Dave Sobel | The AI Gold Rush Needs a Reality Check

[00:00:10] Hey everybody, welcome to the latest installment of MSP Business School. We're going to jump right in here because all of you already know my guest. From MSP Radio, we have Dave Sobel here today. And we're both hot on the heels of just coming back from the GTIA event, Channel Con, that was held out in San Diego. And Dave, what was the biggest topic you heard out there? Oh, you know, people are still talking about cybersome. I mean, I have to do that. Thank you for the flattering. I never assume anybody knows who I am.

[00:00:39] I'm like, you know, spend time on this. I mean, obviously everyone's talking about AI, right? And there was a lot of buzz. Many of my conversations, the interviews that I did were all focused around that. There's so much here in this space. And I implore MSPs to be really thoughtful on this because it is both a hype cycle with too much hype. And then there's real concerns underneath that need to be thought through. And I want to make sure that, you know, I keep warning MSPs, like just be thoughtful here.

[00:01:07] This is a measure twice, cut once situation. And, you know, I will say this, right? So I was out at CCF back in March and then we come out here for Channel Con and AI monopolized both conversations. But you are starting to see some of the evolution, even six months time, where the conversations are getting a little bit more mature. But there's still this uncertainty around it all. And then there's still this almost wariness.

[00:01:34] And I think that's something, as we were talking a little bit before getting into, we need to discuss and look at it as an MSP community. Because it's here, it's real, and it's going to stay. So, yes. But what I'll smile and go is the analyst in me goes, and we don't know how fast, and we don't know how bumpy, right? And we don't know how the conversion is going to go. And by the way, there's a ton of instability in the market here. So, like, while I could say, like, it's here, it's like, well, it's going to ebb and flow a little bit.

[00:02:02] You know, some of the analogies to the early internet is probably reasonable, particularly if you're thinking about it from a market's perspective. Like, I expect it to be kind of rough for a little while. And, you know, you're going to take two steps forward, maybe three steps back, then another three. Like, you're going to bounce around a lot. So, I don't expect this to be completely smooth. There's a lot of analogies also, not only to the internet, but to the early cloud conversion piece, right? Like when we were talking about the cloud stuff.

[00:02:29] And, you know, remember that you don't have to be on the bleeding edge of this to do well with it. You know, most of your customers are on the journey with you. You know, I like these conversions where the customer recognizes it too. Like, the customers heard the word cloud and they understood it, right? Customers understood mobile and they understood it. Customers are not necessarily understanding AI, but at least they're aware of it.

[00:02:57] And they're entering on the, they are on the journey with you. And likely in some cases they might be outpacing you if they are experimenting and trying things on their own. And similar to what you're saying there, I agree with you. You know, it kind of follows the path of internet and cloud. But there were some differentiators there that make it a different journey this time. Internet, it was access ability, right? That really slowed it down. With cloud, it was product lifecycle.

[00:03:22] Most people weren't moving to the cloud until machines aged out and it was logically time to move or their applications finally got there. But AI definitely seems to be much more accessible at the start. But there's a lot of question marks on how to use it. And that's where the MSP community has real opportunity. Oh, sure. I mean, absolutely. And by the way, both opportunity and risk. Of course. And that's where being the guidepost for your customers is going to be super important.

[00:03:51] Like, again, people like to think about all this stuff from the rush perspective. Oh, I got to move. Got to move. Got to move. Let me point out that Apple's success has always been by moving second. Right? So don't necessarily think that you have to be on the bleeding edge to get this all right. Just understand you need to be moving in the appropriate direction. And then deciding where the risk points are for you are going to be key.

[00:04:17] You know, the thing that I think is really important for MSPs to start understanding is the fact that, like, hey, there is a difference in the way AI technologies are consumed from a business perspective. Right? The fact that they are pure consumption and that the economics do not look the same as software. Right? The nice thing we all love about software is that essentially it's near free. Right?

[00:04:44] Like, it has always been, like, it is so close to free as to be infinite. And we've never really worried about the fact that once you start increasing demand, you will have problems. We get a whole discussion about cloud computing and what the hyperscalers do and all that kind of stuff. Just smile and go, they know how to make plenty of money, everybody. They can add money. But on AI, the more you use, it does not scale down to commodity very quickly.

[00:05:11] In fact, the more you use might actually drive to much higher bills. And that level of disconnect in the way consumption has been is something that's really different. It's not the straight economics of software. Right. I mean, you know what your software bill, by and large, is going to be coming in each month, right? Unless you do have some sort of usage component. But even there, there's usually some guardrails that keep it from getting too out of whack.

[00:05:36] Here, we are seeing some things that lend us to believe as more and more of the token costs are really being transferred to the end user. There can be some scary numbers thrown out at people who aren't expecting or at least looking at evaluating, are we doing what we're doing actually worth the spend? Right. And it's mistakes at machine speed, right? That's the – and I have my own example of that. Like, so MSP Radio, we've done a whole bunch of automation and AI on the back end to do a lot of the enhancements.

[00:06:06] I hit this last month. We had an agent go little rogue, and it got caught in a loop doing analysis on something. Yeah, it had been taught the right thing to do, but then it kept testing something. It burned through our monthly budget for that in an overnight. Now, I put in place all of the proper catches so that it did not run – it was not allowed to run to infinity. But that was a mistake that cost money last month, right?

[00:06:33] And so you have to understand, like, hey, mistakes at machine speed can be incredibly costly when they are metered at this level. And, you know, that's something that's a bit uncertain, I think, for everybody at this stage because it's not – there's not a true understanding of what's going to be built, how it's going to operate, what those rules underneath are going to be, obviously, driving these automations and other outputs that come into it.

[00:06:59] So that mystery, because there's not enough stories like yours yet on the street, you know, is really the challenge that we see the community facing right now. Yeah, and 100%. And by the way, you know, for MSPs, I'll always like to – let's put ourselves – you can see this confusion in the fact that all of the vendors you're working with are trying different things. You know everybody's got it figured out when they all come back with the same business model, right? But, like, right now, you can see every little – every vendor is trying different things. So we'll use Microsoft, for example, right?

[00:07:29] So they've added a consumption component to M365 in the E7 licensing, right? So everyone's going, oh, it's not me. That's them experimenting, everybody. They put it in the enterprise customers with the most consumption where E7 is now not just a per-user price, but it also includes a usage component of Copilot that they bill through to their customers.

[00:07:51] So, yeah, sure, you're not seeing that right now, but if they start figuring out that the economics makes sense of that up at that level, I am confident that they're going to roll it out across the entire mechanism, right? Because they are optimizing for what makes the most financial sense. You can see this in MSP vendors, right? Like, I can see the difference between, say, an Atera and a ConnectWise. Atera has done a ton of automation investment. What they're doing is they've told MSPs, we're absorbing that, right?

[00:08:21] You're going to pay a fixed fee, and we are going to absorb the token economics of that, okay? That is a strategy. We're going to see how it plays out. Additionally, you'll look at somebody like ConnectWise, right, who has rolled out their predictive AI platform. They're billing by the workflow and then trying to tie that to the outcomes. That's a consumption-based billing, right? And their argument, that I've had many on the show, and the argument is that they're going to manage that.

[00:08:48] And they believe that they can drive the cost down enough through their efficiencies on token economics to make sure that those workflows are significantly cheaper than people. Cool, right? Sounds awesome. Sounds awesome. We'll see how these all play out, right? Because, again, I'm not endorsing any particular strategy. I'm just telling MSPs, hey, we're all figuring this consumption component out. You need to be aware of it.

[00:09:17] Right, and I think everybody's trying to figure out, as vendors, and I'll just put myself into that conversation here, we're trying to figure out where do the AI calls come in versus pure play and integrate automation and tooling that we can manage and obviously be able to present along with that mixture of AI enhancement, if you will, in that conversation. And then what are those costs, right? And once we figure those kind of things out, I think it's going to change some of the models for everybody out there across all that. Oh, sure. The whole thing.

[00:09:44] And everyone is – I mean, if we're getting into the weeds on this, which is way fun for engineer types, right? This is where you also figure out model optimization and which model is right for the right job and the cost efficiencies. But that whole component, MSPs who are listening recognize that is a whole skill set you probably have not had to think about.

[00:10:06] Depending on your size and engagement, the vendor may absorb that for you, or that will become the value that you bring to the table, or the value is you helping your clients select the right tools that manage that risk. But you need to understand the economics behind it. So let's maybe stay there for a minute, Dave, because one of the challenges I'm still seeing from the MSP community I interact with is really the where to start. You know, got a customer.

[00:10:35] Obviously, there's an AI usage of some sort in almost every business at this stage. But these are folks that haven't delivered their first AI project publicly. They're still building out some workflows internally. And there's a lot of mystery out there. Do you have any thoughts about how to kind of enter into that conversation? Well, the first bit is you need to have the conversation, right? So I have to smile. So my research at the business of tech, we've looked at the MSP market.

[00:11:02] 68.5% of MSPs do not even put AI on their website. Okay. The test we did was are the letters AI with spaces beside it, right? So we're not catching the word said or something along those lines. Are the letters AI on your website in the context of some business offering or discussion or information?

[00:11:29] 68.5% of MSPs don't even have it on their website. And the reason is, is like, hey, if your customer or potential customer has questions and they go look to talk to you and your website doesn't even discuss it, they may think you're not involved or you don't know anything or you haven't done anything. So let's just start with the, hey, you need to clean it open. Out of your 68% or whichever number you just said, the bigger problem is 50% of them haven't updated their site at all in five years. And that's a whole other different horse of a different color.

[00:11:59] But yeah, you know, if you don't have it out there, you can't expect your customer to know you're going to be a part of it. Right, exactly. You have to make sure that you're engaging your customers proactively to have the conversation. Like literally just make sure you are talking to them. You know, and there are a bunch of questions. Like we're, and I'll say we've just launched the V5 of the 68 point checklist at the Small Biz Thoughts.

[00:12:27] And we've included questions specifically around AI governance to ask your customers. Things like what, you know, what products have you, are you using? What are your employees using themselves? Do you have rules and guidance for your employees around AI usage? Literally just have you thought about it. Have you thought about it? And for your customer, have they talked to their employees about it at all? Exactly. Start literally there. It's just the start.

[00:12:57] Like we give you the basic starting questions, right? Like, so that's the idea is you need to be engaging to have those conversations to start the conversation. And I'm going to say, I say this all the time. We need to make sure, particularly as an MSP, you need to make sure you are not positioning yourself as the department of no. Like, your goal in a compliance conversation is not to go in there and start saying, talking about the dangers and how bad everything is going to be and how you shouldn't use these things.

[00:13:26] Or go in there from the element of saying like, hey, you've got a shadow AI problem. We have to wipe out the AI. Like, you're positioning there as the department of no. Nobody likes those guys, right? And most of the shadow AI problem today is, hey, go have a conversation with the user and find out what he's using it for. And do they have a pro edition? And if not, start there. That's your simple first step. And understand what you're doing with it, right?

[00:13:54] Because there's a philosophy of the use of the tools where some of the big benefits that we're seeing initially are on individual productivity, right? Where an individual is able to use some tools and be more productive, right? And you need to examine that and understand what the actual risks are and what the risks are for that business itself, right?

[00:14:19] Because the difference between a small retail organization and their risk level versus a healthcare organization are very different. And you don't want to come in there thinking, oh, we have to put together all this compliance for a small business that may actually see a lot of benefit from some of the personal tools because they don't have the same risk profile, right? And you want to make sure you're matching. We were laughing beforehand. Like, Claude has a flip switch, everybody.

[00:14:48] Like, it's the don't share your data. And for some group of customers, that may be all they need. It's a deliberate conversation you need to have. Yeah, and we've talked about that this is, you know, where an opportunity does open up for the MSP. You know, we've heard as QBR guys, let's just start there, that there's this pigeonhole mentality. We can only get to the middle manager and we can't get beyond this.

[00:15:14] Well, this is the greatest place to get a new set of conversations started because it's, hey, and you've said it already, where is your line of business using this? The department heads have the visibility into the end users that are really having the mundane tasks assigned to them daily. Let's start there. Let's build a pipeline, right? So there's this huge opportunity to deepen your relationship while having this conversation as well.

[00:15:37] Well, and, you know, don't, again, if you're putting on your philosophy of I'm trying to help these customers, these employees have figured out something that they think works. Well, let's find out. Let's measure, let's get them enabled to the point where we can measure the work, see if it's actually like making a difference, and then what are the downsides of doing it that way, and actually go through a very deliberate process. By the way, for everybody that's listening, this is what we mean when we talk about, you know, sell outcomes, right?

[00:16:07] It's what we mean is, is like go through, understand what the work process is, and what the result on the other side is, and measure that so that we can say if we make changes, are we making that process better or worse? And by doing so, can we iterate in a way that improves the outcomes?

[00:16:28] Customers love this stuff, right, where you're actually coming in and say, what we're doing is to work with you to make sure we're using the right technology in the right ways, in a managed, correct way to get you better outcomes because we want to drive your business. And then these projects become easy to implement conclusions because if they can see the outcome and they see the opportunity and they see either increased productivity, efficiency, or some sort of savings, things just happen, right?

[00:16:57] And you're not selling very hard to do so. No. And by the way, I am a big believer right now that old school consultancy is going to be incredibly valuable for a period of time. You know, I get it. We're all addicted to MRR. We love the high of getting a contract and coming every single month. I get it, everybody. Like, but I want you to know that when things are not defined, that's where good advice comes in, right? And that is a billable activity.

[00:17:26] And you can go forth and you don't have to lean so hard into making everything a pure packaged MRR easy. It's okay to charge for expertise. This is a good time to do so. You need to gather the expertise to do it. And then you can go around and help your customers out. And I say amen to that because that has been one of the biggest challenges I think a lot of MSPs have had is, hey, we don't charge for the projects. Well, I've never understood that to begin with, David.

[00:17:53] That's just me because I'm old enough, right, that we sold consulting to get a project. And then we sold the project and then hopefully we got MRR as the MSP scene came to be. But you're going to spend, if you really want to do an effective upfront AI assessment of a company, you do have to spend some real time with your customer. And as I said before, with those people that are in the seats doing the work to understand where optimization opportunity lives.

[00:18:22] And you can't do that for free. Right. And however you want to put that together, I am confident that there are smart MSPs that have figured out a way to roll that into a monthly. Like, sure, I'll believe you if you know your numbers and you can show me how you're cost managing that. Every customer is profitable. If not, know that not everything has to go into your bundle. Right. That is the base that keeps your business stable.

[00:18:50] And you do additional things on top of that. And by the way, like I'm going to leverage, service leadership data shows that the best in class have a healthy mix, not just of managed services, but also of professional services. Yeah. So we're getting near the end of our time, Dave. And I just want to make sure we cover off on everything you'd like to share around this topic. I will also highlight, listener, that the link to Dave's information in his new guide will be in your show notes.

[00:19:17] But before we go there, I want to make sure that we've covered off on everything you want to share today. Well, you know, Brian, we're always up to cool stuff, right? So I will tell you, I've spent a lot of time thinking about this particular consumption billing piece. You know, the other bit is really, I want our listeners to really be thinking a lot about the fact that this is going to be a period of time of some upheaval. Right. And I don't want you to rush. Right. I think the real bit is being deliberate and thoughtful and test and iterate.

[00:19:47] Right. This is a good time to start experimenting. It is a good time to partner with your own customers. Right. They have questions, too. The one thing that we haven't talked about as much is I really do think a lot of customers are on the journey with us. Right. This is in many of them are exploring AI in the same way that we are. They have access to many of the same tools, particularly in the small segment of the SMB market. You know, that is a ripe space for them to be testing things out.

[00:20:14] I keep hearing everybody talking about like, oh, shadow AI is like shadow IT even worse. I mean, OK, that's a pessimistic view of it. I will look around the other side and say that is an opportunity to work with our customers and say we're going to help get together, work through what makes sense and do this in a logical way. You know, there is the incentive to drive this faster that's being pushed on us by others in the market. You know, I'll smile and go.

[00:20:44] Remember, everybody that's telling you to move fast is highly motivated for you to buy more tokens. Yeah. Right. Right. So just remember that. Right. From the they are highly motivated to do that. That doesn't mean that isn't necessarily an interesting area to work on. But know that you can move on the timeline that makes sense for you and your customers. I do think not moving isn't an option. I want to say that baseline is I don't think you can ignore this.

[00:21:13] But I want you to lean into the fact that you're going to be deliberate at a good pace that works for you and your customers and not fall into the speed hype. And I think this, you know, one of the greatest opportunities that comes from this is it's not enterprise architecture either. So you don't have to do a clean sweep. Start with something small. Get your feet wet. Something with low risk in your portfolio of projects. Understand how that works. Move to the next one. Right.

[00:21:39] This is a journey, as you've said a couple of times, take it in those foundational steps and you can come out with some great outcomes on the other side. And if it takes 18 months, the market will be much clearer, I'm sure, at that point. Right. Exactly. Today anyway. And remember, the goal here for both from my perspective for you and your clients is profitable. I keep saying my theme of this year has always been profitable is enough. Right. And so, like, if that's what we're achieving, that's a win. Right. And we do that on the pace that makes sense for us and for our customers.

[00:22:08] And if you're focused on that quarter after quarter, you're going to do just fine. Well, Dave, I really appreciate you joining me today. You know, this is it's always fascinating to get an analyst viewpoint of what's going on as you're talking to many of the people that we get the advertising hype for as vendors and MSP. So I really appreciate you coming on and joining us for the listener. Dave's information will, of course, be in the show notes, whether you're watching this up on YouTube or anywhere where you get your podcast.

[00:22:37] We'll have a link to his latest version five of the 68 point guide as well. And if you want to meet up with Dave, just reach out to him. He'll be happy to get back with you. Dave, any last final moment or final statements? I'm always you to find. Tuesday's way to do is if you're interested in the checklist, free MSP checklist dot com is the link that redirects you right there. It is entirely free. Just sign up and sign up and give us your email address and then you get a free guide.

[00:23:04] Additionally, if you're interested in the also free analysis that we do, it's always on the business of tech that business of dot tech. There's a big blue subscribe button to get it on whatever podcast platform or YouTube you prefer. And we've got email versions if you prefer to get it in your inbox. Awesome. Well, Dave, thank you so much for joining me today. Always appreciated. And we'll see you out on the road. Thanks for having me, Brian. Thanks for having me.