MSP Dumpster Fire Season 1 Episode 9 - Microsoft Missteps: AI, Privacy, and the Race to Market

Welcome to the MSP Dumpster Fire series. This is season 1, the impact AI will have on your MSP. Welcome to the MSP Dumpster Fire podcast. This is season 1, the impact AI will have on your MSP. My name is Robert Chiaffi, CTO and co-founder of Progressive Computing Inc., an MSP based in Yonkers, New York. I am joined once again today by my friend, colleague, and co-host Alex Farling. And today we have a super special bonus episode we're calling Microsoft Missteps, and also a super special guest who needs no introduction, Matt Lee of Pax8. Welcome, everyone. Alex, can you please start us off on how and why we decided to come back once again for this special episode?

Yeah, yeah, it was, uh, who thought we'd be doing bonus episodes when we started this thing, right? But, uh, super excited about this one. As I was kind of talking to Matt at an event, and I can't even remember which one because I run into this fella everywhere, um, but as I ran into Matt at an event, we were talking about the podcast and, uh, he said, you know, have you covered what Microsoft is doing with their Recall product. And of course, this was a month ago. And so things have changed 9 times with Recall and Microsoft and AI in the last month. And by the time we publish this, it will probably all change again 'cause I think it'll be 2 or 3 weeks before this episode actually makes its way out. So things move fast, things go crazy. For those of you on the video episode, I run into Matt so often that I literally have his beard on my shelf. That is a joke. But you know, this is a thing that is absolutely insane. It's absolutely crazy. Microsoft basically came to the world and said, we're going to do this thing called Recall. We're going to record everything that happens on your computer and we're going to save it. What could possibly go wrong?

I, you know, thanks for having me on the show, y'all. I think this is a poignant topic. I think it goes like dog years when you think about how quickly AI advances in the world around us right now. Dumpster fires burn quickly. Not faster. Dumpsters, fires burn quickly, smelly and smoky, right? That's the way it works. So they're warm and warm when you touch the dumpster. It's really, really challenging. So, you know, this thought of, for me at least thematically, gentlemen, is that it's a race, right? There's like, put it in, do it now. We're going to do this thing that snags everybody's screen, makes amazing images and information about it that we can look at and search forever. What about my doctor's screen? What about that health image? What about those things that are sensitive? What about, right? And the challenge is that for me, it wasn't that they included it. Now, we talk about secure by default, and I'll go off on a rant for a second. They didn't by default turn on MFA until a very short time ago, right? And they didn't by default, but by God, they're going to turn on Recall by default was the initial statement, right?

They were by default.

They were on, you had to opt out.

Yeah.

So not secure by default, AI by default, right? Because that's the rave. That's what sells. That's what people want. That's what they're buying. But you saw how quickly that snapped back, right? Just since you and I first talked about it, even when we premised this show, it was by just the will and sheer public, oh my God, that they said, oh, you know what?

Maybe buy one by default. Well, security experts, rightly so, banded together and basically, you know, screamed from the rooftops that this was a bad thing. And Microsoft kind of backed down from it. It was actually funny. I was reading an article as we were prepping for this where an Apple executive was asking— Apple senior vice president of marketing was asked, you know, has Microsoft's blunder of the rollout of their AI product damaged or caused— made it harder for you as Apple to roll out your own? Like, is this going to make your life harder? And his response was pretty funny. He said, wait, do you mean that the failings of Microsoft competitor might be a problem for me. I think we're good, which is pretty entertaining. But Microsoft just continued to step in it on this one.

Yeah, yeah, yeah. You know, and you see it, and the joking aside aspect is, at least from my perspective, I don't think anybody could make a misstep too large right now. There's so much fervor, so much hype, so much, uh, just belief that even with that misstep People are still buying it. People are still doing it. People are still moving forward. People are still all in 100%. I mean, I get to talk to a new AI-based vendor probably 6 times a week, right? At least reaching out or in some touch or some form or fashion. All have some new application of it. All have some new thing that they're bringing to market. I think, though, this was such a misstep, and it shows this just like general endemic abandon of giving a shit about privacy. And security. That is the way I would describe our state of the union globally today, Alex, in my opinion.

And well, Rob, I think we talked about that in a couple of these episodes, right? I mean, if you think back, this is not the first time we've talked about AI being the forethought and everything else being the afterthought.

