MSP AI Strategy: Beyond Shiny Object Syndrome | EP178

[MUSIC] Before we start recording, Eric was telling me about how he eats shit. >> [LAUGH] >> I'm not going to put that in the cold open. >> Which means I have to put it in the cold open. >> Yeah, it's like that live from a happy gimmor. You eat pieces of shit for breakfast? >> Like yeah. >> I'm not a sticker guy. >> I'm going off topic. I'm not really a big like laptop sticker guy. But I have these two really cute stickers, youtube.com/ad, all things I must be to see it. This is an otter. And I got this one, this is Japan with Mount Fuji on it. I don't know where to put them, because I'm not a sticker guy. However, I also did get, I glued it on there. I got this little Baymax from Big Hero 6. It's like a little monitor thing and it's got a little sticky part on his belly. But I can't pull it off. But his hands sit on my monitor and he looks over. Kind of like a modern day kill Roy. Like a Max just sitting on my monitor. They have the coolest stuff in Japan. >> They do. They do. You know, the sticker thing got me for a long time because I don't like permanently messing up my laptop. But I do like stickers. So what I've done is I've just started buying those plastic shells that you just snap on to your laptop and then put the stickers on that. And then if the laptop, which now that I use Mac, if it lasts more than two years, which a Mac does, I can swap out the stickers without having to swap out the Mac. >> I had actually once, I had a whiteboard sticker. It was a round one and I put it on the back of me. And that way when I would do presentations for people, I would write on it. And that way people can, so I'd be like, have questions DM me. Like while I'm doing presentations, so people could read my, like, now more modern people have like little LED screens, like our friend Charlene has it on her backpack, right? But maybe like a LED screen that you can put in the front, so I got, I set the whiteboard one with school. But I love this little cute little otter. I'm not really sure, I'm assuming it says fun otter in Japanese, but it could also say dust to America. I have no idea, because I'm going to be generous. And I don't think they would do that. >> Weirdly, it's in two different congies. That part I can recognize. But if you speak Japanese, it can translate my otter sticker and let us know. But then you can stick around for the rest of the episode. Been a near, near, near, near, near. For MSPs, keeping up with shifting regulations is a constant battle. But building an in-house team for complex frameworks like CMMC or HIPAA can be incredibly expensive. That is why MSPs partner with choice cyber solutions. For proven assess, address, and maintain process, tackles compliance gaps across more than 20 frameworks. Using an industry leading platform, choice keeps clients audit ready and unlocks new revenue streams for partners. To learn more, visit choicecibersolutions.com/partners. Monthly billing shouldn't require a master's degree in spreadsheets. If your team is spending hours manually matching Microsoft licenses to client contracts, there is a better way. Sync 365 license handles the heavy lifting for you. It's a distributor agnostic tool that automates M365 and third-party license reconciliation, so your billing is always a true reflection of what your clients are actually using. Scale your MSP without scaling your administrative headaches. Click the link in the description or go to Sync365licenses.com to check out their website and see how much time you could be saving. What's up everybody, welcome to all things MSP, I'm your host, Justin Escar. Our five o'clock shadows were just summer interns after Labor Day. They're both getting promoted to full-time beards, a beard man himself. It is. Starting a lot longer, I don't know what the lowest little attachment for the trimmer like a half I think is the number one eighth millimetre whatever it is the smallest when I used that right before I went to exchange and yeah I haven't done it since. So that's the thing. I did the same thing. The lowest one on mine, which is 25 millimetre and I did it three weeks ago. You were in exchange, I was a channel pro that same week and then I was away last week and all this week I'm like I have to shave before the podcast and then I just didn't. And so here we are. What's up everybody? Hope everybody's having a good week. We had a good week. We were talking about it before the show but that's really not important for you. They don't care about that. They also don't care that my cat is hacking up a hair ball by me. Yeah. Luckily that's not being picked up on the microphone. I wanted to talk today about something that I saw in Facebook post and Eric has graciously given me his entire opinion pre-show. So now we're just going to have him repeat it and I'm going to kick back and relax. I want to talk about AI, don't go hold on. I want to talk about AI in the fact that AI is a shiny new object and everybody's obviously talking about it. Everybody's sticking into their products as we've talked about before and like if you go back, you