Listen to "All Things MSP" on Your IT Podcasts!
Eric Anthony:
Hey everybody, Eric Anthony, co-host and producer of the All Things MSP podcast here. And I just wanted to remind you, yes, you that you can come on the show and talk about what's going on with your msp. If you need new ideas for your msp, tell us what you need. It's like therapy for your msp. So go to atsp link slash podcast and submit your request now on with the show.
Justin Esgar:
Number one, McDonald's chocolate chip cookies are far better than most other chocolate chip cookies. And I'm wondering why And if they deep fry them in beef tallow like they used to do to the french fries.
Eric Anthony:
Okay. And how does that relate to being an msp
Justin Esgar:
That the second thing I wanna talk about is you ever get like really invested in a TV show and then they have their series finale and at the end of it you just go, well, and usually I kind of just start the show without Eric knowing like going, what's up everybody? Welcome to the All things MSP podcast. I'm your host Justin Escar with OG host Mr. Eric Anthony, and we have a special guest today. Yay. Kermit Kermit Wave. Yay. Mr. Dave Sobel is here. Dave, what's up buddy? Oh, thanks for having me guys here to rock and roll. Just hanging out.
Eric Anthony:
Um, this is one of those where I was really excited to have Dave on the show. Number one, cuz I think he's probably one of the most prolific, uh, you know, influencers in this space and his content is not like ours, right? We talk about whatever we want to on this show. We're very casual and we just, we do this for fun, right Dave, you bring it to just a higher level. Like you elevate the entire, you know, I'm gonna call it the influencer part of the MSP channel and the IT channel in general, uh, just because of the things that you cover. And
Justin Esgar:
Well, it's kind of you to say cuz it's also a coded way of saying dude talks a lot like <laugh>. But oh, I was thinking more along the lines of he's, we're gonna bring you down to our level <laugh>. Uh, it's, it's all, I mean, look, there's, there's a certain degree of particularly cause we work together, Eric, like, like for, from a distillation perspective, like what I do now, I almost sort of took like my favorite five minutes of my old job, which was like the keynote bits where I got to do market analysis and I sort of always thought like, I'm kind of not bad at that. And I figured, well, let's see if I could lean into it and do a lot of it and see if it helped. And the second bit is, is I think a lot about the fact that MSPs and IT services companies can't Gartner, Canales, Forrester, like these really big consult, you know, the big research firms, but have that need and, and I literally was just covering some data about how low their r and d budgets are and I figured, I was like, well, you know, I think it mass there's some value here if I actually spend some time and do that for them.
Like, do that work and put it out as a digital product versus one-on-one consultant.
Eric Anthony:
Yeah. And and that's why I recommend your podcast and your YouTube channel so much is because you cover the data, you bring in the research that is going to allow MSPs to look six months, 12 months, 24 months down the road to see how the industry is evolving and keep skating to where the puck is going to be to use the Wayne Gretzky quote.
Justin Esgar:
No, that's not, I, I also smile and go like, look, I'm a capitalist at heart. I like making money and I want to keep as many MSPs alive as I possibly can. And that means like, let me, let me tell you where this is all going and give you enough warning that you can start doing stuff about it.
Eric Anthony:
Yep.
Justin Esgar:
Which I think is a great way to intro into what we wanna talk about today. Which knowing you Dave, uh, I wanna talk a little bit about AI and how MSPs are using it. And I actually, after what you had just said, I was thinking, well I could just cue you up with this and then go take a nap for 30 minutes and then come back to end the show <laugh>. But Liz, you're in my wheelhouse right now cuz because it's, it's been a long time since I've been jazzed about a piece of tech and also had the data tell me that it's a real thing, right? Like, you know, because I mean, look, we, you know, carcasses of bad ideas over the last few years. You know, remember big data, how about we talk about the metaverse? Like there's so many things that were like the next big thing and they had no data of of what's going on.