Yeah. And, you know, one of the things that I can't help but tie together is this notion of, you know, this rapid application development or, you know, minimal viable product. Fail forward, fail fast, you know, all of these mentalities of, you know, I recently did a LinkedIn post around there's something better in between good enough and perfection. And I try to shoot more closer for perfection. I had some pushback, right, from some people who are like, no, you'll never make it to the market quick enough. And so I think this is the dark side from, you know, what I'm hearing today, that the dark side of moving that quickly is you've, you know, the law of unintended consequences still applies, and you may have opened up Pandora's box in some way.

And this, this is not new. Like, we act like this is new as a— and maybe I'm a fairly old soul, but if you think about back to the origin of cars, and I've talked about this a million times on stage, being hyperbolic, but I've talked about this a bunch of saying, In the beginning, anybody made a car. If it fell apart, it was kind of caveat emptor. It didn't matter. And as you started making a scale and understanding how this changed the world and how much it matters, how big it is, how large that is at a scale— like, imagine just driving down Dallas freeways right now. How many cars are around you at all given seconds? That tells you the scale of this. And eventually, you just had a point where, no, you have to have a process. You can't just want to rush to market. It's going to take me a whole year to make another model car. It's going to take me that whole time to get through testing and get to a— you start getting to a point where airplanes don't fall out of the sky every day unless they're made by Boeing, uh, but they don't fall out of the sky.

Only parts of those fall out, but because there's little loose parts. Yeah, that's true.

Sometimes it's just a door chunk, right? But the point is, is that in, you know, in Boston, we, we rushed to market to put in a molasses tank to be able to deal with the amount of molasses demand that was happening, and, and it coated 3 foot of, of molasses up the walls of Boston, right, all the way to the water. And so at some point you have building codes, and instead of having studs wherever you want them, they're 16 inches on center in hurricane areas and 12 inches on center when they're in high flood and hurricane areas, right? Those types of things happen. In our world right now, we're still at the end of this age of building technology without anybody giving a shit at how horrible it is. Consumers not being aware or able to understand the delineation. Nobody's dying yet. And now we tacked on the speed of going to market of AI in its infancy. This is a perfect cacophony of where we sit that drives Microsoft to have to make that decision and Apple to have to make that decision, um, and even us as practitioners sometimes to have to make that decision. I think though, we're at this time where this will probably hasten the change that'll happen like it did to the auto industry.

And I think we're starting to see some of it, right? We were talking in the green room about this news that just dropped this morning. Now all of a sudden Microsoft and Apple at the same time are saying, whoa, whoa, whoa, maybe we don't want to be so—

We need to be off this board seat.

Yeah, yeah. Now they're giving up board seats at OpenAI because they're worried about SEC regulation. So it seems like Daddy's now paying attention to what we're doing and they're hearing something maybe faster than we are, that oversight, if not some regulation, may be coming.

Yeah. And in this case specifically around OpenAI, right, British and US antitrust watchdogs having concerns about Microsoft's influence over OpenAI. But it doesn't just end there, right? I mean, it continues to go on as that goes. And one of the things you brought up before, Robert, of this is Copilot PCs. Right? This concept of bringing actual silicone capabilities at the endpoint. Alex and Robert, what are y'all's thoughts on this? And I have mine, but what do you all think?

Well, Apple's going there, too, right?

I want to take the flip side to all of this. I think it's cool. Remember that even things like Microsoft Recall, as much as you guys might hate it, and yeah, I have my reservations about it, too, But there's a value and benefit, at least there, you know, it started that way, right? So while I have concerns about implementation and making sure that the product is sound and secure and privacy and confidentiality is all there. And I don't want to sound like I've got a blindfold on that I'm just going to ignore, like just assume that they're going to protect me. But let's start with the premise that a lot of this technology is super cool and has the ability to enhance productivity and allow us to be more profitable, grow faster, things like that, right?

100%.

Let's not lose sight of the fact that, you know, this is good for us, but counterbalance it with this whole conversation that you got to be really careful about how to implement this stuff.

Well, you know, I think the point that came up to me when Microsoft said Copilot PCs and Apple's going, "Hey, we're going to do it with on-chip," you know, on silicon, we're doing on device, all that. Basically what I thought of is I can't run away from it anymore, right? And it was one thing when Apple said we're going to put it on all of our devices, and I'm like, okay, those are mostly soccer mom devices, and that, that's fine. When it lands in Windows, it's in every, every Fortune 500, every, you know, every business. Basically it's on every desktop, and we can't hide from it, and we can't really well turn it off. Although we are seeing Microsoft start to roll some Intune services and some things around being able to identify and manage and lock it down. We're not to a point where we can turn it off yet, and we're not to a point where it freaking works. So we're rolling out cars that'll do 60 miles an hour with no seat belts at this point.

