Episode Transcript
[00:00:05] Speaker A: Welcome to Monument Matters, a podcast produced by the Monument Builders of North America for all things memorialization.
Each episode is an extension of our monthly magazine, MB News. Monument Matters invites everyone to listen and share. You'll find all of the episodes on Apple, Spotify and YouTube.
[00:00:25] Speaker B: I'm your host, Mike Johns, CM AICA from the Johns Ciarabelli Company Cimarano Monuments and Flowers in Cleveland in Ohio. I'm also a past president of Monument Builders in North America.
Our guest today is Mike Lannan, the founder of Eternity, a leading web design and AI consulting firm based in Burlington, Vermont.
After Mike presented at MB University in February, he worked with MB&A's board of trustees to write a four part live but virtual AI Boot Camp series specifically for Monument Builders. Today he's sharing details about the series, but also sharing practical tips to utilize AI tools. So Mike, welcome to our podcast.
[00:01:09] Speaker A: Thank you so much. Pleasure to be here.
[00:01:12] Speaker B: Glad to have you.
Mike, what did you hear from Monument Builders after your MB University presentation that convinced you a deeper industry specific AI training series was needed?
[00:01:24] Speaker A: Yeah, great, great question.
So, kind of a combination of, of several things.
Some folks were asking for that in kind of an indirect way. They were like, do you have any kind of ongoing training that can kind of keep us up to speed on all this ever changing AI technology? Other attendees from the conference were wondering if it could be longer if that. The session that we had, which was multiple hours, they were like, hey, could you just make a whole day of this? So there, there definitely was a desire for, for more information, more knowledge and in kind of ongoing, an ongoing manner. And I started to see that pattern in other associations that I presented at and kind of just came up with the idea sort of on the fly and offered it to several associations. And I've done it with many in the past to date and that's kind of where we are now.
[00:02:15] Speaker B: All right, well, I think you have your work cut out for you today. At least from where I sit, I'm still on the fence.
AI good or evil? I don't know.
Let's see where I end up by the end of today.
So when people hear artificial intelligence, they may picture complicated technology. How would you explain AI?
[00:02:37] Speaker A: Yeah, so I would say AI is just like any other tool really. It kind of depends on what you want to do with it, what your expectations are and any, any kind of pre disposition to. To AI through science fiction slash the news can kind of create this sort of jarbled view of like what is AI? Is it evil? Is it going to End the world? Is it going to use all of our water? Is it going to replace our jobs? I don't have a crystal ball though I do pretty strongly believe that the cats out of the bag, the toothpaste is out of the tube, insert any other acronym or any other synonym there for, for that, that it's going to keep going and competitors are going to use it in some way. And there's, there's definitely ethical, efficient, practical ways that, that you can use AI, especially if you view it more like an assistant that's there to help you and not a replacement of you or, or a replacement of your teammates. It's not perfect. Humans are not perfect. But if you put in enough effort, you can get these systems pretty, pretty set at like an 80% effective output right now in like 60 seconds. You know of doing a proper prompt or pro, a proper training. So you put in enough, enough initial time, you train the system, you give it access to documents or past examples of things it will start to learn.
Just like when you hire a brand new employee, day one, they're not going to be like producing for you at 100% or even like 10%. They need time to kind of learn. And that's my kind of basic mantra and model in my training is to, to show folks how to train their AI, how to set it up for success instead of it just kind of being a generic robot sounding output of a system.
[00:04:40] Speaker B: Okay. So that I'm going to go off script for a little bit. So please stay with me. I hope you can. So what I think I hear you're saying is that there's a certain amount of teaching that I as an AI user will need to do.
So AI will work for me, is that what you're saying?
[00:05:02] Speaker A: A hundred percent. So kind of imagine the first time that you set up Chat GPT or Claude or Copilot Gemini, whatever your tool is, that is an employee or an assistant, you know, for you coming to work the very first day. But they also have no access to any information yet. They don't have company history, your target audience, examples of what good looks like to you.