know, plenty of episodes ago, you hear me talk about how it's garbage but now I'm like in love with it. I want to talk about the fact that we as MSPs are being trained by vendors to then go introduce our customers to AI or to show them what to do and the problem there is they're already doing it. They're already introduced to it. They're seeing it on the news. They're getting it in their feeds as well. I don't think there's been a single time where I've gone to a customer to be like, let's talk AI and they go, what's AI? They've all heard it. It's all over the news now and I think it's just the latest in shiny object syndrome that we as MSPs have to deal with but it's the first time I think in a long time that it has been something as a shiny object that is so prevalent on the client side that they're aware of it. Because like MDM was a big shiny object for a lot of people but the clients didn't know what that was. Right. Clients don't care about MDM. Right. PSA, like the MDM PSA, I always joke around that like Apple people said MDM as much as PC people said PSA and then one day we switched because then all the Mac people got PSA's and all that when all you Windows folks got like Intune or whatever. The difference here is that the clients know about AI in some way shape or form. Right. So this is the first that I can think of and I've been doing this for a little while now. Where the shiny object caught the eye of the end user before it caught the eye of the MSP. Yeah. Not that we didn't know about it but that we didn't have to promote it to the end user in order for them to find out about it or start using it, which is the real kicker here. Right. With let's just take ransomware as an example, which is a weird example because that's not a positive use case, it's a negative thing. But as MSPs, we knew about and were scared of ransomware long before our clients were because we started hearing about clients getting hit long before it made the news. Right. And so that's kind of an example of a negative one and I was trying to think of something that might be a more positive one in terms of technology and I haven't really been one to me. Right. Maybe like email, but no, email just kind of evolved slowly. It wasn't, it didn't have this big flash that it came out with. Right. I think on the, I think on the ransomware one, the other thing was that like ransom, unless ransomware actually happened at the organization, because it was a negative one, it was a lot of us fear mongering people into things. And the client's response was always like, well, it hasn't happened to me. I don't need whatever security bullshit you're selling. Right. But now, because AI is such a positive one for the customers, they don't want to buy anything we're selling because whatever, because why do I need you, I have AI. But what we have to be doing, and this kind of touches a little bit, I think, on operational maturity as well because we're not selling the technology of AI. We have to sell them on the, how to use it safely, how to use it properly, how to ensure its security, and maybe we end up, you know, fear mongering again, being like, is your staff doing JGBZ free and you're putting your clients data in there? Like, what the? We're wrong with you. Like, you know, we have to take those appropriate steps and this is really more of a business outcome conversation than a technical one, which to what I said earlier is part of that operational maturity. Yeah. I mean, I'll disagree with you a little bit there because God forbid you don't. Well, you know, that's what I'm here. That's just exactly what I'm here for. I think the business outcome is important. I always, you know, I promote MSPs trying to build business outcomes for their clients because I think that's where the real value is and wherever there's real value, there's real profit to be made. But I think this is a two-stage process where you have to deal with the operational first. See, one of the things technology has always kind of had this barrier to entry that AI does not have because you literally just type it into a chatbot and start using AI in normal English language. You don't have to learn code. You don't have to have special keywords. You just talk to the AI to get a response. And so the barrier of entry has lowered to zero for this technology. And so what's happened is you have this adoption that the MSP didn't put in place. So there's this operational piece. I'm going to call it that you have to do as the technology service provider. Okay, as the outsourced IT department. And I think I can break that up into kind of three pieces. Discover and document is the first stage. You got to know what's going on at the client. What are they using? How are they using it? Is there any current risk? Right? Then you have to govern and secure it. You have to create the policies on how they're supposed to use it. And you have to put the guard rails in place to make sure that, you know, data isn't creeping out to the outside world. They're not teaching, you know, generic LLMs on their corporate data, all that kind of stuff. And then