.

Ryan Shennefelt: Spending 10 hours on a task that saves you 15 minutes, like, yeah, that's really frustrating. When you realize the time that went into it. But, it's that muscle. Maybe it'll then take you eight hours the next time to build a task that saves that time.You wanna be targeting those tasks that are recurring. It's 20 minutes a day, right?

Ryan Shennefelt: That it takes you and maybe 10 hours over the course of a month will then be like, oh, that has paid off in a month. But you do get faster with it as you go. And that's something leading AI efforts at an agency, does get difficult because we don't want them all coming out of one person or one committee.

Ryan Shennefelt: The, purpose of our AI committee at De novo is [00:25:00] teaching people how to get inspired. Inspiring those people and teaching them the steps in order to leverage AI in their daily work lives. Sure, we create processes that can be used by everyone at de Novo, but it's about inspiring that curiosity and truly teaching them like.

Ryan Shennefelt: It won't work right away. If this is something you need done in two hours, do it yourself. Do it yourself the way you used to do it. If it's a client deadline or something like that, you need some time. You need some iteration, and you need some grace in order to mess up and, and tell it it's wrong.

Ryan Shennefelt: But those are those muscles that you have to train and just get familiar with the more you play around with it.

Erik Martinez: Yeah, I think that's fantastic advice. So, Ryan, let's pivot over to decision making. Because I think now we're starting to get into the, how do I use AI for decision making and should I really trust the output? When you are using AI to evaluate a decision, I want to go this way, or I want to go that way, I'm not quite sure. What are some questions that you would ask AI [00:26:00] to help you pick between some choices?

Ryan Shennefelt: It kind of comes back to that control. I think that people, when they listen to this episode, they're gonna be like, oh, he's a control freak. Huh. But,

Ryan Shennefelt: I go with my gut a lot and I have historically gone with my gut a lot and I have gotten very lucky. I like to think it's because I can consolidate a lot of information in my head and make a good choice. It also might be luck, on a lot of things. And so that means when I'm making decisions, I have a hunch. I have an idea of where I'm at, and I like to leverage AI to prove me wrong.

Ryan Shennefelt: I will straight up tell it, I'm thinking this. Why not? Give me five reasons? Give me 10 reasons why not. And it's doing that gut check. It's telling me okay, you think this, what about this? And that's for very specific situations when I have an A, B, or C option. It's that, why you shouldn't do this.

Ryan Shennefelt: I'm asking it why I shouldn't do something, not why I should. Usually I have an idea of where I wanna go with something. So I need it to, to be that checker on me. But when it comes to managing people [00:27:00] and things like that, My gut is always to help, but really it's to give my opinion and that is not helpful when it comes to managing people because that can be very annoying for people who are very similar to me.

Ryan Shennefelt: So, it is going down and just saying what are questions that I can ask in order to, make this experience go better without just giving my opinion or helping solve the problem. Because I tell myself, I like to help solve problems and make things easier for people, but that also comes down to some control issues that I have, right?

Ryan Shennefelt: I wanna be able to control the situation. But, that's not helping my team members grow and be able to solve these problems for themselves. So I do regularly also use AI as kind of a leadership coach when it comes down to these things. What questions can you ask in order to inspire the answers from your team?

Ryan Shennefelt: And that has been very helpful as I have moved into this new leadership role. It's realizing that everyone has a leadership style, but everybody also has a way that they like to be led. And it's different for [00:28:00] every single person. That has been very helpful, eye-opening because it's quickly made me realize not everyone sees things the way that I see them.

Ryan Shennefelt: And, just saying, well, that makes sense is not a good way to inspire people.

Erik Martinez: I'm curious, is there one favorite question you love to ask the AI when you're challenging yourself on a decision?

Ryan Shennefelt: The same way that you have panels of different audiences, right? I've kind of boiled down my team to how they react to things. So, I have asked it questions, if this is the person you are talking to, how would they respond to this?

Ryan Shennefelt: How do you think that they would prefer, some of this information being sent back to them? We do a like a personality test of some sorts and it gives you results on how you like to be communicated with all of that. Very similar to true colors, a little bit more proprietary.

Ryan Shennefelt: So, we had to, anonymize it, but then leverage that. And it's like, okay, if you're this employee, how would they respond? And I think that from a [00:29:00] leadership standpoint, that has been very helpful. But also when it comes to vetting opportunities, vetting data, vetting choices.