So if you don't give, if you didn't give a human access to that stuff, how could you ever think that they could create good output for you or things that sound like you look at AI the same way? I think, I think understandably folks will see things either like in the news or on a show and just kind of expect they can say computer, make me this perfect thing, entertainment. And it just come out in like a one shot, perfect, ready to go.
And that's kind of a quick way to set yourself up for disappointment. Because just like an employee, it's not going to be perfect the first day, the first week, or even the first month. It can take a while to get that kind of onboarded. But once it is, then it can consistently again. Maybe it's never 100% every single time, but neither are humans. If you can get something to be at 80 or 90% and then you have to fill in the blanks with your own human brain, you still have to work.
You can still save a ton of time on that kind of initial grunt work or output or whatever that kind of content might be.
[00:06:35] Speaker B: Okay? But I mean there's, there's already my mountains and mountains of information on this Internet, this wide world web, right? Why isn't that information enough?
Why do we need to do our own teaching?
[00:06:52] Speaker A: That's a great question. So think of it. Think of these kind of base models or the base kind of setup of these things.
Like a generalist. It has access to, like you said, lots of information and lots of different processes, lots of different, of what good looks like. But your organization might have a very different view set or policy on like what good looks like, what your who, your target audience is, your tone of voice, colors, fonts that you should use. You can train these systems on all of that information.
So when output comes out, it looks like somebody in your organization actually created it.
[00:07:33] Speaker B: Well, I, I get that part. But you know, I think too often we assume that something that we find on the Internet is factual and accurate, but that's not always the case. Right. So is there an AI police? Is there someone out there that says this information is true, this information is false?
How do you, you know, when, when you're talking about facts and figures or those kind of things, how do you determine that the sources that AI is pulling from are accurate?
[00:08:09] Speaker A: Yeah, I, I think it really hasn't changed that much from, you know, from the past. I'll date my own self here a little bit. When I was in like grammar school, in high school, we had the, the card catalog, still the Dewey decimal system, periodicals. What do they call, I think it was called microfiche, where you could like look at articles and things like that. And you had to go look up sources, do research, cite those sources.
AI can do that very, very, very quickly, like at, you know, at an exponential rate. Can go do research for. You can go cite sources. This might come to a surprise as some. But not everything on the Internet is true. There could be a source that looks like it's a legitimate source, that is framed as legitimate, that is written as legitimate, but very well may not be.
AI may choose to look at that as a center of truth and pull that in. What I always do when I'm prompting. And there's anything to do with numbers, financials, dates, things like that. Even though most of the time these AI systems will cite a source in the right hand column while it's doing research.
I will always kind of explicitly tell it, provide the sources, make sure to have a clear link.
So if there's something that I'm like, I don't know, that sounds a little fishy, I can click on that resource, go to that website.
There's still kind of human brain involvement there of like, does this look like a legitimate source? Can I find three places saying that this is true to at least back it up? So all that to kind of say it hasn't really changed that much, but it has. It has become significantly easier for anybody with two hands and fingers to just sort of button mash, write me a report or do research and get something written and not double check all of those sources because they might not have read them. They went through and read it for them.
So trust but verify is sort of the takeaway there.
[00:10:15] Speaker B: Right? I got it. All right, so what are some of the most useful tasks AI can help with in a monument business today? And where should a company avoid using it?
[00:10:25] Speaker A: Yeah, so I would say I'll start with the avoid side, you know, especially in monument and memorial business. Even though it could be appealing to consider like an AI website chatbot, you know, on, on a website or an AI kind of phone reception answering system.
I help a lot of like hospitality and kind of large, large turnover companies set up things like that.
I don't think that that has a good place at least yet in the industry. Like maybe give it another 50 years and folks will become accustomed to talking to a robot about the, you know, the loss of a loved one. But I'm not there yet. I use AI all the time and I would want to talk to a human being that has gone through this process of grieving themselves and can kind of walk through it. So kind of starting with the, the stay away from side of things. I would say anything to do with like direct communication, meetings, things of that nature. On the flip side, there's tons of behind the scenes things that the, you know, the consumer doesn't see or never needs to see that could be supplanted with AI that just speed things up. So some of those examples could be proposals, estimates, things of that nature, depending on like the software, or if you're not using software to create estimates, you can create something very easily and very custom Inside Claude or ChatGPT that has dozens of examples of your proposals, Excel sheets with your price sheets inside there. And then after you're done having a meeting with a potential customer, take all those notes and dump that into your system that you've built. And you might be able to come up with like a 90% perfect proposal in like 60 seconds. And then just kind of double check things from there. Yeah, exactly. So any, basically I, I see it as anything that is not like, at least for this industry, not super kind of consumer facing behind the scenes things, administrative stuff. And then of course, on the design side, there's tons of things that can be implemented into workflow for creating mockups, for designing monuments. The AI tools have come so far now that you can upload like a basic example of an existing monument or a blank, you know, of a monument, and you can just give it a clip art, give it a graphic, give it text, and you could mock up 15 different versions of something in a matter of seconds. Will that version be like perfectly ready to go to send to engravers? No, but you can again send somebody a bunch of ideas really quickly, and then once that's locked in, have the actual like producer designer create that in the system that they're using so it's, you know, set for integrating. So those are kind of my examples for things of where it kind of can make sense to use it.
[00:13:30] Speaker B: Sure. I think, you know, when you talk about the chat bot and the answering and that kind of thing, you know, what pops into my head is live agent.
[00:13:41] Speaker A: Live agent, human.
[00:13:43] Speaker B: You know, so you have to put yourself in a position of being a consumer when you're making these decisions. Right. How do you want to be treated on the other side of the equation? And if your expectation is this, then, you know, that's how you should be modeling how you're facing your customer.
[00:14:03] Speaker A: Exactly, exactly.
[00:14:04] Speaker B: Makes sense.
All right, so the first boot camp session focuses on AI fundamentals and security.
What information should monument companies never enter into a public AI platform and what other safeguards should they establish?
[00:14:19] Speaker A: Yeah, so this is a great question and one of which I find a lot of folks very understandably not realizing that the free versions of these tools, there's many, most of these versions have a free trial or a free forever version. This, this was kind of the case before AI, but for better or for worse, we humans have kind of become currency in a way. And in data of like, what we do, what are, what actions we take, what our emotional state is. I'm in digital marketing and it still freaks me out how much information is tracked about us. But inside these free versions of tools, by default, when you scroll through that little legal agreement that most folks do not read and they say I agree and create their account, you've given it permission that anything that you dump into that chatgpt, Claude, Gemini, whatever it might be, that that company can, can use that data for whatever they want and they can train the system. Parts of your data might come out inside somebody else's output. Probably not all of it, but a piece of it could.
So if that doesn't feel, if that feels like ick like it does to me, you can turn that off in the free versions in the boot camps, we'll be showing in all the systems how to turn off that, that feature where by default it's giving access to these companies of your data. It's literally just a little toggle button that you can turn off, but it's under like six menus of hidden things.
[00:15:56] Speaker B: It's there, but they don't make it easy to find.
[00:15:58] Speaker A: Exactly. And they also typically call it something like improve the model for everyone or make the model better. When like in small print it says we are tracking like every bit of data that you use.
[00:16:11] Speaker B: But there is ways, there are ways to be more private, even in those free versions.
[00:16:20] Speaker A: Indeed. Yep. So you can feel, you know, I guess, as safe as things are safe, you know, in this kind of digital world, like as safe as you might feel putting it into copilot or into OneDrive or Google Drive. Yeah.
[00:16:34] Speaker B: So then does it pay to pay it?
[00:16:39] Speaker A: Yeah, I mean, I may be like bias of this a little bit, but I think if you, even if you are a business of, you know, of one to sign up for these paid versions, usually the monthly cost is no more than like 20, 25 bucks. And not only does that by default turn off those kind of sketchy tracking systems that are tracking all of your data, it makes it so folks can't accidentally forget to turn, turn that feature off. It makes it secure out of the box.
The paid versions also give you near infinite extra abilities that you can do that. Turn it from like a, like an unpaid high school intern in the free versions to like a $25 a month super employee that's working, you know, 24, 7 for you securely and trained on your data.
[00:17:31] Speaker B: So then you do get what you pay for?
[00:17:34] Speaker A: Yes, very much.
[00:17:37] Speaker B: Well, it's good to know. I mean, you know, listen, if I could find the cheap way, I'm all about that. But, you know, especially in today's world with privacy issues and identity theft and intellectual property rights and all of that stuff, it's way more complicated than saving a dollar here or $10 there.
[00:18:00] Speaker A: Exactly.
[00:18:01] Speaker B: It's good advice to watch that and I will be looking forward to learning how you could do some of those things to make the freer versions safer for use.
Very good. So each boot camp session includes a volunteer complete completing a task live. All right, very good. Why was it important to show the process in real time rather than simply present finished examples?
[00:18:24] Speaker A: Yeah, great, great question.
As much as I like kind of getting up in front of the, you know, whether it's a virtual room or, or a physical room and people seeing my smiling face talk about AI, I've done like 11 billion of these before and I've always found folks seeing another peer do it, it just kind of files differently inside their brain or like, oh, if that person can do it, I can definitely do it. Or they might even know that person and they may know that that person is like, maybe hypothetically, not always the best with technology, but they're able to do it because kind of at the bottom, bottom lining this, this technology is not that complicated. It's more behavioral in thinking about this software as, as more than just like Microsoft Outlook and kind of framing it into more of a, like I said, an assistant in training it.
[00:19:17] Speaker B: Yeah.
[00:19:17] Speaker A: So I found that as a way to kind of find something that's very relatable to others that are watching or wanting to learn. It also gives me really good practice on kind of behind the scenes how somebody is using it or wanting to use it in their workflow.
[00:19:33] Speaker B: Yeah, I think that makes sense. You know, when you see the finished product or you see step one, step two, step three, there's. I don't care how good a presenter you are, there's always something 3a or 2b exactly. That you didn't write out, that you didn't say, it's right here. And so being able to watch someone go through the process, I'm sure you, you, you don't miss or you have the opportunity to review.
Exactly. See those, the, the, the fine tuning, the little piece that, that gets us
[00:20:13] Speaker A: to the finish line.
[00:20:14] Speaker B: Get. Yes. And that frustration line for sure.
[00:20:17] Speaker A: Yes.
[00:20:18] Speaker B: Very good. All right, so by the end of the four AI bootcamp sessions, which are
[00:20:23] Speaker A: how long each they are going to be 90 minutes each, I believe is what we.
[00:20:28] Speaker B: Wow. Okay.
[00:20:29] Speaker A: Yep.
[00:20:29] Speaker B: And I assume that you can start and stop and you don't have to digest 90 minutes at a crack.
[00:20:34] Speaker A: Correct. Actually, let me course correct that we are 75. 75 minutes. Sorry.
[00:20:40] Speaker B: Okay, 75. Very good. That's still a lot of. A lot of information.
[00:20:44] Speaker A: Indeed.
[00:20:46] Speaker B: So what should a participating monument builder feel more comfort confident doing after the four sessions? And what would you say to an MBA member who is interested but unsure whether they are technology technologically experienced enough to participate?
[00:21:03] Speaker A: Yeah. So 30,000 Foot View folks will have a better understanding of security and privacy and how to set up their AI to be not only effective but safe.
[00:21:13] Speaker B: Safe, exactly.
[00:21:15] Speaker A: Which I feel is like the most important thing. Before you show any fancy features like your data should be safe, then I will be talking about how to practically use it inside workflows, both marketing content creation, kind of outbound social post, things of that nature as well as from the design side of things. Folks will be able to create kind of quick mock ups of monuments or of scenes for monuments to go into just to kind of see how things will will look. So a very kind of wide net but laser focused inside each session on one particular thing. So again to kind of recap that you'll understand how to use it safely, securely and effectively in both marketing content design and operations. So folks will hopefully feel a little bit less like they're just blindly walking through a tunnel of AI and have some directions of where to start and and where to go from there.
[00:22:13] Speaker B: Sure. There's safety in numbers, right?
[00:22:15] Speaker A: Indeed. Yeah.
[00:22:16] Speaker B: Got it.
So I guess it doesn't really matter what AI platform for. For lack of a better term you're going to use. This will be this introduction. These sessions will be translatable across different AI platforms.
[00:22:36] Speaker A: Great question. Yes. Ish. So some of these systems can and can't do things. Some of them are really good at creating like photos or photorealistic outputs.
[00:22:48] Speaker B: Okay.
[00:22:49] Speaker A: Others are significantly better at human sounding content like say you were wanting to write a blog or any kind of like long form content.
So I will be showing off some differences between Claude, ChatGPT, Gemini and Copilot. So folks will kind of get a sprinkle of each.
I am kind of Claude bias in that I'm like team Claude kind of all the way. The one sort of kind of negative of Claude is it does not have a graphic kind of output creating photo system like ChatGPT does. So I'll be showing some things inside ChatGPT that are much more photo focused and some things inside Claude as an example that are much more kind of human content planning strategy. It's more. Claude is more of a thinking partner.
[00:23:42] Speaker B: Okay, so can we, can we dumb it down and say that this AI tool is a hammer and this AI tool is a screwdriver, and if you want to do, if you want to screw screws, you're going to use this platform. If you want to hammer nails, you're going to use that platform because they do it better.
[00:24:03] Speaker A: Exactly. I use that analogy all the time, that it's kind of like a tool belt or a toolbox and there might not be one, one tool that can do all of the things. You know, if you're going to be, you know, creating a deck and all that you have is a hammer and no saw, that's going to be. I mean, you could do it, but it's gonna, yeah, it's, it's not gonna look great and you're gonna get really frustrated. So also, another thing, another thing that folks will be much clearer on is like, which tool is better for what? And not to say that you need to have four or 40 of these things. You might be using two of them, you know, tackle the thing.
[00:24:41] Speaker B: But understanding their strengths and weaknesses, I think it's a very valuable part of the process, for sure.
All right, so I'm pretty much out of questions, but I don't want to necessarily wrap just yet. Is there anything else that you want to tell our audience about the process, the program, why it's beneficial? You know what I miss?
[00:25:04] Speaker A: Yeah. So I would say regardless of where someone is currently at right now, whether they've never used A.I. or they are like a never A.I. er, that maybe has like some negative, understandably connotations to A.I. just to show up to see kind of what's available. It might kind of open your eyes. And the things that I can show you on how to save resources as well when you're using these tools.
And if you are a super advanced user of these tools, because those folks are out there, I still think that you will get that, that particular person will get some good kind of gold nuggets out of this. So I will be presenting it in a way that really works for everybody, no matter where they're at.
[00:25:48] Speaker B: Awesome. All right, Mike, well, thanks so much for joining us today. Our audience, thank you for listening to today's installment of Monument Matters. And like I said, Mike Lannan, thanks for this really interesting content.
You know, you've piqued my curiosity for sure. I'm still not convinced, but I'm definitely convinced that I need to be a participant in in the series for sure.
[00:26:14] Speaker A: Indeed.
[00:26:15] Speaker B: I want to see what's going on because I'm sure that my bias is mostly that so the July and August issues of NB News have articles about the upcoming AI Boot Camp series in four Monument builders that would be members and non members.
Non members will have access, but at a different rate than members. So another benefit of membership in mbna.
I encourage you to read these issues and look at the Boot camp details on www.monogrellers.org and if you have a topic you'd like to have covered in a future podcast, you could please leave a comment. MBNA invites you to stay connected through Facebook and LinkedIn, or visit us at www.monimalbuilders.org for upcoming events and webinars for MB&A and Monument Matters. I'm Michael Johns. Thank you for taking time out of your day to listen in again. If you found this content worthwhile, please share the link with a friend for comments and feedback. We'd love to hear from you. Please drop a note to infoonumentbuilders.org again. Mike, thanks for taking time out of your day to thank you, Mike, enlighten our listeners and me.
[00:27:26] Speaker A: My pleasure.
[00:27:27] Speaker B: And all of you have a great rest of your day.