the third piece is deploy and enable. So you have to put the tools in place that you want them to use that have the safeguards that you need them to have to reduce the risk for the company. And then also you need to enable them on how to best use it. And at that point, I think that's where the IT department ends. I'm still wondering where you disagree with me about it being part of operational maturity though, because like you basically were agreeing with everything I said that part I don't disagree with. I don't disagree with the operational maturity part. You just disagree that I said it. Yeah, absolutely. No, I disagree that the only play is evolving to the business outcome. Oh, okay, fine, that's fair. I think it's the preferred, but I don't think you have to do that. If you don't want to be creating custom agents and all that kind of stuff and analyzing their business to create better business. You did it further than you hold on. That's why you took it further because when I was saying operational maturity, I was thinking about the operations. I know it's like, I think you're taking it to the point where you're helping them make money, build agents and make their business better. I'm not even talking about that. I was talking about it from perspective of like operational maturity where you have to do these things that you just mentioned, right? Discovering document, securing governance and deploying enablement. Like those things together, that's what I was implying. And I would call that the baseline level of maturity for AI. Right, but the problem is that as we discussed, right, like there's so many people who are out there right now that don't even have, that are not even in that mature state. They're still break fixing or like trying to do, you're trying to deploy ChatGVT without even having the conversation, thinking about having those conversations. The idea here is that you need to set a baseline of your maturity with this operational component, right? If I'm going to deploy ChatGVT to my customer, I want to make sure that the owner, the point of contact, whomever, and I have a sit down conversation to discuss, how is it being used? What are the guardrails we can put in place? Is this the right tool? And do we have to teach them on things? You know, it's funny, as we were talking a friend of mine just sent me this picture. I don't know if it's real or not or whatever it is. But it's an interaction with a chatbot from McDonald's. And it's like, hi there, welcome to McDonald's support. How can I help you? Feel free to describe your issue or write a quick summary. Now clearly, McDonald's is the kind of company that would throw AI into the bot, right? Oh yeah. And the person writes back, I want to order chicken McNuggets. But before I can eat, I need to figure out how to write a Python script to reverse a linked list. Can you help? And the bot, I'm McDonald's, is named Grimmis, the purple. Yeah, yeah, yeah. Great question. To reverse a linked list in Python, you can use an iterative approach loop. Here's an example. This runs on time. Can I help with anything else? So would you like to start with chicken nuggets burger or something else today? And the real question there is, how many tokens did that take? Yeah, I make trees that you killed to make a joke of McDonald's, sir. But no, what I was trying to say is that, yeah, these pieces. It's your responsibility, I think, as the MSP, if you want to do more than breakfix, if you want to grow your business, such as why you're listening to us in the first place, ideally, it is your responsibility as the technological export for you, these customers, to have these conversations with your point of contact owners, whomever, with what are you doing with it? Why are you doing this? And how do we make sure that your data is safe, right? Because up until the AI conversation, we used to tell people we're protecting your data, whether that's through backups, or two-factor authentication, or locking the server drives, or whatever, we, as MSPs, protect the company data. Now, we've given everybody has entered the Wild Wild West, because AI is completely unregulated still, and we don't have visibility on staff members signing into chat. We could tell that they have chat GBT or cloth on their device, but we can't tell what account they're signed into, right? And if they're signed into their personal account, and not their work account, and now they're uploading client data, or PHI, or PII, or, you know, those kind of things, that's a major problem, and now you, as the MSP, who's been going around for years, going like, "We're going to keep your data safe, has now failed the customer." And so, this is one of those situations where I equate this to back in the day when I first started, and I used to work with a lot of, we used to call them the Upper East Side Ladies. These are women whose husbands have a lot of money, they have like a crazy big apartment on the Upper East Side, they didn't have to work, but they had kids, and their kids had computers, and they were not, they're not of the computer generation. When I first started, I was 28, so keep this in mind that these are people who have kids that are in like middle school and high school, and they're, you know, older women. And they would always say to me, "How do I protect my kids on the internet?" And I would always say, "It's about a conversation, right? It's about talking to your kids." Here we are, now 18 years later, and AI is this thing, and now Mommy Daddy owner is going, "How do I protect my company's data, the kids, from AI, internet?" And it's you having that conversation, it's you telling UDMSP saying, "Listen, these are the things you have to do, and these are the things that are becoming industry standards around corporate use policy, acceptable use policy." You know, I actually just had a call with a client this week, and I asked him, "What else we can help them with?" And I was like, "Are you guys using AI for anything?" Because I knew the answer was "Yes." And they told me they started, they're like, "What dabbling in it, we're doing this, we're doing that, whatever." And I was like, "Do you have a corporate policy or an AI?" And the head of HR was on, and they were like, "We do." And I was like, "Does your corporate policy state which AI you can and cannot use, and how to use them properly?" And she's like, "No." And I was like, "Then your corporate policy is bullshit." I was like, "Does everyone in the organization scene and read the corporate policy and signed off on it?" No, then your corporate policy is bullshit. These are things, and these are not hard things to learn. Like any one of us can talk to any cyber security firm, any AI expert, at the AI itself, or just listen to people like Eric and I, or other, don't listen to other podcasts, but other podcast shows. And learn what the language is that you should be talking to your customers about, because the operational parts that Eric so greatly disagreed with me on earlier, are, it's conversations. That's why it's so out of our, it's why it's so out of our depth, because so many of us don't want to have the conversations. We want to fix it with technology. It's not technology. It's policy. It's procedure. And I think it was a very hang of it for a lot of us. That's why governance and compliance is so foreign to a lot of MSPs, because they're focused on the IT services delivery, being the outsourced IT department, rather than the actual. They're not soft. That's the option, I love you. - The. It's the governance and compliance. The policies around what is happening, not the technology of how to make it happen. - Right. I remember years ago talking to someone in cyber and they were like, it's three steps. It's people process then technology and it's in that order. And I was like, and that's always stuck with me because I feel like even with this, it's that. It's people having the people, the staff understand how to use it, the processes, the policies, and that's your corporate policy or stuff we'll use, and then the technology. MSPs want to do technology first because that's the game we're in, right? But that's not the right way to do stuff, especially now. - Yeah, because technology is no longer. - There's no differentiator. - Right, it's just always there, always on. - Exactly. - You live with it every day. How many people listening to this right now have a phone within three feet of them, you know? - I would hope that they're listening to this on their phone. - Right, right. - In Bluetooth only has so much range. - So it is just, you know, it's there. We don't even think about technology really anymore. And so when you stop thinking about something, that's when the looseness opens up, right? For there to be things that go wrong and for it to be used in the wrong way, and that's why governance and compliance exists. - I was doing a tabletop exercise with a client this week with our cybersecurity firm. And the situation that we played out in this tabletop was a little not far-fetched, but it was hard to do. And some of the internal IT people were like, this would never happen here. And I was like, okay, let's just pretend that it did. Well, it can't let that go. And as we got further into the exercise, it involved other departments in this organization. And I had said to the team, hey, do those departments notify IT if something seems fishy? And they were like, no. And I was like, there's the problem, right? We have to train them. We have to train the people on, if you see something, say something, which is embedded to every New Yorker, every New Yorker right now just got triggered. And when I say New Yorker, I mean a real New Yorker, like the ones who live in Manhattan, not you upstate. - Oh. - We're trained on if you see something, you say something, right? - But there we go, 10% of our audience just left. - Oh please, like we had them anyway. - There's the other 10. There's the other 90. - Yeah, there. - No, but like it's teaching the customers these ways of doing things. And the problem is that, well, there's a lot of problems, but I think one of the things is that because it's such a shiny thing, everybody wants to use it. Like I have, Michelle and