Ryan Shennefelt: I ask why, but, it's not just as simple as asking why. It's giving it the context of why you're skeptical. So I don't have that one killer question because it's different for each instance, but it goes back to my gut. There's a reason why something's not perfect and there's a reason why I don't think something makes sense.

Ryan Shennefelt: And it is saying based on X, Y, and Z, why would you recommend this? And then it'll gimme the answer. It'll gimme the prompt, and then it's why would you recommend it? So I always like doing the, doing the pro and the con and having it explain it, because then that's helping me build this Venn Diagram in my head of the full picture.

Ryan Shennefelt: So ask it one question and then ask it the opposite question. My one question is two questions, and it's why and why not.

Erik Martinez: Yeah.

Erik Martinez: I think that's a really great way of using it as a use case. That for whatever you're doing, you can ask why or why not. If it makes a recommendation on a tactic, you can [00:30:00] ask it why or why not? You could just literally, without putting any more context in, you could ask those questions and then you can ask some more questions. And that is one way that, again, low risk, right?

Erik Martinez: You don't have to follow any of the recommendations or the advice or thoughts that you're getting through, but you can learn from the tools by asking, the why not question? I think that's fantastic.

Ryan Shennefelt: And the piece I want to add to it. I don't ask it in the same question. I don't say why or why not. AI generally tries to give you information in a similar format, right?

Ryan Shennefelt: It tries to keep your character limit roughly the same. I don't want it thinking partially on why and partially on why not at the same time. Right? I want it to be fully in the camp of why. Fully in the camp of why not. So, I have separated those and I feel like I get better success by separating them versus keeping them all in one question.

Ryan Shennefelt: Depending on how many, prompts you're limited at, I know my co-pilot will limit me after I think 200 prompts. Maybe you don't wanna [00:31:00] do it in there, but, my Gemini, I haven't been limited yet, so I'll keep going.ButI do those two questions in two separate prompts.

Erik Martinez: I think that's actually super important. I grew up in direct marketing and segmentation is hugely important. Paid search, paid media segmentation is really important. Audience, is one of the most basic forms of segmentation there is.

Erik Martinez: And when you write your questions in chat GPT or Gemini, or Claude. You've gotta think about starting to segment your questions.

Ryan Shennefelt: Yeah.

Erik Martinez: I think you're right. I get much better results when I am segmenting questions than trying to ask it to do too many things at one time. In fact in the training group I got yesterday, somebody asked me how much context is too much?

Erik Martinez: And I'm like, there's no such thing. Keep dumping. However, when you're giving instructions, you need to be very explicit about your instructions and what you want it to do. Because, think about the boss that they just dump 50 priorities at you without any other [00:32:00] information.

Erik Martinez: The AI's brain does the same thing. It's gonna try to answer every one of those questions, but it doesn't know how to prioritize the information. It's getting better, the higher end, more expensive models do a better job of this, but they're not great at it.

Ryan Shennefelt: Yeah.

Erik Martinez: And so, you know, breaking that down into smaller pieces, I think is a really great thing.

Erik Martinez: All right. One last question before we close out. Is there one thing that you are like, Hey, you didn't ask that question, but I want to tell the audience about it anyways.

Ryan Shennefelt: Hmm.

Erik Martinez: You know, It's the typical last piece of advice question.

Ryan Shennefelt: We've skirted around it, a few times and we have skirted around the topic of trying new things, vetting different programs, etcetera. And, it's just start, right? I had been looking at vibe coding for so long. I have a brief computer science background.

Ryan Shennefelt: I took, two computer science classes in college. Realized it wasn't for me, but that gave me enough. I know how to work Python, but I do not know how to write my scripts, write my code myself. [00:33:00] Vibe coding was something that I really wanted to explore or look into, especially when it comes to connecting APIs.

Ryan Shennefelt: We use a bunch of different systems at de Novo. Workamajig, many of your listeners might be leveraging that for agencies. But, APIs, I know enough to get messy and I wanted to explore the space of vibe coding. I subscribed to so many newsletters, so many people on YouTube about vibe coding. Over our winter break we're off between Christmas and New Year's.