Yeah, yeah. Or the wheel, like, doesn't turn every time to the left.

Most of the time. Sometimes when you turn it left, it goes right, and that creates a problem.

It'll randomly hallucinate and turn right.

And that's my problem is, to, to the point of earlier, I think we're rolling it out faster than it's ready. You know, you mentioned talking to a bunch of new vendors, and a vendor came to me and said, hey, we do all these things. And we looked at their product, and it looks like a combination of 5 or 6 other security products. Like, we do them all. Oh, and by the way, we use AI for SOC, for, for SOC analytics and all that. And I went, great, well, my AI lies to me all the time. How's yours doing? And they went, oh, well, it's getting better. I said, Well, getting better doesn't give me the confidence that on a certain 4th of July weekend when things go sideways and as an MSP, I need to know that there's a human staring at glass to make a hard decision, that they will make the right decision. I don't trust that yet. We're just not there.

This goes back. Well, this goes back to one of the things that Sunil, you and I did in a talk that we did together about using AI in the right way. Like, for example, around security specifically and a little bit tangential here, it's this understanding that AI is not deterministic, right? It is at this, at this time, it is non-deterministic, at least LLMs, right? And so if that's the case, why use it for deterministic functions? Use it for non-deterministic functions, things like having great scenarios to train for in a tabletop game or having ability to identify potential things you want to dig into. Those are great machine learning tasks and great things like that. But saying that's a threat versus not a threat, that's deterministic, right? And I just don't think that's the appropriate use of that. Side tangent there because I want to dig back into something that Robert said too, in the sense that AI is hope, right? And it is absolutely the next iteration, evolution of something we've been chasing since the beginning of computing power, right? Somebody said, at what point does singularity happen? At what point does a computer have the capability of a human and have this efficiency of scale? And we're scared of it. I said something on a podcast with Will Dowling and the folks at ConnectWise, and it came from a TED Talk. That I watched where somebody was talking about, you know, just general technology. This was 10 years ago or so, but she said, hey, you know, when, when we invented the tractor, if it wasn't for the tractor, we wouldn't have had porn. And this was this premise of this discussion. It was an interesting and kind of humorous take on it, but saying that the tractor empowered the ability to, instead of having 17 kids, that those kids would live on the farm forever, they'd have their own farm tangentially, we'd all work the farm enough to have enough food instead. You have one dude on a tractor and that does the same thing. And so people moved to the cities to support industrialization, the change in industrialization. They then lived within a mile of where they were going to be born and died. But it created all these new jobs, right? You now had a pickle packer and you now had a— like all these things that came out that wouldn't have happened, that would have happened at a micro scale on the farm, became operations, became all these things. And what the point was of this talk is that humans are two things. They are extremely creative. We are extremely creative and we are greedy as hell. And when it comes to those two things, they always invent new things. And it was her premise, really not digging much into the porn, that all of these other industries got created because you let humans be free, having enough time and some money to do things. And boy, they're inventive and greedy. And my argument is, as AI becomes that capability, we will still continue to move forward. And I don't mean the negative connotation of greedy, but we will still continue forward to take all those people that may be in a new role or new capability to be inventive as hell and make something really cool. Yeah, sorry, I'll get off my tangent.

No, it's okay. That was wonderful, Matt. But I do want to actually go back to another point that you made earlier. But before I get there, it's the creative destructive forces of capitalism at work.

Yeah, amen.

That's what I learned in economics in college. And I think it's spot on, right? You mentioned pickle packers that also created jar manufacturers, right? And the trucking industry.

Yeah, yeah.

You know, the people who, you know, pack them in stores.

Yeah.

I mean, I'm a firm believer that I'm not worried about the destructive parts of it because I think the creative forces are always greater than the destructive forces. But you said a word earlier that I want to fixate on just for a minute. You said that AI being used in deterministic ways. And I think that's brilliant that you brought that up. And it really should be a focus for all.

I credit Sunil Yu for that. I don't want to steal any thunder.

So I just want to make sure and say that. Totally. Sunil, I'm sure you're listening because the entire, all 7 billion people on the planet are listening to this podcast. And if not, we're going to come find you.

Last time I checked the numbers, that's accurate.

We're rounding up a little bit.

It is. True news.

But it illustrates a point that we've been making on this podcast is that we need a common language and we need to understand the theory of all of this stuff. And we need to understand that AI just isn't a thing unto itself, that there are things like deterministic AI, and what is the implication of that? What is, what is that? How does that impact the world? And at what point do we draw lines? Until we have a common vocabulary, we can't even begin to start wrapping our arms around these other larger issues of privacy and confidentiality and, you know, taking things too far or not taking things far enough. Yeah.

And we're an industry that struggles with language anyway, right? I mean, we already have too many nerd words that none of our customers understand. True story. Now we have to add some more.

I get it. And you know what? I've struggled with that a long time. But you know what? Unfortunately, our other competitors, let's say other countries, are going to get their arms solidly around this. So, yeah, I don't think— stop your whining.

I don't think you're wrong.

Actually, I think it's—

I think actually, I think this language, being able to speak this language across our entire community or even outside of our community, folks who are not IT nerds, is probably more important than being able to talk about the difference between a hard drive and an SSD and how many gigs, right?

It is and it isn't. And it is 100%. So I'm 100% in the is camp. I just want to make some counterpoint here. There is a certain amount that my doctor is not going to say, "Matt, you need to understand the nature of Staphylococcus aureus and how it comes from, you know, Staphylococcus-resistant aureus." I don't even know, but I don't expect to know those words, right? I know that I have methicillin-resistant Staphylococcus aureus if that was the case, and it would not be great. I would have something I have to deal with, but I wouldn't want to understand it. I wouldn't understand the methicillin resistance at the cellular wall. I would No, none of those things. That's not something I need to understand. We're into an industry right now. And this goes back to my point of AI coming on the tail end of us being a just fledgling industry in general, even though we've been around 80, 90 years. That's typically when something gets to that point. The automobile industry was around from 1897 until 1966, after the National Motor Vehicle Highway Traffic Safety Act in the United States. And so the point is that this is typically the way this plays out. So my argument is that because we're at the end of this where anybody can be a technology person, like you call your kid because to your point it was flashing 8s on the VCR and you were the one that could fix it. Like that was our world, man. Like at least mine. And so when you think about it, we're now at a point where we have to regulate this industry in some way. We have to have some barrier to entry. You can't just say I'm a technology provider. You can't just say you're a doctor. You can for a couple of days, but probably go to jail for that second or so inspection of prostate that you shouldn't be doing.

Right.

So there's, there's some, it's a weird doctor guy. I don't know. I came up with this word. It's strange.

I'd tease.

But the point is, there's not that in those worlds. We're in an industry where all of the terms that we talk about are fucking requisite for how we do what we do. They are requisite. They're the language of being a doctor. They are the language of being an attorney. That is the way it is. And yet we have to educate someone to get them to believe us because they're shyster attorneys. Imagine you had no bar. Imagine you had no ADA or AMA or globally, you could take the same extensibility. That's where we sit, boys. And now you're going to tack AI on top of that as if it's going to fix it when you say we have no concept of privacy. We have, because there's nothing that forces us to, or the big players or the little players.

Anyways, Matt is screaming at the top of his lungs for bigger government. That's what I just heard.

Not bigger government, self-regulatory motion.

Okay.

I love that.

I love it. I would love to.

Come on, chant it together with me. SRO.

I'm SRO, baby. Self-regulatory organization. It's definitely. But we're definitely there, right? And but my point around the language isn't that we have to— that a customer has to understand all the minute details. What they do have to understand is where are the limits? What's it capable of today?

What's the decision to make tactically?

Yeah, right.

For them.

And where it starts to fall down. Where's what hallucinate— what hallucination means to you, what the inability to be an actual deterministic tool means to you. As a business. They have to understand that. And whether that means that they have to understand the language or that they have to know that there is someone with the right expertise who they can call on who will legitimately give them right answers.

Here's a great simple example of deterministic versus not. I need to make a PowerPoint. I actually have to make two more in the next two weeks. I can either ask it, hey, make this follow a format that's mostly green, has a left-right division that has a blue and a red on it, and show me these graphs. That's non-deterministic. That's creative. That is the nature of making it easy for me. And that saves me an hour, let's say. Right? Let's say that saves me an hour in doing this. Same thing with a DALL-E image. Instead of me trying to search for the perfect image, I just make it. Right? Those are very fast, non-deterministic things versus I want you to describe all the things about cybersecurity that an entry-level person needs to know. That's trying to be fairly deterministic about stuff that should be my subject matter knowledge. So what I'm saying is use the tools for making my pretty graphics and getting it the way I want and getting my vision on paper, not being the subject body of knowledge because it frequently hallucinates, which means you have to have a subject matter expert go figure out if it's lying. So that's my point of deterministic.

Well, and even there, right, we can use it to defeat the blank page. Hey, help me build this outline and I'll figure out where things go, right? Where I need to supplement it, where I need to divide things out. I'm working on a monster-sized course that tracks revenue through MSPs all the way from marketing to account management and renewals and sales and everything in between. And asking it to help me fill in the blanks or asking it to help me, you know, kind of lay out the big picture so that there's a logical talk track and a right order for all this. Super valuable.

It puts me at 80% of the work, right? And I'll do the last bit.

And then I'll find the real big gaping hole in the middle that you missed.

Yeah.

Yeah.

Yeah.

Or the thing where you really took me down the wrong road.

We talked about that earlier in previous episodes too, right, Alex, about it's still up to humans to validate the data, right? Or the outputs. We cannot assume that AI is just going to produce perfect output, that it really is up to us. And I think what I've been talking to my clients about is acceptable use policy when it comes to AI. So at least even from a training perspective, that we inform end users that they need to understand that the outputs are not perfect, especially in those deterministic ways, Matt, and that you must validate it because your name is going on it.

I want to talk about something here that also comes out of this, gentlemen, is that everybody misses the giant elephant in the room. If you believe me that everybody has had a stupid S drive, data has just been jumbled together. Nobody knows where their data is. Nobody's ever focused on that from a security, privacy, or otherwise perspective far before stepping into AI. We call that shared SharePoint now. And you believe me—

what's that? SharePoint now.

True story. And you believe me that now people are rushing to use AI as an MSP. And I will say to any of you that are actually in companies out there watching this or listening to this, as an end company, you need to get your shit right before you turn on AI. And if that is the case, there's billions of dollars— sorry, it's going to come at the expense of you not dealing with your S-Drive for the last 10 years. I love you. I'm sorry. But the reality is, is that you've now got this tech debt that is built up that you need to deal with. But for MSPs, that means profitability, that means value, that means being able to bring more value out of their data by helping them sort it. There is a business value to doing that. And I believe, Robert, this is a proserve. I believe there will be ability to go to a customer and say, yes, I can get you AI ready, but this is the process we use. Here's the things you need to do, and here's how I'm going to show you how to do that. And that's going to be a proserve. You have this many hours on it, it's going to be this much time based on this much volume, or whatever the number is. What do you think about that, Alex, or Robert?

It's not just cleaning up data either, right? It's data classification because there's certain parts of data.

Amen.

There's certain kinds of data and nobody's been doing data classification unless it was absolutely regulatory mandated. But if we don't do some data classification, then we're not sure what we can feed into AI and what we can't. Amen. And that overzealous intern is going to paste the formula for Pepsi into that AI. And next thing you know, ChatGPT is going to be giving somebody else answers that relate to the formula. Wheel for Pepsi.

First thing I typed into Copilot is, what is John Street's salary? Didn't give me the answer, but I was looking for it. Somebody did a good job.

Somebody over there either gave it incomplete information on purpose or on accident, but regardless, they did a good job.

Well, in the spirit of the podcast, not the whole series title, MSP Dumpster Fire, right? Matt, you just threw another match into the dumpster fire there, started a new kindling about what are MSPs going to do to provide professional services around AI and building out those types of services to help someone along their AI journey to talk about all of these things that we're touching upon, not even in just this episode, but across this entire series. How do we formulate that? How do we create a service around that? How do we build process around that? Who's got the expertise to be able to go in and do this? I mean, listen, I think as we started this whole podcast, I admitted very early on to Alex that I felt completely lost last year and went into this full learning mode on my own, just trying to absorb as much as I can. And while— and thank you, Alex, for giving me a lot of credit for, you know, being maybe smarter than some on this. Topic, but I just still feel like I've got so much to learn. And I can't imagine how the average MSP out there is truly prepared to have these conversations and be a thought leader or a guide to their customers on this topic.

I can see Matt's got something to say, but before he does, I want to ask you both a question. A big void right now. I want to ask you both a question before Matt carries this on. Rob, how many hours do you think you have invested? Invested in learning around AI today?

That's, um, is it easily, is it 100? Yeah, it's easily, I was about to say it's around 50 to 60 and it could possibly be more.

Okay.

Matt, what do you think you've got invested in learning and education around AI?

Tangentially larger than that, right? Just from impact and use and other things in that nature. Study-wise, probably 60, 70 hours.

I'm probably where, so the MSP who's listening to this and you've put 8 or 10 hours into poking around in AI, like, think about this from the fact of you may not be ready, but this is one of those things that's going to require education and upskill and work. And we just have to go, we have to go take this thing on. And the only way to do that is one little bite at a time.

Yeah.

And I was sparked by Robert's point of saying, Robert, your question of how do we get to all this? How do we deal with the brevity or the breadth of it? How do we deal with the depth of it? That's why they made jar makers instead of my pickle guy making jars. That's why they made lid makers instead of the jar maker making lids. That's why they had salt makers that brought salt that I could make in the brine. Like, you know what I mean? I think this is the point is we're watching.

Yeah, yeah.

We're watching the supply chain exist. There's literally a company that— but who was it that just started a company to, to do AI implementations for people? You saw Jimmy Hatzell, right? You see people that are taking on and creating supply chain for that knowledge.

We had Jimmy as a guest on an earlier podcast. So if you haven't heard that one, go back and check it out.

Yeah. So yeah, that's my answer to that, Rob.

And if you take that another step further, like IT has gone through this revolution once recently, right? It used to be the IT guy was the IT guy. Now we have the IT guy and the security guy. And I don't think we're far from the IT guy, the security guy, and the AI guy.

And the AI guy.

Yeah.

Yeah.

But you're also seeing this at the same rate. There used to be the Exchange guy. And now it's the licensed procurement lady or guy, right? Now, what a difference. And that's my point is like, I genuinely, like you can even get now in Copilot, I can get my custom Copilot that I get to train. I get like, they're commoditizing that. It's that same commoditization journey that plays out. Like at some point there was a massive commoditization of almost every market. And I do think that what we're seeing is, If you believe me that some of the— well, this is a long side tangent, but if you believe that the only way to really solve that secure by default is by choking 3 throats rather than millions or billions, then you start seeing that that same commoditization plays what makes the average consumer not die when they drive their car. It's the same thing we talked about earlier. And so, I think anytime you have something this systemic, you usually see it play out in that Pareto principle. You usually see it play out in that, you know, 80/20 rule that 20% of the companies will hold 80% of the revenue. But in the same breath, you'll have that capability enable the fact that when I turn on that light switch, that light I screwed in that common socket turns on, right? And that's where we're starting to head towards. You know, I'll get off my pulpit here before I fall and break my neck, but I do think we're heading down that path of that supply chain getting created while we're trying to heal from really sucking at dealing with security and privacy.

Yeah.

And I think that's an interesting mix to be in the middle. Well, I'm not sure how you said that.

Like, we're not— even though AI is the new, you know, thing for us to be hyper-focused on, let's not lose sight of all the basics and all the other stuff that we're doing too.

You and I live in every day. And by the way, that is new.

Those are things that we're not—

go ahead, Rob.

I was going to say those are the areas that we're not even really complete subject matter experts and perfect at execution on So before we go abandon things like cybersecurity and jumping on the AI bandwagon, you got to remember, you still got to be good and get better at that other stuff.

We've only been working on that for 25 years, right? I mean, we were trying to solve random admin passwords floating around at my MSP in 2004, which was the year I bought it, not the year we decided we should start worrying about that problem. It's been around forever. AI hasn't been around forever, so it's new and we like to talk about it. And to Matt's point, every new product that you touch has some AI advertising in it. That doesn't make it ready. It doesn't make it right. It just makes it the marketing buzzword of the week, which I think kind of gets us to a point where we ought to be wrapping up. We're running low on time here. But I want to end up with Matt. If you're an MSP today, like you put yourself in Rob's shoes, you're running Progressive Computing. Rob, you just got fired, buddy.

Sorry. Old one.

What's the conversation you're having about AI with your customers? Take out all the nerd words, take out all the fancy stuff that we're talking about. What's something that MSPs can take back to their customer tomorrow and start a good, healthy conversation around AI?

