Inside the Pipeline: AI as a Sales Conversation Starter

So welcome back. Here we are, we're doing some Inside the Pipeline stuff. You guys should see my screen, I hope. And, you know, this is basically the livestream where we kind of unpack what's working in sales and what's not. You know, some of the stuff that we hear online or see online looks good in a Reddit thread and doesn't really work in real life. So our goal today is to get into some things that actually work in real life. And today's topic is going to be the AI sales conversation starter. super excited about AI because it's the one place really where it's a pull motion right now. Our customers are asking us for it as opposed to us having to go shove it down their throat like cybersecurity. And, you know, there's a lot of really cool stuff going on. But when I was researching for this episode, I found a really amazing number. 91% of businesses that Deloitte surveyed plan to increase AI spend in 2027. And as MSPs, they're always looking for a couple extra bucks in margin, something else to sell, another opportunity. I think that is just an absolute number that we need to focus in on and really dive into. So, you know, as AI investment goes up every quarter, we got to figure out how the MSPs can get a little share of that, and I'll talk through it today. But, you know, my real belief is that we don't have a technology problem. MSPs are adopting AI. We're starting to tinker and figure it out. We certainly don't have it figured out yet, but I feel like sales delivery is the last place that that we've got an understanding on how to go talk to our customer about it because we're still tinkering and figuring it out, which leads me to my guest. I don't usually bring guests on this thing, but today I've got my guest, John Harden. He's been in this industry for 17 years. He started out as a NOC tech at an MSP. He built a little company called SaaSlio that quickly became Auvik SaaS Management back in 2022 and then ran AI strategy inside of Auvik. He's had the engineers and the budget and the interest in growing this in a real company. And you know, you've kind of had a front row seat to this, John. So John, welcome. You wanted to give everybody a kind of a heads up as to what you're up to today?

Sure, Alex. You kind of gave the story of how I've gotten here. You know, serial entrepreneur like yourself, love building solutions and technology, and particularly seem drawn to some of the most complex problems in our industry. I would say what we're talking today about— I don't know if you finished your sentence, not to call you out, but I'm going to finish it for you because you said it's not a technology problem, but you never said what it was.

Well, I think it's a sales problem, right? I think it's a how do we have the conversation with our customer problem.

Yeah, I agree with that. That's where I was going to go. I was going to say I think it's a delivery problem. I don't think it is what to deliver. It's how to have the conversation so our customers understand the value so that they can take advantage of the opportunity. And that's what we're up to today. That's what I'm building here with my co-founding team here at Lemai is programs, process, and platform to help it all come together.

And I feel so connected to what you guys are working on because, you know, my last gig, I felt like account managers didn't know how to talk to their customers and I went and tackled that problem. And so like we both are building SaaS solutions that are— have built SaaS solutions or whatever I built and you're building SaaS solutions that kind of help people walk a conversation. It's like I always used to call Lifecycle Insights like bumper bowling for account managers and, you know, it was literally literally like, hey, if I can get the ball on the lane, the tool will kind of help me get from where I am to where I need to be. And it sounds like that's kind of the take that you're working on over there at Lemhi. And you guys kind of touched on that at SaaSlio with just helping folks understand how to have the conversation about SaaS and rogue SaaS apps inside of an organization and understand where their data's at and all that. But with AI, This is, I guess it's kind of like, I say it's not like cybersecurity 'cause it's very much a pull motion, not a push motion. We had to push cybersecurity and cloud for that matter at our customers. This time their friends are talking about AI, their boards are talking about AI, their investors are talking about AI, and they're coming to us going, hey, we need to do that AI thing. And I love this meme because it really kind of exemplifies where everybody's at with AI today. Like, hey, I need to do some AI in my business, John. I've heard it can replace all my people. I've heard it can make me 10 times more efficient. I want to do that. And, and that's kind of where it stops. And I think that as MSPs, if we can't figure out how to have that conversation, we're doing everybody a disservice.

Yeah, I agree. I, I see the problem very intimately. So I'm going to say in general, I agree here. Downstream, we're seeing a lot of noise. So they're asking for help, they're just not asking through the right channels. And let me re-communicate that. Most of the way that an AI conversation is starting today is not the rep having the QBR or the prospecting conversation. It's the rogue high pot inside of their company that's Claude coding away, creating some unstable, fragile process and asking for admin M365 full access to their tenant while they're all doing it along the way. And that spawns as a ticket. The technician says no, the customer gets frustrated, and nobody walks away any better.

This is— this sounds like the website developer in, you know, 2010 that was like, hey, I need the DNS login and I'll just go fix it myself, right? And then the next morning email doesn't work.

Yeah, exactly. And that's what I'm thinking though, is most people are missing the signal. They're not having the conversation. They just don't know where the signal's coming in. It is coming inbound.

I'll say for— That's interesting.

Some of their clients, it's coming inbound. I'll also say, while we feel like AI is moving dramatically fast, and it is moving dramatically fast if you're on the edge, you're trying to keep up at the edge, it's moving very fast. But if you're kind of somewhere in the center, that edge is super irrelevant and it's moving at a technology pace.

That's a great point because as somebody who's a tinkerer and like I'm not a technician by any stretch, like you don't want me to implement anything. Everything I build sits here in my home lab and I don't expose it to the internet because I understand that I'm a rusty old fella when it comes to my techie skills. But, you know, I'm playing with, you know, OpenCLAW when it came out and with, you know, Buzz now that some of the newer tools are out and with Hermes and some of these kind of things I'm tinkering. And I feel like it's moving super fast. But yeah, for the average customer who's just using ChatGPT and watches the model iterate by a .1 every 6 weeks, it's not that big of a difference.

It's not that big of a difference, but it's also important to highlight that we feel like it's moving fast because we're tinkerers. Our customers hear from the tinkerers that it's moving fast, and that's where this we don't know is coming. But I also want to make a point, like a lot of people are having sales conversations and they're not going anywhere because the customer is also not ready to go anywhere. I say we're right now, we're like circa the risk era of the MSP industry 6 years ago. Well, you know, 6 years ago.

That queues up this next slide. So let's go there. You know, I was looking at numbers and some of this informational data as I was trying to figure this out and build this thing. And I realized that I left the Deloitte slide that was too old. on the attribution in the bottom corner. So don't follow that Deloitte link, or you can, you'll find 2025 data. But when I went and dug up 2026 data, Ramp's AI Index was actually the best place to get this data. And it was actually super interesting. Some of the data that they showed was the percentage of companies that are deploying AI, and some of it was how much they're spending. And being the guy who listens to an AI podcast, I listened to them say in Q1, Uber used their whole year's worth of of credits in Claude and had to adjust their spending expectations. And, you know, you hear just the fringe, like you said, kind of the edge issues, but in reality, the average median company out there is only spending $12 per employee on AI as of right now. And you see how much that number has gone up just in the last couple of months. Now, the top 10% is spending $660 a month and the top 1% is spending $7,500 a month. 1% is Uber buying, you know, unlimited credits for developers and just letting them rip, right? They're spending $80,000 a year on that developer to make sure they have AI. But the rest of the world really is just in its infancy. We're spending less than we spend on our Microsoft subscription, the bottom 90%.

Yes. I want to jump to my point I was about to conclude earlier, and then I'll transition into this. Like I said, we were in the cyber risk era 6 years ago. 6 years ago when we went to all of our customers and said, hey, you're gonna get hacked, you need to secure your business, only 5 or 10% of them bit on it year 1, 10 or 20%, maybe 30% bit year 2, 30 to 50 by year 3, and then you're passed into the late majority. I just wanted to come back to that point and then I'll, I'll address this chart.

Yeah.

The reason I wanna bring that up is because I feel like a lot of salespeople are being disenfranchised with the conversation because they think they're gonna go, and bat at 100 when it comes to AI because everybody's talking about it. So surely everybody's ready for it. But I think a lot of people are still in that experimentation phase. So if I'm in the pipeline and I'm having AI conversations, I need to mentally frame that this isn't going to hit as hard of a close rate as cybersecurity was 3 years ago. But in 3 years it will. And the importance of that is to know that while right now my customers may not all be buying it immediately, If I don't have the acumen and the skill set to have the conversation in 2 years when we hit that early majority in that peak, we don't have the skill set to go close when it's ready to get getting. So I just want to highlight that. That's where I was headed.

That's a great point, and it reminds me of selling ThreatLocker at my MSP, because at my MSP we bought ThreatLocker in 2018. It was barely a product, right?

8 years ago.

Yeah, yeah, yeah. So, you know, we bought ThreatLocker, we rolled it out to exactly 2 customers. I had 2 customers that were super heavily regulated who were, you know, exactly the customer that needed it in that day and age. And I sat down with every other customer in a QBR and said, we've got this new thing. I want you to understand what it costs and know that you're probably not going to buy it this year, but you might buy it next year. And if not, you'll definitely buy it the year after. Start thinking about that.

Yeah.

And if I hear you right, that's the conversation we should be having around AI, which is You may not be ready today, but we should be talking about it and you should be planning for the spend.

Yes, we're not going to get through any slides today. Don't skip this slide because I want to answer this and I want to answer what you just said too, because there's some interesting things here. What you just said is what we call the defensive motion at Lemhi. You need to— I don't care what you're doing. I made a post today on LinkedIn. Nobody's judging your AI offering based off of perfection. They're judging it based off of silence. I say that to say that your offering, I don't care how good it is, you need to tell your customers you have something circa Farling 2018 with ThreatLocker. Just come up with something. If it's we're going to implement AI and we're going to— sorry, if we're going to implement a generative AI tool, I don't care if that's it. Just come up with something and tell me you have something because what's going to happen is your customers are going to get chirped at from AI consultants who are going to get that CEO's ear and they're going to steal your wallet and erode the trust with your client. And you're going to still have to do the cleanup on aisle 9 once they leave and leave a tech mess behind. So first point, defensive motion. Second point, offensive motion. The new MSP conversation is AI, in my opinion. If you're selling MSP services on a net new funnel, you got to have the acumen to talk AI because they're not looking for a managed IT service provider to just support their network and their infrastructure anymore. They need somebody who can help them do AI implementations. And going back to That meme earlier, you said, do what? You know, AI now, like, bridge that gap. So that's that answer, and I'm ready to answer the chart whenever you want.

No, no, you're good. If you guys don't follow John though, make sure you grab his LinkedIn handle. That's why I came back here, because he's always posting great stuff. He's got a great livestream series he does too. Coffee and what is your coffee? Coffee. Coffee and chat.

Coffee and chat. Yeah, it's pretty simple. Yeah, it's not as polished and clean as this.

Oh, I don't know about that. It's— but it's great. And I've been lucky enough to be asked to come hang out on it and it's always a good time. So make sure you follow him. But yeah, let's get back to the slide.

This chart, it's important more than what you're reading here. So you brought up the spend. What I look at is the package. So on the right-hand side, let's start there on the median. The package at $12 is a Claude or a Copilot subscription. And if it's a Copilot subscription, it's actually because it's $30 right now, you can get it like—

Yeah, so it's a subscription for 1 out of 3 people in the organization.

That's what I see in that data point.

Yep.

That's an important distinction from the left-hand side. Why is the delta so big on the left? As you move towards the middle of the top 10%, you're still in what I would call SaaS subscription AI land. And what I mean by that is I upgraded to Salesforce Einstein and that cost me another $120. I added Lexful or Harvey subscription to my AI feature for my legal firm. That $200 to $600 barrier is still probably AI SaaS.

Yeah, I'm using Clay to go find leads and it's got some AI components, right?

Those kinds of things. Yes. And so far right-hand side is I have a GenAI subscription, one in every 3 employees. Middle is probably now that top 10% is I've invested in some SaaS tools for AI. The left-hand side story is I'm into tokenomics. And so you're into the real tokenomic landscape on the left, which means you are actually running agentic AI. And that means you're doing autonomous workflows and automations with AI on the left, on the top 1%. So if I'm reading this chart as a sales rep supporting small-medium businesses, I know that enterprise is always ahead by a year or two. So I can look at this data point on the far left or probably the middle and say, this is where my clients might land in a couple of years or a few years. But I also know that like I need to be equipped to understand agentic workflows and automation and those conversation points in a couple of few years. But right now your customers are in generative AI land. They haven't even— maybe they haven't even adopted a generative AI tool of choice across the business. And that's some of the easiest, lowest-hanging fruit in the world to help your client with. So that's how I read this chart. Yes, it's spend, but really it's packages of what they're using.

Well, and I also look at the trend, right? You know, when things start to go parabolic, they typically continue to go parabolic until something gets a little crazy. So, you know, our $12 a month on the far right, probably becomes $20 to $30 a month in the next year where everyone gets a license.

Yeah.

Which means the blast radius for an employee doing something stupid gets bigger. The blast radius for someone being uneducated and making a wrong choice gets bigger. But also the opportunity to enable these people gets 2, 3 times bigger too. There's still a lot of opportunity in that median on the right just to help those folks understand how they could use AI throughout the rest of their business.

And I'll also mention inversely, risk probably goes down a little bit as spend goes up, which is interesting. There's like risk goes up in one category, goes down in— yeah, risk goes down in one category and up in another category. The risk that probably goes down in one category is this perceived risk that if you're going to put data into a free tool, it's going to train into the model and it could get leaked. Look, I get it. I'm going to argue if you're telling your customers that you really should just stop because the risk is infinitesimally zero, but it exists, I suppose, in a crazy black box Doctor Who multi-universe type scenario. It exists. Realistically, the risk that is existing today isn't data training the model, it's tribal processes and tribal knowledge being created in the model. That goes away when Sally in accounting leaves and she had a bunch of processes built around FP&A inside of her free OpenAI license. Like, that's the real risk, but you can call it shadow AI if you want to scare your customers. Well, on the right hand, the new risk goes up, which is as everybody gets in a subscription, people start putting more and more processes in, which creates more process risk, but also they start integrating into other tools and there's a new layer of risk that comes in there.

So they start building their own tools and now there's more weird things that we have to manage and the potential for more support tickets. Interesting thing to think about for MSPs is, you know, is the thing that somebody went and built with their AI model, is that covered under your help desk agreement? Because that is probably not in most people's agreements one way or the other.

No.

But your agreements should have something that excludes custom software. That is not a regular help desk issue. And if you're not careful, that expectation is going to get set that you're able to handle and help with those things.

Yeah, I've heard of a couple MSPs who are actually going to their customers and during their QBR they're saying, we have a managed AI offering and we want to introduce something for you. And if you don't want it, that's okay. But I need you to sign this agreement if you're only using AI in your— yeah, you can sign this agreement if you're using AI in your business and acknowledge that you didn't need services and support around it. And so we'll move to like an opt-in, opt-out only model here probably in a couple of years again, just like everything else.

But we'll start to bundle it in renewals and support agreements and things like that. As we bring on new customers, it'll just be— we'll wrap it into what we offer them. But we have to be aware that this is— man, nothing is more opaque and needing for time and materials than trying to support somebody building some weird wonky SaaS product and coming to you with problems.

Well, that's what my post this morning was actually about. It was about an MSP that we were talking with and coaching through building their program, and he just got frustrated and he said, you know what, be damned, I'm doing this. My techs are getting frustrated because they don't know how to answer the conversations and they're yelling at me saying I need to figure something out to offer them because I don't know what to— because I as a tech don't know what to tell my customers we can do.

Mm-hmm.

The techs being the right people in the right seat want to help but have no idea how to advise or guide them. And so the CEO of this MSP got frustrated and he said, we're just going to tell everybody we can do it. And if we don't have the perfect offering, we don't have the perfect offering, but I'm done letting AI manage us. We have to manage AI.

And that is such a good statement.

Yeah. And that's where they're at. And that's where a lot of MSPs are right now. AI is managing them because their techs are answering questions. During their QBRs, they're offering free advice. Like, MSPs are being managed by AI in the current state of things, and that has to change.

Yep. Yeah, completely. And I found this number pretty interesting too. I want to help— I want you to unpack this one with me. There's this myth going around, and I don't know, maybe it's TBD. It could prove itself to be real, but there's this framing going around that AI is going to replace people's jobs. I, for one, fully support it. If AI would like my job, it can have it tomorrow. tomorrow. But I think this has got a lot of people worried. And the RAMP data actually spoke a little differently, and I thought it was super interesting how RAMP approached this and how RAMP kind of talked about this. Interested to hear your take on this, but what they're saying here is the long-tail effect on headcount with AI is the more that someone adopts AI, the more people that they hire. In fact, high-intensity AI companies grow their headcount at least 10% after 2 years, over 2 years following heavy AI adoption. And this just to me was, it goes against the narrative that the news is spinning that all of us are going to lose our jobs. Interested to hear your take on it and how you unpack these numbers because Ramp a little critically didn't give us a ton of data on how they unpack this, but I have some ideas.

Yeah, I'm wanting to read— I'm reading a little bit of the data before I just answer. My reaction is this. If you've ever met a company that is truly adopting AI and using it to run their business, Alex, I said this backstage before we jumped on. I said I've never felt like I've moved more fast and I've never felt more overwhelmed in my life. My ability to solve and create customer value is at an all-time high. I can create customer value insanely fast compared to 10 years ago when I built my first SaaS company and 5 years ago when I built my second SaaS company.

Yeah.

Thus, my ability to create value for my clients is very extraordinarily high. Thus, I need to support all of the growth going on. Thus, I need more AI and more headcount combined. When you think about like this right here, high AI adoption, I guarantee they're using that 1% or 10% spend number as their metric here.

Looks like it.

My assumption is this. Yes, headcount growth is going up for companies that are using AI because they have an impeccable ability to create customer value at breakneck speed, which customer value is the whole world of currency we are in in B2B anyway. So of course, if I can create more customer value, I'm going to need to grow. Again, secondarily to that, like as I create more customer value and have more need to grow, I spend more money on AI to create and automate more internal efficiencies to create more customer value. So it becomes this like, it's this AI flywheel. And the last, and the big thing you're seeing if you're looking at these high heavy count growth companies, I guarantee if I look into those, the big thing that's missing right now for those kind of companies and growing is people in like success and enablement and all like in delivery of like true human-to-human value. So I bet a lot of those hires, not only are they growing more, but the roles they're hiring for are higher human touch.

Yeah.

Whereas the companies that are stagnant, the reason isn't because they haven't adopted AI, it's just because they haven't joined the AI flywheel yet that gives them this hyperbolic growth. Like it's an interesting time, but there is the reason everybody says that has adopted AI aggressively, you have to get started because It's the new way of doing business. It's not even an option anymore.

Nope. You know, inside of our Slack now we have a robot that when somebody reports a problem, the robot goes and reads Empath code and comes back and says, hey, I can reproduce that problem. Or, you know, I can't reproduce that problem. Your user probably has something weird going on in their web browser. And just little things like that make my frontline support team able to get responses back to customers infinitely faster. And so if I take that and give that level of efficiency to somebody like you, instead of you looking around around and going, I'm going to fire everybody and run the company all by myself. You start looking around and saying, man, I can grow this thing way, way faster if I can be 10% more efficient and I can hire 5 other people and make each of them 5 times more efficient. Then I'm going to go hire some people and make them super efficient and spend money to grow my company, not just ridiculously shrink my company and go, well, solo founder, I'm going to chase $100 million as a solo founder. Like, yeah, the most miserable sounding thing in the world.

Here's where this whole paradigm started. It's because you go— and I'm not trying to get like tinfoily hat here.

So, conspiracy?

No, you know me, I'm too straight and narrow for all that. You aren't reading news articles about startups cutting headcount with AI. You're not reading news articles about companies scaling and cutting. What you're reading is Fortune 500 cutting because they already had a lot of bloat anyway. But the reality is they're also at a different side of company. They're in a PE, they're in the PE machine at this point, in the publicly backed version of a PE machine for the public companies. And so yes, cutting inefficiency is a predominant driver, but I guarantee you where they're cutting, they're also reinvesting and creating new headcount in new categorical areas, which is why I said it's the new way of running a business. You can't run a business in 2027, I'll even say in 2026, you can't create a new business. And not just use all the AI stuff that's there. So I don't know.

I don't remember who they were. There was a European company that fired all their partners or customer success people and were like, we're just going to do all this with AI. It failed miserably and they went and hired back all their customer success people. Like, you know, you're seeing some of that. But the other thing that we see that we probably give too much credence to is the noise out of the AI factories, right? You've got Microsoft and Google and Anthropic and ChatGPT out here going, hey, we're going to replace everybody's jobs. One person is going to be able to build a business like they have all their marketing hype that these products and tools just— it doesn't land in the real world. It doesn't really apply in the real world. The last thing I'll say on this slide before we move forward though is if I'm an MSP looking at this, there is one thing I want to take away from this. If I enable my customers with AI and really get them going and they come with me on this journey, the folks who are running around telling me my seat count is going to go down and my customers are going to have less employees, and my managed services business is going to be in the tank, man, I'm going to look at them a little bit, a little bit more suspect because now I have some numbers that say if I enable my customers with AI, they will have a better and healthier business and they will grow and thus will grow their managed services business with me. And that's my takeaway from this, is that the thing we've been worried about all this time and folks talking about, we got to get to outcome-based billing because seats are going away, man, are we— we are the furthest thing in the world away from that, at least as far as base managed services goes.

Yes, and I have 2 points. First off, hey, you gave me a really big aha there, so thank you. That was a very good learning. You're gonna hear me say Alex said this and he's really smart, you know. But I wanted to introduce 2 points to that before we jumped on. One, I'm working with an MSP based in Australia, brilliant MSP. They went to every one of their clients and said, hey, Your per seat cost is not going down this year because we entered— period. Your per seat cost is not going down this year. I want to give you some facts. We introduced ticket triage AI and we introduced support AI that's now automating a lot of our ticket and help desk. We are now introducing AI as a service offering as part of your managed service agreement moving forward. We're increasing your per-seat cost by 10% because of this, but we will be doing less ticket support and more business transformation. And they went to all their customers on renewal and said that, and they defended it. And they didn't have—

I love that they're being proactive about it.

Yeah, I've met 2 MSPs who have done it and they've had phenomenal success in resonating with customers because nobody hires the MSP for the support desk. I mean, they do, but like—

But that's the commodity.

Yeah, it's the commodity.

So if you can— It's electricity. I can't live without it.

Yeah.

But I'm not going to, I'm not shopping on that commodity. I'm shopping on the other values that you bring.

Yeah. And so the other thing you said is you challenged outcome-based pricing. And I agree with that. I've been saying that since the beginning. I said outcome-based pricing is pure CEO talk.

It might fly in the enterprise, but it is really hard to bring something like that downmarket.

I don't even think it really flies in the enterprise, man. It's like pure like CEO theory. Like, it's not real. Like, if I go to you and say I can create, you know, $100,000 in efficiency and it takes me 20 hours to create this agent, you're very quickly going to go, great, I'm going to go shop this till I get to the bottom dollar and it's going to go back to per hour. It's going to go back to—

Like, we have outcome-based pricing, John. It's called commission for a sales guy.

Yeah, it's just nonsense. Like, yeah, I'm sorry. It's just going to margin compress immediately as the supply and demand curve balances out here. So don't be ridiculous.

I'm glad to hear somebody smart agree with me on that one because I always thought it was obnoxious and silly. Well, think about it. CEO theory, I'm stealing that one.

Yeah, just like think about it. If you say, okay, dear Mr. Business Customer, 20 hours of effort, $100,000 in efficiency, you're going to give me $20,000. Say they say yes. How pissed is that customer? Sorry, am I allowed to cuss? I guess I am.

You can say whatever you want on here.

How pissed is that customer going to be in a year when it becomes a feature of Claude and they spent $100,000? So I've been telling people, Like, if you're going to build agents and automations for your customers, you need to clearly communicate to your customer that they are leasing a year of innovation at this price tag. And if you put $100,000— uh, thank you, Tim— if you put a $100,000 price tag on it, they're going to churn so quick on you next year for overcharging them.

I agree.

Then I come back to my whole— I've jumped around a lot of points, but I'm still touching the dots here. You said outcome-based pricing, and I'm really glad you said that because I disagree aggressively with outcome-based pricing. But I've had an aha point, which is we won't have outcome-based pricing, but we will have outcome-based selling.

Well, and we'll have outcome-based loyalty, right? That is the thing. If I deliver outcomes, that customer's staying, they're spending more money with me, they're going to become incredibly loyal to me. if we deliver outcomes, which MSPs have traditionally been kind of bad at. But this is a place where we can really shine and deliver that outcome and build a loyal customer.

Yeah, the whole world translates on value creation and currency exchange, right? Like, so if I create more value and you'll give me more dollars, it's pretty simple. But I want to— I agree with the outcome-based— you said trust.

Loyalty.

Loyalty, yeah, similar. I was going to outcome-based selling because I think it's really important. There's a distinction here between all 3 of these different things. Outcome-based pricing being how you bill, outcome-based loyalty being the output of what you're offering. Outcome-based selling is really important because when people hear output— sorry, outcome-based selling, I think they infer outcome-based pricing. Meaning they hear that and they go, oh, I gotta sell at an outcome base. It's not that. It's selling outcomes to the business that are different for each client and relevant to their business and their line of business. It's not charging them an arm and a leg for a small work job, but it's selling them outcomes that drive their business forward. Whereas before, the managed service provider industry was selling insurance, for lack of a better term. Like you were just selling You know, you don't want— your company can't go down for a day, it'll cost $100,000, so you're going to pay me $10,000 a month on insurance.

Yep.

Now we're selling outcomes, which is your business could be this chart here with AI, and we want to help you get there. And so this goes back to one last final thing. I know I say keep it final things, but this is actually a really big thing that I think people need to reprogram, and the sales frontline Because for the last five or six years, we've been selling on scarcity and fear. We've been selling to our customers. Yeah, FUD. You know, we've been saying to our customers, "You're going to get hacked, and your business is going to go down, and all your data is going to be risked, and you better buy my threat locker thing." Risk, risk, risk.

Give me money, right? That has been the conversation for the last five years.

And so we're trying to take that square peg and just shove it down our customers' throat by saying. Oh, if this comes back to my gripe from earlier, oh, if you put your data in that model, it's all going to get exploited and you're— no, that's crazy talk. A, it's infinite, like Doctor Who, Marvel Universe, rare multiverse scenario that happens. But B, you're just taking the old way of selling and trying to shove it into the new square peg hole. I don't know where I'm going with this, but you get my point. But here is the beauty of it. FUD sucked to sell. There is far more FOMO than there ever was FUD. And the FOMO is a hell of a lot more fun to sell than the FUD was. Like—

There's a slide I expect you to have in the Atlanta deck is FOMO greater than FUD.

Yeah. Oh, it's there. It's probably there. I have a few random stump speeches I go on, but yes.

And if you guys haven't heard, John is going to be joining myself and Wes in Atlanta for a workshop on basically You know, opening the door with AI and how to get in and talk to new customers, new prospects, and what to do with your current ones. And John's going to be joining us to teach a big part of that day. So we're super excited to have him for that. There's my pitch for that.

You did. Yeah, it's just, it's such an opportunity right now to sell. All you've got, all you gotta do is—

I want to touch on opportunity again.

Just jump on this.

No, no. Well, I want to follow the opportunity because that's That's really the right conversation, right? So now we're talking about on that last slide, if you were really paying attention, what we see is that when these folks start to adopt AI, headcount grows and we get to sell our traditional managed services, which if we're good at it, we make, you know, 16 points of EBITDA on. We really get to take home money on this. It's not a project that we have to learn how to scope and do anything crazy with. Like, this is just our moneymaker as an MSP when headcount goes up. And what we see is that small businesses are less likely to adopt AI in the first place. I see that as they don't typically have super technical resources in-house. They don't have an internal nerd that they can hand this to. They're not Avik who bought John's SaaS business and has a genius in-house that they can hand this to and go, John, figure out AI and teach our business how to AI. Small businesses are much more technology averse, or at least risk technology averse. But when they do adopt AI, they use it more intensely because it's really easy for me to take AI and buy 10 agents for 10 people in a 10-person office and really dive into it in my business. With a big business, man, it's hard to scale that across 2,000, 3,000, 20,000 employees. But a small business that adopts AI, man, it just goes like this. And once they get on that, once they get a little taste, they get a little dose, they just run with it. And there's so much opportunity for project and for AI-focused managed services here that I don't think we've realized once we set the hook that there's just a huge amount of opportunity for us as MSPs here.