Chat G B T just celebrated six months in the marketplace and it was the fastest adopting consumer technology of all time. Like, yeah, I, that should tell you what you need to know that there's something here. I think that there's, there's so much to cover here and I know, you know, for a short podcast we're gonna do, let's try to, let's try to keep it towards, let's talk about chat. C B T I wanna just, I really like, uh, what's the, what's the mid journey? Um, that's always a fun one if you're doing art, but let's just talk about chat c bt specifically and, and how it works for MSPs. I'm just gonna start this off with, I see a lot of my friends using chat b t in my opinion, the wrong way, right? A lot of them are using it for social creation.
They're trying to get chat G B T to create content for their blogs as opposed to all of us paying for syndicated content or thing, or, or even writing our own content, right? So that's number one. We'll talk about that. And the second thing is how chat G B T should and should not be used as your tier zero tier one support within your organization. So let's talk about social from the first, for the first part cuz I, I, I know you and, and I know I, I can see it in your eyes for anyone who's, who's listening, I can see it deep into Dave's eyes here that as soon as I said chat G B T for social, his eyes roll to the back of his head. Right? Uh, well cuz it's, it's like, I, I kinda laugh and go really? Everybody, I mean, so, so I like to to sort of boil it down conceptually into where I think these things, I think chat GPT is a fantastic high school writer.
Like, it, it it's an infinite capacity high school writer. Yeah. And I do you really want just some high school kid writing all your social stuff? Cuz that's what you're, what you're saying. I use Chachi PDD every single day in the creation of my show. I do not mind telling people that I do that. Let me, let me give you some. But I would never have it replace my voice because it's not good at that. It's not necessarily great at advanced tone at cre at creativity, but it is an incredible partner. And thematically, I'm gonna bring this up a bunch of different, different times. AI is not gonna replace your job. Someone who knows AI will replace your job and if you keep applying that filter to the way you think of this stuff, you'll be better inclined to use it. Lemme tell you just briefly the way I use it.
I write my own stuff, very proud of that fact, right? But I write my own stuff, but then I, I use Chachi PT as a writing partner. I say, Hey, what did I miss in this story? What's the angle I didn't cover? What are points that I might not cons might not have considered? And I have a conversation about the story with it and then I go, Hey, help me with headline writing. What's a good headline? What's five good headlines? Tighten the headline. Uh, and I iterate with it and I work with it as a writing partner and it accelerates my ability to create. But if I just said, write me a social media blog post, it's going to go to all of the stuff that has previously happened and make a derivative version of that. And that sounds really boring, <laugh>, but you have a, you have a persona that's out there, right?
Like I was testing this the other day and I, and I asked it, who is Justin Escar? And it, and it figured out who I was because I have a public persona between ACEs and everything else I do and shows and whatever it is, it gathered all that information and I don't remember, what are we up to? Like it only reads up to 2019 or something like that. Um, 2021. Yeah, it's, it's, it's got a limit. So it it, it gathered all that information and figured out who I was. It, no, it should know who you are cuz you have a public persona. Have you ever tried just doing like in the style of a Dave Sobel article, write this? Like where is that limitation? And I mean, I, I understand what you're saying about being a writing partner, right? Cuz not everyone has a public persona.
Not everyone can pull this trick, but like, have you tried that? Like what happens when you do that? Well, actually to be fair, when I work with it, I'm not telling it right in the style of Dave Sobel or anything along those, I'm, I'm actually at times I will just tell it things like I need a concise summarization of, of this. I need, uh, how would, how would you write this if you were a journalist? How would you write this if you were an analyst? I ask it those questions and I get different versions and start comparing so that I can say like, okay, you know, what would the journalist voice sound like? What would the analyst sound like? What would the business owner sound like? You can get different ones and ask it in those contexts. And again, always thinking that this is something that I'm using and I'm reviewing, not I'm having it do for me.
It generally is doing editing or, or different or asking questions about content rather than creating content. Uh, I tried that one time. I was, I was trying to do a pep talk for my team about client engagement and talking about like doing the follow through or whatever it is. And I wrote this one paragraph one and a half paragraph thing, and I threw it into chat G B T and I said, uh, write this in the style of a business owner. And it started with, like, it started off with like, folks today we're gonna talk about this. And then I was like, rewrite it in the style of a TEDx talk. And it was like, ladies and gentlemen, today I'm gonna talk about this. And I was like really interested in the fact that it could change my tone in such a way that I can't do myself, right?
Like mm-hmm. <affirmative>, I've never given a TED talk, by the way, if you want me to do your TED talk, just send me an email. Um, but like it was able to like take my original tone and my third grade writing skills and make it into something that like looked and seemed, and maybe it was just because I don't comprehend it well enough, seemed way more polished than anything I could have put out there, right? And th and this is where I look and say like, that iteration is where I think it's really good, right? Where I think it's like I can take material and I can try different versions and I can get different voices and I can work faster with a writing partner to create stuff that is generally the role of the editor, right? Is is that you, they are someone helping you tighten up your prose and do a better job with it.
And that's where I think this excels where I think it's, it, it's also useful in brainstorming activities where I say like, Hey, I've got this idea. What are four other things that I might consider on here? Where, where, or I might say like, you know, I'm, I'm considering this possibility. What are the questions I might ask around that? And it, it comes and it's a great collaborative partner if you ever use it for, I would recommend trying it in brainstorming sessions, adding it to the mix of like, if you've got a group of people around a table as opposed to replacing them. Like it's, that's where it's, that becomes very powerful partner. Where I wear warn people off is the idea of, of thinking that you can singularly replace something with ai. It's like, yeah, that's dangerous territory everybody. That's where you lose where, where you lose the subtlety of what you're trying to intend and getting the messaging really right?
Yeah. So let's lead that, let's lead that right into, cuz I feel like a lot of people are using chat G B T and it's, it's being incor incorporated to a lot of PSAs now to yeah. Be the first responder for a lot of support tickets. So you're not, you're not allowed to go responder. What's the story there? Well, let's take shitty automated emails and make them worse. Like <laugh>, like, like seriously, I mean come on everybody. You are, you're literally in the support business, right? You, you, your value is about the ability to interact with customers. That is the number one thing that ha that retains value over time is your ability to help them. You're gonna automate that away. That's insane. But you can take frontline, uh, customer service people and you can take two month employees and make them like six month employees because that's where the data shows the acceleration is, is that you can give those frontline people tools to be better, faster and more engaged using those tools. But I would smile and go like, you're really gonna put one of these things in between you and your customer. That's a horrific idea. But why is that different than one of those current non-AI related chat bots that like pop up on your website when someone's like, help me with my Zoom. And it's like, here's four articles I found on our knowledge base about Zoom. Like is there anybody, like anybody like that? It's also horrible idea.
Eric Anthony:
Nobody likes
Justin Esgar:
That. Anybody like that experience either <laugh>, like look, particularly when I think about the typical IT service provider in this space, right? Let's break sort of break it down. 80 some percent of of these are people are are le doing less than 5 million in revenue. These are not massive scale organizations. Your competitive differentiation is white glove, better service, personalized, using humans. Like that's literally the bit Now I totally get everybody the make that efficient, make them as, as you know, best as possible. That's where we apply tools to that. But your value is your boot boutique ability to engage them and deliver high value information and recommendations. Do not lose sight of that. That's where you retain value over a long period of time. And that's why I smile and go like, really you're gonna outsource that? Find me humans that like dealing with those automated chat bots because they're not out there <laugh>. Right? What those businesses are doing is trying to achieve scale on a massive front that you are not trying to achieve in the typical IT service provider, right? Right.
Eric Anthony:
So we've, we've talked about how, you know, MSPs can use it, we've talked about how we shouldn't use it. I think the other important topic is how can MSPs help their clients? What's the opportunity for MSPs to help their clients engage with ai?
Justin Esgar:
Yeah. And this is where it gets like super exciting, right? So this is because again, go back to the core value. Everybody remember what the value is. I mean, and everyone always asks me the question like, ah, do you think IT service providers are gonna, are gonna survive? And I'm going, yeah, if they remember their core value, the core value is helping small mid-size organizations and customers with their technology needs, right? That's the core value right there. And their technology should drive their business, right? It should be something in the business that becomes a thing, right? We help them and my, I do a good better, best model of, of, uh, the way you analyze this, but the the best right is, is if you're driving revenue. So what we want to do is we want to help our customers drive revenue using ai. That's what we're trying to do.
And I look at this space and there's a, a theory that the guys over at the Techmeme ride home, uh, Chris and Brian are super smart guys and they've come up with this, i, this theory of AI models as, as similar to the way winemakers apply grapes, right? So they look and say like, different grapes will get different results in wine because you know, by, by the, the production and the taste and all of those kinds of things. And I love that model because I can extend that Dave thinking, thinking of this model and saying, okay, if that's true, a sommelier is a wine expert. That's someone that pairs a person to the right wine for their experience. That is the IT service provider is pairing the right I AI based on its model, based on its capabilities, understanding the underlying models, the technologies, and applying that to customers in the right way.
And last time I checked sommelier are incredibly well paid roles in restaurants. I love spaces that look like that. And so I look at it and say, that's your analogy. You want to think about this as your expertise as an IT services company is understanding the capabilities of the different models, then understanding how that applies when they get implemented into products and into solutions. And you help your customers achieve that, using those tools to do better and drive more revenue. I've now made it mandatory that my entire staff listened to this episode. <laugh>, I also thought about the idea, I also thought about the idea of the IT sommelier and I was in my head while you're talking, trying to come up with like an acronym for that that replaces an acronym we already have in the industry. Like I tsm A A I A I I T sm like something like I don't whatever, but but do your acronym.
I also push back and go, no, no acronyms, everybody. No acronyms. Stop, stop with the acronyms. We don't need more of them. We actually need to talk to customers in these basic terms. It's super important to explain to them in a way that they understand. And you can say, look, there's just, and you can use analogies like this, just like there is different grapes and different wines. You get different outcomes with different models. And what we do is we help pair you to the right solutions. I would also like to put out there, if we're gonna stop using acronyms, can we start putting vowels back into company names? Because <laugh>, I saw some clip on TikTok where some guy was talking about, he started a new company, it was some stock trading up and uh, the company name was giraffe and the people were roasting him and they were like, you got giraffe.com and he's like, it's J R F F and they were like, get outta that really does, and it's not giraffe please.
Right. And by the way, we're all gonna type in the correct, like the, the correct spelling and go the wrong place. Yeah. Put put in, put in numerous acronyms, put in things. Um, so okay, fine. So if we are going to train our clients on how to use ai, right? We now like, it's weird because like we have to follow some set of rules of we're not gonna use AI to generate, we're gonna use AI to edit, but we're gonna train our clients to use AI to work better with their existing businesses. And then we have to figure out, okay, well if they're a creation company for whatever, whatever they're creating, like how do we keep them from doing that? And like I feel like that's a perpetual, it's gonna be a perpetual cycle. It's just gonna keep cycling that like we need, I think we need Dave Sobol's, we don't have to do this right now.
You can write this down. You're gonna take this idea, run an amazing blog article, in fact have an AI write it, but we need Dave Sobel, uh, top the, the the 10 commandments of of ai, right? I wanna see a picture of you. I want mid journey to create a picture of you holding two tablets up like Moses and the big, like the whole thing. But so Justin, a smile and go, let's not reinvent the wheel here. One of the things I cover on my show all the time is kind of smooths that are happening on a legislation front. There are AI frameworks being put out by nist, by congress, by the defense department, by Microsoft. Like there are frameworks you can apply to your customers to guide them on ethical principles for the way that this, these technologies should be applied. Just cuz there aren't laws yet doesn't mean that people haven't been doing the work on this front.