Ryan Shennefelt: I was like, Just do it. Spend the $20, get the Claude Code subscription. The one thing that you're like, oh, I wish I could figure this out. Just do it.It comes down to that Shia Labeouf meme where he's just like, do it. Just do it.

Ryan Shennefelt: Do the one thing that you've wanted to do with AI. You will feel so much better. It also allowed me to then inspire my team to show them this is what I was able to do. I figured it out. It's something I've been wanting to do for a while, and I just did it. Then people they feel inspired to do the same thing.

Ryan Shennefelt: It might not even be the AI piece of it. It's just the one thing that you've been thinking about. Write it down on your post-it note and just do it. And I [00:34:00] know it's stupid listening to somebody on a podcast tell you to do that thing, but, it will make you feel better. And if it is AI related, it keeps building on it.

Ryan Shennefelt: It's that muscle, like you said, Erik, that you, you have to hone. And once you figure out how to do something, oh my gosh, you can do so much more. So, getting into the realm of Claude code, all of that. I didn't wanna add in Claude to my tech stack, but I did.

Ryan Shennefelt: Claude Code, I was like, I've heard enough about it. I know that this can work. Sure, I could use all these other tools. But I had to just get in there and do it. And now I know, okay, this is how you set up your instance. This is how you set up your visual code.

Ryan Shennefelt: All of these things that, all of these other platforms still use. I got in there and did it and felt so much better. Oh my gosh. And, yeah, the one thing, just do it.

Erik Martinez: Isn't that the Nike slogan?

Ryan Shennefelt: Yeah. Correct. Just do it.

Erik Martinez: The funny part, I just got done reading Shoe Dog.

Erik Martinez: Which Phil Knight's, autobiography and story of how he built Nike.

Erik Martinez: Really enjoyed it. And look, you don't have to like Nike [00:35:00] or not like Nike, but it is a fascinating story of how he just did it. Everybody thought he was crazy in the 1960s that he wanted to sell running shoes. It's absolutely insane.

Erik Martinez: His dad loaned him 50 bucks so he could do this. 50 bucks in the early 1960s is a lot more money than you think. But, he started on a shoestring budget. He started with a trip around the world and he learned things and he tried things and he did things, which almost drove them into bankruptcy multiple times throughout Nike's history.

Erik Martinez: I think the key lesson that I got out of that is you do have to try things. You do have to take some risks. But the cool part is none of the things we do with AI is a risk. You can try it. I mean, don't put your clients personal information in there. Don't put their customer list in there, don't do any of that stuff.

Erik Martinez: But, if you're just trying to experiment with something and you wanna learn how to vibe code, I'm actually using lovable right now, and

Erik Martinez: I totally find it [00:36:00] fascinating, in a couple of hours I built a landing page with a system behind it.

Ryan Shennefelt: Yeah.

Erik Martinez: A database system. I also know enough about databases and stuff that's some of my background.

Erik Martinez: But, in just a few hours, I was able to create this pretty useful, interesting, customized to me landing page that has a very specific set of functions. And I didn't need to go buy some other tool to do it. So I'm totally with you.

Erik Martinez: Look, we are out of time. Ryan, thank you so much for coming on the show. This was so much fun.

Ryan Shennefelt: Yes, Erik, thank you for having me. I knew that when we first chatted AI, that I could talk AI for a very long time. So, thank you for bringing me on. I know that we ventured outside of the digital marketing space, but really you can use the things that we talked about in any specific instance in marketing and just in your AI exploration.

Ryan Shennefelt: So, thanks for having me on and letting me gab for an hour.

Erik Martinez: Oh my gosh, it was so much fun and we're gonna have to do this again.

Erik Martinez: The main idea that stuck with me from this [00:37:00] conversation is how Ryan uses AI. He doesn't ask it to agree with him. He asks at why he's wrong. He'll have a hunch, tell the AI where he is leaning, and then ask it to give him every reason not to go there. He then ask the opposite question separately, so he's getting the full picture instead of a blended answer that hedges both ways.

Erik Martinez: That's a different way of using AI that most of us are doing right now, and it's low risk. You don't have to act on anything it tells you, but you might see something you would've missed otherwise. That's it for today's episode of the Digital Velocity Podcast. Thank you for listening and have a fantastic day.

Narrator: