Syncro’s AI Shift: How Model Costs and Adoption Create New Separation Among MSPs

[00:00:01] Every MSP platform vendor is shipping an AI layer this year. Syncro shipped theirs, a connection between live ticket and device data that the AI assistant technicians already use. I spoke with Michael George, their CEO, and DeZep, their Chief Product Officer, about what this is. Does this help close the gap for a 10-Tech MSP or widen it? This is the Business of Tech. I'm Dave Sobel.

[00:00:31] Michael George, your CEO at Syncro. DeZep, you are the Chief Product Officer at Syncro. Both of you, welcome back to the Business of Tech. Delighted to be here. Thanks for inviting us, Dave. Well, I always love having you on. You guys are both now frequent flyers because you always come with some interesting stuff. We're talking right after you've done a recent release, and I want to get a sense of you. You've actually done some work now to integrate with Anthropic and connected MCP in there. Talk a little bit about, you know, in a quick sense, what can a

[00:01:01] technician do today that they couldn't do before the release?

[00:01:31] And believe in an open platform to the benefit of the channel in particular, right? I mean, you know, we've always been very, very open, and we want to integrate and have APIs and do all of those things.

[00:01:45] You know, this is sort of an API on hyperscale. Like, I mean, this is like not even 2.0. It's like 2.0.0, you know, version of an incredible way to go and access all of the things that we do in, you know, within our system.

[00:02:04] And we're really, frankly, just proud of the fact that we've done so much, such great work with it that Anthropic has gone ahead and authorized us. We're the first and only RMM and TSA platform to be now native, a native connector in the Anthropic ecosystem.

[00:02:27] So I think that speaks volumes for, you know, the level of commitment that we have, sort of the openness of it. And then sort of the last piece, and then I want D to hit this, you know, harder, which is, you know, the whole UI, UX issues around, you know, this is clunky, and I don't like the way you set this up or whatever.

[00:02:50] And I'm not talking about us. I'm just talking about in general, in the tools world, you know, technicians have to kind of go from system to system, and they all have their own sort of UI, UX and nuances of it.

[00:03:02] And this sort of eliminates all of that complexity and all of the sort of the, you know, diversity, if you will, of approaches today, because it puts the interaction with the system, not only in the hands, but candidly in the mouth of the technician.

[00:03:22] They can ask in natural language, you know, they can query things, do QBRs in advance, get latest ticket information, be prepared, you know, or all of these things. Just inquiring in very much the way you would want to, not having to go through any level of sort of system or system differential.

[00:03:49] One of the things that we've heard, you know, from speaking to our partners, and, you know, especially our advisory council, you know, MSPs are a great group, right? Creative problem solvers, always like looking for building the next thing and kind of like helping each other to advance. And when I saw what people were doing with MCP servers in the community, it kind of made me really think about MCP in a whole new way.

[00:04:19] And it made me realize it really is like an API 2.0 for these folks. And it removes all the barriers of an API. You don't have to be a programmer. You don't have to get that technical. And yet you can integrate different systems, interact in natural language. And so I saw some of the possibilities play out in our partner base.

[00:04:40] And we decided that we really wanted to go all in on making sure our platform stayed open as part of the ecosystem. I think that, you know, MCPs really help future-proof platform use, right? I think when people are looking at buying in the same way they go, do you have an API? Can I plug you in with other systems? Or am I locked into you? I think an MCP provides that next layer of protection for folks as well.

[00:05:09] Now, Dee, I want to follow up a little bit because you've kind of you've talked about the fact that the MCP adoption right now looks a lot like slides. You've actually sort of said 5% to 10% of partners, and they tend to be the bigger, sophisticated jobs. Now, Michael, you've also said that AI is going to be the greatest separator the industry has already seen. So I'm kind of asking the two of you to put this together a little bit for me. Is this something that helps the 10-tech MSP catch up?

[00:05:33] Or is it something that helps the top of the market pull away and get further better at what they're doing? Candidly, both. I think both of those opportunities exist, candidly. And it really gives the pen technician an opportunity to scale their business like they have 100 technicians.

[00:05:56] It does allow them to scale, to grow their business, to add more customers, to add more endpoints, to add more cloud and everything else without having to add a lot more people. You know, they're just going to create huge efficiencies within them. And it gives them a unique opportunity because in the smaller organizations that you know, they're kind of caught on the hamster wheel a little bit, right? Like trying to get a new customer, but I'm trying to figure, you know, whatever.

[00:06:25] And this like gives them that breathing room to go out and grow their business and to scale, right? And so, but they need to take advantage of it to do it. It's not going to happen naturally. They have to adopt. You know, they have to go after it. The point about separation is just, and Dave, you know, you were with us back in the day when, you know, at Continuum, you know, when D&I were there.

[00:06:50] And we launched the very first cybersecurity solution in the entire, you know, MSP business. We partnered up with Sentinel One. Now they're the number one, you know, you know, EDR, I would see in the space and everything else as a result of it. But they were never part of it before we brought it in. And we said that this is going to really create separation, that this is going to be a security first industry.

[00:07:13] If you ask anybody, any business owner, what matters most to them being, you know, proactive and preemptive on a security basis. And then when there is an incident, be able to immediately react, respond, isolate, remediate, right? Like, like this is like number one. And then number two is, is just provide great service and everything else at a really reasonable cost to me.

[00:07:39] And so AI gives people the opportunity to do both of those things really, really well. And again, there became the formation of an MSSP, which most MSPs really have, obviously, cybersecurity capabilities within their, you know, portfolio offering.

[00:07:58] And in the same way, they're going to need to have these AI capabilities to allow them to provide a super high level of efficacy and advancement in the way that they serve their customers. But do so at a better economic even than they can today, needing labor to do that work. So, again, is this going to be, and this is moving so fast.

[00:08:25] You know, I will say, even as we speak today, it's the slowest we're going to be seeing it in the next three to four years. I mean, this is just accelerating. And so anyway, that's, that's what's, this is going to create a massive separation in the industry. We'll be right back after this message. Here's a reality every MSP knows. Native Microsoft 365 security leaves gaps. Those gaps land on your desk.

[00:08:54] Proofpoint 365 total protection closes them. It's an MSP first platform that unifies Microsoft 365 security backup and compliance into one integrated console. High security efficacy, less operational friction, and multi-tenant management that scales as you grow. Protect clients against modern threats and stop stitching point tools together. Built on Hornet security, now part of Proofpoint.

[00:09:20] See it at Proofpoint-total-protection.com. And we're back. You put some numbers on the record last year, which I want to say I highly respect someone who's willing to actually put some numbers out there. So I want to follow up a little bit. Last September, you said Syncro expected to autonomously handle 30% of level one and level two technician work by the end of 2025 and 50% by 2027.

[00:09:49] We're talking in August. Did the 30% happen? Did the 30% happen? What's the number today? And how are you measuring that? So I think we've learned a lot along the way. And we are on track for our 30% number, but it's for this year. And so as we're going through and looking at the work, it's for this year. And then we believe we can get to the 50% in 27 as well. And so it's taken us a little more to get to the ramp that we wanted to see.

[00:10:16] And as we've learned along the way, we've started to experiment with different models. We're feeling really good about our ability to automate the work. And the way we're approaching it is based on what we're hearing from our partners, which is they want to keep their hands on the wheel for now. And so we are allowing them to have complete control over what gets run, what gets fixed, the audit trail for that, and really having that complete picture of it.

[00:10:44] Because the idea of building trust with the system is so incredibly important. And so we're making sure in everything we do that we're kind of allowing people to easily decide what they want to use and then have complete oversight into how we're doing it. Okay. And thank you again, by the way. I very much respect the transparency that will be reported numbers. So thank you for that.

[00:11:08] And I'm going to use that to baseline the next two set of questions because I've been looking a lot about the way the economics of this all work, right? And it's interesting now because you've got a component of what you're doing that I would say is within the system, right? Now the 30% that you're targeting this year. And now we've got another component of this which happens outside of the system, right? That could happen in Claude, for example, on itself. So I kind of want to get your take on the way you're thinking about the token economics.

[00:11:37] Because if I'm trying to now understand the way different vendors are approaching this problem, because I don't think any of us know exactly how this works. And I would say like now Synchro has a new answer for me. There's a token component within your product, which I assume you're absorbing, but I'd like to know your strategy. And then there's the token component that the MSP has to manage within Claude. Talk to me the way you're thinking about token economics as it deploys for your customers.

[00:12:05] We're thinking about it in two ways. And so let me talk about the MCP and kind of outside of Synchro first. Okay. And in that, you know, mode, it's really people building, you know, kind of custom for them, either building custom for them or even using natural language to do simple things, right? I want to prep for an on-site visit, read over all the tickets from client X and let me know what I should be prepared to go do while I'm on site, that type of thing.

[00:12:31] And so in those cases, people are managing the economics of their AI tools they have today, right? In terms of inside of Synchro, we are looking at this across a number of trajectories, right? We really want to make this so that costs are 100% predictable and fully, you know, contained. However, we don't live in a world right now that's going to allow us to do that in a way that's going to be sustainable. You know, you've pointed it out in a lot of your work before, right?

[00:13:01] The way the world is operating right now is these large models, the frontier models that we're dependent on aren't running out of profit today. And so what we're doing is for the specialty work that we're creating, we're looking at having a credit model where we are basically, you know, kind of charging them credits based on some token use and based on the things we're doing for them, right? How many tickets are we resolving for you? How much time does that save?

[00:13:30] We really view the cost of the MSP in three layers, right? So there's your software cost, your traditional cost, there's now your model cost, and then there's your labor cost. And so we think about the model cost and the labor cost as kind of coming together to come up with what's a good COGS number for you. And so as we structure this, we're trying to make it super transparent, putting it in the control of the MSP and allowing them to actually measure, is this helping my COGS today? If it is, great.

[00:13:59] I want to turn it on and I want to use it. And if it's not, no, I don't. And I want to put it away. Now, the other thing I'm going to follow up because in September's guided resolution, you projected at 20% of tickets resolved autonomously. I kind of want to get a little bit of a sense of like projected from what and the pilot with real models, particularly because that'll give us the data to talk a little bit about tickets and tracking. So where are we with that?

[00:14:22] So the guided ticket resolution today, we have deployed out to a number of our partners and we are seeing somewhere between 15 and 25% of tickets kind of based on what we're looking at. It does go across all different types of tickets and then we're adding new resolutions as we go. And so you'll hear more from us throughout the fall, but we are tracking a lot of the data.

[00:14:47] We have, you know, kind of daily, you know, dashboards that we're looking at and making sure we're fine tuning to make sure we're hitting the numbers that we want to hit. But I ask that because that then baselines on the fact that I want to talk a little bit about autonomous. Right. So in the area that I keep seeing discussion is some some vendors are declaring the death of the ticket. Right. And what I sort of look and say is, OK, I believe that we can achieve a world of proactive IT where we're doing a lot of things with systems.

[00:15:14] Yet if we make the work all disappear so that no one ever sees it, well, then we've lost all of the measurement. And one of the reasons we've all used the ticket for so long is that it is a unit of work that can be measured. Help me understand the way you're thinking about this resolution of work autonomously. You know, what does it mean from a measurement perspective? How do we baseline that? Give me your kind of philosophy on the way Synchro is building this.

[00:15:43] Yes. Yeah, absolutely. You know, I do think when people say the ticket is dead, it's just a kind of a gross overstatement of like and then it all magically happens and we don't need to track it or describe to our customers what we did. And we don't need to baseline where we are today versus where we are, you know, tomorrow. And so I think the ticket is going to remain an important unit of work. I think it's a way to communicate. I think it's a way to baseline performance. I think it's a way to have an audit trail.

[00:16:12] And I also think it's a great place for people to have visibility over what AI is doing, because I do think we're going to want to maintain some level of monitoring and connectivity so that people can understand what's happening with the AI. We'll be right back after this message.

[00:17:01] If you're rethinking your tool stack this year, it's worth a look at logmein.com slash MSPGrowth. And we're back. So, Dee, the other thing that I kind of wanted to talk about is you've been talking about ticket summarization improving time to action. Like, talk to me about what you've seen in terms of measurements there and what the improvements have looked like now that you've got some real world data.

[00:17:25] Right, right. So, ticket summarization has actually had an even bigger impact than I imagined it would, especially in long-running tickets and tickets that, you know, are escalations. And so, we're seeing on those types of tickets, we're seeing improvements of, like, up to 40% or 50% faster, time to action and time to resolution. And then across the spectrum of tickets, we're seeing, on average,

[00:17:52] tickets with summaries versus without summaries are resolving about 25% faster. And anecdotally, we're hearing it just isn't like an unlock for, like, I open a ticket, it has, like, everything that's been tried and kind of what the state of the ticket is and what's next. And so, it's just saving folks every time they have to context switch. And, you know, as a MSP technician, like, the amount of times you have to context switch is high.

[00:18:18] And so, anytime you have to dip out and come back in, it's just this automatic, you get to pick up where you left off. And so, the data is looking really good around it. And we're hearing anecdotally really positive feedback as well. So, as we sort of wrap up our time here, I want to get a little bit of a take on something. Michael, you wrote two years ago that MSPs shouldn't need to be AI experts, right? That it should work behind the scenes. And your analogy was the idea of not needing to know the principles of physics, you know, in order to understand flying, right?

[00:18:48] But there's an element here of connecting a PSA and RMM platform to Claude does ask a technician to think more about, like, the data and the way AI works. Like, has your thesis changed? Or is this more of a transitional phase that we're passing through? I don't think our thesis has changed at all.

[00:19:11] I do think that people need to think about AI in a very orthogonal way to the way we traditionally thought about other tools and technology. And so, I think it's a little bit of a mindset adjustment, you know, versus, you know, a complete shift.

[00:19:35] You don't necessarily need to understand all of the dimensions of what makes AI work and how it works and all the other things. You just need to really understand what are the points of leverage? What are the things that you need to challenge? And it's almost like human interaction, Dave, right? Like the trust but verify. Like, you know, we got this response. Does this make sense?

[00:20:04] I will tell you that Sequoia is one of the foremost, I mean, you know, 50-year history of record results as a venture capital firm. They're an investor in this category as well. But independent of that, they did a spectacular job of publishing a paper. And it's called Services as Software by Sequoia.

[00:20:31] They put a chart together of the things that you outsource versus the things that you would insource. The things that require judgment versus the things that require intelligence. And all the things that require intelligence that you typically outsource are going to be the things that make the most sense to automate using these, you know, great technologies. But you can't lose the sense of judgment. It still applies.

[00:21:01] And that's why, like, Dee talks about the human in the loop and everything else. We need the technician to be in control of these things so that they can use good judgment around, you know, any of these activities as well. And then it's only until we've done it enough times, like once, five times, 25 times, 30 times. And it seems like the system's getting it right where they might say, you know, I still want you to report on it, but you don't need to come to me anymore. Check it out.

[00:21:30] Just like you would with a human, right? Like when you're bringing somebody in and training them and everything else. You get to a point where your degree of confidence in their ability to apply good judgment to the intellect and the outsource, you know, element of it, you know, works. So I think it's on a different sort of spectrum, if you will, of thought as to how people are going to need to approach and deal with this stuff.

[00:21:55] And no, they don't need to know all of it at the deep level, you know, at the core level, but they will need to approach how they use and think about the technology quite differently. Well, Dee, Michael, I always enjoy talking to you. Thanks so much for joining me today. Really appreciate the conversation. Thank you. We certainly hope this is helpful to your audience. We think this is an important, you know, period.

[00:22:22] Like you said earlier, you know, this is most transformative technology to come in our careers and in our lifetime. We're super excited about it. But, you know, we think things like transparency and regulation and all of that is going to be critical as we step through this together. But thank you for the opportunity to be with you today. Thank you both. Take care. The hardest part of running an MSP? Doing it alone.

[00:22:52] The Small Biz Thoughts community has been the room where independent operators compare notes for nearly 20 years. Real peers, real numbers, real answers from people running businesses just like yours. Pull up a chair at smallbizthoughts.org. Interested in advertising? Head to mspradio.com slash engage. The Business of Tech is written and produced by me, Dave Sobel, under ethics guidelines posted at businessof.tech.

[00:23:22] Thanks for listening. I'll see you on the next episode. Produced by Picture This Video. Part of the MSP Radio Network.