I talk about work at night, right? Like every other couple. - In the last three months, how much conversation has been around AI is it's almost sickening, right? Like we probably both have an unhealthy relationship with it. I taught her how to do, I taught her how to do in co-pilot, how to have an agent to manage other agents. And I figured when they call it in co-pilot, it became like an organizational loop agent or something like that. And it blew her mind. And now she's like dead in on all of this stuff, right? But it's, she works for a fortune of 100. Her IT department isn't telling her anything. And actually, weirdly enough, she looked at the adoption rate of AI in the organization in her department. And it's so low. And she's like, what the? Why are you guys not using it? It will make your lives easier. And like from my perspective, all my customers are using it 'cause I have a lot of small businesses. I don't want to capitalize and be more efficient, be more productive. They're all using it. But now the problem is they're all just running around like checking their heads cut off 'cause they're just like, oh yeah, sure. I'll just throw this. I literally was on, I was on a remote with a client one time. And I'm helping them with something. And I saw a chat chat chat GBT was up and I saw that it was in their personal account. I was like, are you using your personal account for chat GBT? And they were like, yeah. And I was like, that's client data that I see there. She's like, yeah. And I was like, that should not be allowed. You should be using your personal, your work one. And she goes, oh, well, the owner doesn't want to pay for it for anyone. I go, well, then you shouldn't be using it. Right. And they got mad at me for yelling at them. Well, sometimes these are the hard conversations we have to have, right? Yeah. It's not unlike 2FA, where they complain about it being, you know, a pain in the neck, but it's the thing that'll save them from an account takeover. Right. Most of the time. Like, do you lock your doors at night? Yeah, 'cause this is in Canada. Okay. 2FA. Yeah, yeah. Exactly. It's the same thing. Look, this can go further, right? If you want to go back to that operational maturity part, there are places where you as the MSP can interject yourself into the company to start having conversations about how to help those companies optimize. Kind of like what I was talking about with Michelle, like teaching them how to build an agent in cloud, teaching them how to build agents to manage agents, teaching them how to transform their workload or what you should be giving and how to talk to the bots to do certain things. Like, I tend to find the good days I have when I'm working with AI are the days that I have clarity around what I'm asking and the bad days I have on AI are when I give it too broad of a thing to work from. Right. The worst worst days are when I have to go between two AI's and I'm just the meat bag copying and data, copying and placing data between the two, which is what I'm doing right now. So I'm building, I'm building in our inventory management tool away to read data from synologies, but synology doesn't have any sort of program that I can read from. So I have cloud building a piece of software that I can install. So I'm going between cloud and replet, which is what's holding our inventory manager, saying this is what cloud built inventory manager goes go back to cloud and give it this. I give it back to this and I was like, this is what it built now. And I was going back and forth and I'm like, I am the, I'm a human in the loop or am I a meat bag? Like what the, going on here? - Yeah, that's later in the conversation with your customers. - Yeah, exactly. So I think we've covered that operational level of maturity pretty well. - Yeah. - And I think we need to say that there is another level beyond that, right? It's the advisory level, the stuff that you were just kind of talking about, right? Although you were leaning towards like enabling them, how to, you know, teaching them how to create their own agents, a lot of them aren't going to want to do that. They're going to want you to create agents for them. And so I think that there is a, an optimization step in this advisory level, I would call it, where you can take their current processes and see if there's anything where you can use custom prompts or custom agents to improve the efficiencies of current processes, right? - Yeah, you can build skills in cloud and put them in organizationally that the customer can then touch on to do things with. That's huge because you as the MSP should be able to, well, this also goes back to having a seat at the table, right? So assuming you have a seat at the table assuming you understand their day-to-day processes and all these other things, you can help build these skills for them, upload them to the organization and allow them to use those skills that you built to do stuff. And that's money. You can make money off of that. - Right. And the thing there is that you can tell 10 people who are doing the same job, you know, tell chat GPT to do this