You don't want Dave's version of it. You want NIST's version of it, you want the DODs version of it. You want the one that actually has been vetted and been, you know, that your tax dollars have gone to creating take that stuff and apply it to customers. That is literally what it was designed for. Yeah, but I can't go my friend, the people who work at nist, I can go my friend Dave <laugh>. Well I mean lastly check. They've got way better clout than I do. I'm working on it, but they've got way better clout than I do. I think, I think in the Mac world people would just go, I don't know who either of these people are. <laugh> it's fair. Well,
Eric Anthony:
But once the laws come into place, the laws are gonna be based on nist. You know, a lot of the D O D C M M C is based nist so it is important to adopt those frameworks, um, as your basis point because those are what are going to evolve into the regulations.
Justin Esgar:
Right? Exactly. And plus by the way, every, you know, like look and I'm tend to be pro-regulation for a lot of reasons, but putting that all aside, don't do that work. Somebody else has already done the work <laugh>. Yeah. Like, like go take the guidance that's already out there and start applying it. That's what it is for literally what it is for. I like playing with, I did a presentation and I, and I, I precursor it. I used a, there's a PowerPoint, well, a slide deck kind of, uh, one called tome TME app, I think it is. Mm-hmm. And uh, I, I said, write me a 10 slide PowerPoint presentation on why nonprofits need to beef up their cybersecurity. And I just let it run, I let it do its thing. And I started the presentation with, Hey everyone, I purposely let the AI make this entire presentation.
Uh, not because I don't like you, but because I wanted to see what would happen. And truth be told, I ran out of time before I could make this presentation. And they, it was weirdly enough they loved it because they got to see now cybersecurity, which, you know, for the most part it was right. And then also, uh, what AI could do and how bad AI failed. Cuz there were some, right, there were some like images where like, uh, it was supposed to be like a person with a shield and a phone and then it would all like melt together like a creepy horror movie. Like there, there's a time and a place really, I think is what this is coming down to. And even with that, the reality is you're right, I should have edited, I should have edited that, uh, that presentation because there were a hell of mistakes <laugh> in it on top of everything else, right?
I mean, there was a, there was a law case recently. The New York Times did a whole article about a lawyer who literally had his brief written by chat g pt, uh, and just submitted it, like didn't check, and it was full of like references to laws and cases that don't exist. Uh, it, it's, that's not how it, there was a medical one I saw on TikTok. Yes, I'm old and I'm on TikTok. Um, there was one where, uh, it gave chat G B t certain symptoms and it figured out that there was some, the, the thing that was wrong with this person was this very rare disease. Yeah. And chat G B T was right. And so then they pressed on it and they said, what were your references to figure this out? And it spit out a bunch of references, but all the references were fake <laugh>, right?
Well, they, but what's interesting is there's also, they've been running some tests, some researchers have run some tests, and I literally just covered this on the show where they were running questions from consumers out against chat GPC versus doctors, and they're actually able to get better diagnosis and more empathetic responses out of the bots than out of doctors. Now, I will freely say that where I think this gets interesting is combining the two, right? Because one of the biggest issues, particularly if you think about both from an IT perspective in terms of our own techs, but if you use doctors as a proxy for that, it is really hard to keep up with all of the information that's hitting you in a fire hose. But if you're training models on all of that and it's able to provide information much faster, it will take you down new ways of thinking and provide information much, much faster.
And they are iterating and getting better and better on the hallucination problems. Open AI just announced the new changes to their way that they're testing and they're seeing new improvements in the way the models work. We're going to get there in terms of increasing reliability. But everybody always talk assumes that we're looking for perfect reliability. We're not, we're looking for it to keep pace with humans who also make mistakes and then improve that. And I like this idea of bringing the two together to make our doctors much more informed, make our technicians much more informed. It's, see, we're seeing it in law cases. One of the areas that it's great at is doing summarization. Like AI is really, really good at simplifying these things, taking inputs and, and summarizing them. Again, you can move much, much faster with the two together. I'm really thinking about WebMD at this point because like you, you could put in whatever you want into WebMD and it told you you had cancer.