Yeah, I think it is about being deeper in the business, right? What I mean by that is if you already understand the commoditization is happening, and I'm talking to a customer at some point and saying, hey, you have the same tools as everybody else. You have the same Microsoft Teams, you have the same Outlook, you have the same software, and all those things are enablers, right? And if I took them away, you definitely believe they were an investment. That's a reality as a business person. Much in the same way, there was a time you didn't have these computers. There was a time you didn't have this software. There was a time companies existed and did this every day. Would you ever start a company without that right now? Hell no. Okay. AI in the very near future will be that same statement. How do we prepare you for that? How do we get you to where your competitor is going to start natively? Right. The other thing I'd want to make sure people are hearing, and this isn't the security buzz, is I like to talk about the two companies, and I'll go quickly. If I was a legacy company running on old technology, servers, infrastructure, building up things, and I wanted to start a new software versus a company starting tomorrow that is basically Azure AD or Microsoft and external SaaS applications. That's it. Tying those identities together, making it easy. I want to add a new application over here. I call my MSP, I get a quote, I figure out how much storage and compute I have, I spin it up, I wait on software to be installed, I call the third party, they test it, they make sure it's ready. We now go into training and we're $30,000 and 6 weeks in. This guy or gal turns on a switch for $20, hates it, turns it back off, and it's done in 4 hours. I want you to bet which one wins. So if we start there and know that this one probably wins— yes, they don't have the business, yes, they don't have the momentum you do, they don't have the profitability, they're faster, more agile, more modern, more capable, more securable, all of those things— you are the loser. And that's okay. That happens all the time when major new technologies happen. You can be the loser. So you're already on the end of the wave of not adopting cloud strategies. You're already on the wave of using shitty old line of business applications with shitty technology and applications and risk and all the things I can get into deeply. Or you're over here. Which one do you want to be? Well, if your strategy is to look like this in 2 years, my strategy is to get you there, and it's going to take a lot. You don't have to win, but we can help you. So the argument is, is that AI just accelerates that conversation. Right now, I'm not ready. My stuff sucks. It's not AI ready. It's not already even in the cloud where it could all be done that way. I don't have the capabilities to make it valuable. And now I'm also behind the 8-ball and losing to the new guy can start up. Now, I've been screaming this for a long time. And I'll also say that my VP of Operations, Tony Miller, always told me, Matt, I'll never disagree with you on the what, just the when. That's a challenge. But the point is, is that you could imagine my conversation with a client of saying vision from a vision perspective. Which one do you want to be and which one are you? Same thing with AI. Are you ready? Can you just turn it on? Is it safe? Is it secure? You're going to waste intellectual property. You're going to give away PII, PHI, get yourself sued. Are those things there? No. Okay, then let's get ready for that. Let's talk about our journey over the next 12 to 24 months to get where we are so that when it does matter in 2 years and you're not behind, you've already gone through the suck, that valley of suck that you have to go through to take that change that you need to change. That's my thoughts, Alex. I don't know if they're valuable or not, but that's how I'd approach it.

Rob, did that spark something in your brain?

Yeah, all sorts of things. I thought of at least 2 new businesses that I can start, and as well as the change that I can influence here in our existing MSP.

So, tiafi.ai.

Just whip those up this weekend. I'll just, you know, start at, you know, just fire up some AI tool and say, build this business.

That's the entrepreneurial mindset in America these days. I actually do have a really good idea out of one of them.

What he said. I'm going to keep that quiet and close to the vest.

My takeaway was just a single quote, and that was, Mr. Customer, you know, how do we get you, the, the old company who's been around a long time, to the point where your new competitor is going to start? That is a powerful statement, and I don't think it's been said enough. So you may hear me talking about that on LinkedIn one of these days and stealing from that. Guys, I think we did it. That is a banger of an episode. I think we've really picked this apart. Microsoft, Apple, enjoy your not having to sit on board seats at OpenAI and trying to keep all this at arm's length.

And we love you for what you're making.

Well, we watch the dumpster fire burn.

It's warm by the fire.

It is warm. Gentlemen, thank you so much. It's been a pleasure. We've enjoyed season 1 and we'll be back with another season here in not too long.

Live long and prosper.

You can do that. Oh God, it hurts.

I'm a Star Wars guy. I can't do that. I can't I do that. can't bring myself to do it. Take care, fellas.

Gotta be able to have a great one.

See y'all.

Bye.

This podcast was created by me, Alex Farling, and my good friend Robert Jaffe. This podcast is produced by Francis Belland. The music and artwork for this series were produced by artificial intelligence and edited by Francis Belland.