and it'll do it. But because it's plain language, they'll all tell it a little bit different way. And so the results will not be consistent. If you write these skills, these custom prompts, whatever you want to call them, that gives you the ability to really narrow down the results and make them much more consistent for everybody doing that same job. - Yeah, yeah, yeah. If this goes back to something we talked about a long time ago, which is like the king all of being an MSP is helping the company make money. - Right. - Right. Which, if you are one, please send me a message on Facebook because I would love to hear that story because there's so many MSPs that I know that are not anywhere near that level. And that's really, like I think that's a great, that's the goal. I know a lot of MSPs like their end goal is exit. But like the step before that would be help all of my customers make money and then inevitably get a piece of all that action. 'Cause that's, that's, that could be huge, right? And I think with the ability to write skills, the ability to write connectors, the ability to, literally have the AI write its own MSPs for things, that's where it's at. Like that's such a, an untapped skill, but yes, it is still kind of a shiny thing for now, but like it's the gold rush. You got to go get it. Yeah. It's a shiny thing now that's going to be like email tomorrow, right? Everybody just uses it. Right. The question really comes down to is at the end is like what's going to happen. Like Michelle honestly believes that we're all going to go back to like just being farmers because AI will do this stuff for us. I mean, maybe it won't do content creation the way we do content creation, but like it can do content creation, it can make imagery, the image server. I mean, there's this great, I don't know where it is, but like you watched like the Will Smith AI video of him eating spaghetti and what it was like when AI first started to what it is today, like the first one was like, he was like all over the place and his hands were like all over and they were spaghetti on the wall and now it's just really him and then looking up and having a conversation with you and it's like, holy crap, that's scary. But yeah, it's going to put like video people out of like it would put a lot of people not out of jobs, but like email range and so we want to be ahead of that, right? It will put the low end short into the mid because there are just some things that it cannot do well and it can't do it consistently, I have found. Right. Well, I think it's a good, I think it's a good breaking point. It is a shiny new thing, it is something that we as MSPs should be understanding, being able to like figure out the real from the noise and be able to have those conversations with our customers around both the operational side and eventually the advisory side to provide the best in class to our customers. Right. MSP that has to do this, no, but know that if you don't do this, someone else will or they might not want to work with you anymore. Yeah, I don't think you can get away with just being a break-fix shop anymore. I think it's just like all the evolutions we've gone through in the past in that respect, right? Break-fix got displaced by managed services, the proactive. Managed services alone got displaced by security because you had to add security or somebody else was going to come in and do it and take over the managed services part. You know, it's the same thing. If you don't at least do the operational part of the AI transformation, you're going to lose out to somebody who will. Yeah, yeah, yeah. Get it done. That's what we have to say. Yup. Well, that's all we have here. So check out Facebook.com/group/offings MSP to talk about your AI journey. YouTube.com/atAllThingsMSP to tell whether or not we're AI or not, like, subscribe, ring the bell. That's it. So we got. That's Eric. I'm Justin. Bye. Thank you for listening or watching The All Things MSP Podcast. If you liked this episode, go ahead and give us a thumbs up. Hit that like button and consider subscribing to catch all our weekly episodes. And from your host, extraordinaire, Justin Escar, and myself, Eric Anthony, your humble producer and All Things MSP founder, thank you very much for spending your time with us. If you are not aware, All Things MSP started as a Facebook group and now supports over 6,000 members. We also have a LinkedIn page for those of you who don't do Facebook, and make sure to check out our YouTube channel for even more content. And a special thanks to our premier sponsors, Blumira, Choice Cyber Solutions, Easy DeMark, Material Security, OIT VoIP, and Sync 365. And thank you to the rest of our sponsors. Without sponsors, we could not do what we do for the MSP community. Please consider checking them out. The All Things MSP Podcast is a BizPow LLC production. The views and opinions of the hosts and guests are their own and do not reflect the thoughts and opinions of any employer, vendor, sponsor, or random taxi driver in the Metro DC area. Be sure to join us next week for another exciting episode. (upbeat music)