Like it did not matter what Right. Was the end state. You always end up with cancer <laugh>, you always end up with cancer and the amount of people who have gone to doctors with their web, with their printed WebMD results because, you know, they've broke their finger. I'm fearful for doctors now because people are gonna go into chat G B T and try to figure out what's wrong with them and bring that into doctors as well. I, this is not a show about doctors. We'll, we'll spin up the all things doctor podcast another day, but, um, well, but, but again, let's, you can for every look again, that's all humans using it the wrong way, right? Right. It's humans saying, I I assume it's gonna gimme the right information where I'm not an expert to do the analysis. I'm gonna give you a counter example, uh, again, one that I just pulled from the show recently where researchers were working on new antibiotics and what they did was they, they put their information into to a generative AI and did analysis on it, and they asked it to give likely candidates for testing for improvements to the antibiotic.
And the, and, and the AI spit out a number and they found one, they found they, they used that information to narrow down the area that they were testing and they used that to get a much more effective antibiotic much faster. And that's where I look and go, that's combining expertise with data analysis and moving faster. Your example is the one where yeah, of course we put non expertise who's unable to understand the information and that's what happens. Yeah. My example is like my next door neighbor, <laugh> throwing stuff into what her kid found, her kid showed her Chachi bt and she decided to throw a bunch of stuff in there and then think and takes it all fact, which, you know, it never really, because it's the old adage, right? Garbage in, garbage out, garbage out, right? Yep.
So this is this, this is why I'm excited about the space, right? Because it is complicated. We like complicated problem. Anybody in our space, we should like complicated problems, right? It's the ability to pair technology with expertise and people to deliver better results. And there are good ways and bad ways not correct. There's not one gonna be one solution, which is the final reason why I like the space when there are lots of different ways to solve the problem. That's good for us as solution providers because there's lots of different ways to make that work for customers. And all of that equals high value services.
Eric Anthony:
And high value services is what drives profitability.
Justin Esgar:
Exactly. <laugh>. Exactly. It's why I got excited about this because again, you see that, you know, adoption shows us it's happening more and more. Surveys is showing us that users, people are actually using it at, at their jobs. There's an element of they're not willing to tell their bosses about it, which makes it even more complicated and interesting. And the adoption numbers are so wide that it would say, okay, there are actual real cases where people are using this and being effective with it. Love spaces that look like that. Yeah.
Eric Anthony:
Well, as expected, this topic has been something that we could probably talk about for a very long time, but people typically like podcasts that are less than 30 minutes and we've already exceeded that. So, um, Dave, any, any last words on ai,
Justin Esgar:
You know, keep leaning into this. This is, and there's no one answer love spaces like that, but that means you're gonna have to keep learning and iterating. You're gonna have to lean into this. It is not gonna necessarily be easy. That's good news for your profitability. And real quick, I was gonna say, there is one thing you should lean into and that's listening to Dave SOL's podcast and shows, uh, I'm really good at those. Dave, where can people find you online? I'm easy to find. You can find all the shows at Business of Tech. It's a big blue button. You can catch it on all your favorite podcast platforms or on YouTube as video show comes out every single day in the news form. And we do long form interviews on the weekend, about 15, 20 minutes with experts in these various topics to help you learn a little bit more.
Awesome. Thank you so much, man. Uh, and you can find all of us online also. We're on all the podcast apps Dave stole online, all the podcast apps plus YouTube, uh, youtube.com/at all things msp or find our big old Facebook group at facebook.com/group groups slash all things MSP and join the group. Find out when this show is, uh, dropping. Join the, uh, live streams. There's always new content happening in the group. Leave a like, leave a sub subscribe, hit the bell, do all those things. Leave us a one star review. I don't care. We just, we're really egotistical and I promise we'll do better next time. That's it for me and Eric. Bye.