Yeah, I love this. The reason that's happening, that curve is occurring, in my opinion, is because there's just less guardrails, there's less compliance, less oversight, which means they can run faster, create value faster. I don't know. I would actually really love to see this juxtaposition against created value because there's a lot of AI, the 2-letter, the 2-capital-letter AI things happening that aren't really creating value.

Yeah. That's one of those things that is in my research has been really hard to report on. It's really hard to track value created by AI considering the fact that it's baked into your Microsoft stuff, it's baked into Notion, it's baked into Asana, baked into HubSpot. Like, figure out what value came from the AI thing and how do you report on that in a business that is 9 employees and doesn't do any real reporting anyway?

My best analogy here is it's like trying to defend spending money on Microsoft Excel. Like, why would you even go do it? Like, what— where are you trying to get down to the brass tacks on if Excel is doing ROI or not? Like, obviously it is. It's like part of doing your job. It is the modern toolkit.

I'm less worried about the AI being able to say $6 of margin came from AI and more interested in being able to say my gross margin went up from adopting AI, right? All I want, all I want MSPs to think about is like track your EBITDA, track your gross margin, track your growth rate. If you're familiar with Rule of 40, like go do some Rule of 40 math, which is, you know, growth and margin at the same time. Right. Go combine those 2 if you're hitting near 40%. Like, hey, if we can get to 45%, 50% after we do AI, maybe we're starting to make some headway here.

I'm actually— I mean, I agree with you as like a totality of a statement, but I'm going to have like a different take on this. Good.

If we all agreed, it wouldn't be any fun.

Yeah, you know me, I'm very opinionated. But you said focus on gross margin. I actually have them and I agree there's opportunity there. You should focus on all areas of the business. But I'm just Zoning in because you said that. My mentor gave me a really good piece of advice, and I think it translates to where we're heading here. And then I'll come back to the chart. And it's why PE companies suck. So now I'm going to go down a random rabbit hole that's going to come back and be a good sales strategy, I promise. He said a good company spends 60 to 70% of their time creating customer value, 20% of their time focusing on internal operations to scale customer value, and 10% of their time on FP&A. Okay?

Okay.

A private equity-backed company spends 60% of their time on FP&A and internal operations and 20% of their time on customer value.

Yeah.

So And the reason I'm just going to push back right now on your gross margin conversation isn't because it's not right. I don't think it's right right now. And what I mean by that is if you are an AI-forward company, you can go back and fix gross margins later with AI. You can increase operations, but 70% of your investment on AI, in my opinion, should be focused on customer value. And then 20% on internal operations, then 10% on FP&A.

No, I think we agree on that. What I would say on gross margin is you're not going to measure it every day.

Yes.

But I want to see we did an AI initiative in 2026, and by 2027 or 2028, we're able to show a couple of points of additional gross margin and hopefully growth to go with it. I'm not saying AI should create gross margin tomorrow, but it should create a world in which You can do the same projects over and over. You can build some repeatable things. You can execute your business more efficiently because ultimately there's no reason to do it if I can't make more money at it.

Correct. And I'm saying the rule of 40 is 2 inputs. And I'm arguing if you're going to go heavy on AI, focus on the growth input right now because you have a monster advantage compared to the people that are not. And then by the time they focus on the growth input, you're over here maximizing your gross margin output.

Look, Rule of 40 applies everywhere, right? I could be negative 5% margin and 45% growth and I'm still happy. You know, those are— that rule is kind of built to handle and manage that. And we may want to lean that direction for the next 18 months while we figure all this out.

And we've skipped like— we're way out of practice.

Way out of it. So I wanted to pull up this slide. This has been my favorite slide since 2021 when I found it. I stole this out of a Forrester research paper. As MSPs, we typically tend to think about growing our business as going out and acquiring new customers. Kyle does a wonderful talk where he talks about the cost of acquiring a new customer for an MSP, and on average it's like $28,000. It's an insane number. It's a 25-month payback at most MSPs. It takes you forever to make any money on it, and the average customer only stays around for like 37 months. So there is a significant cost to going out and getting new business, and specifically where AI is considered, Yep. I want everybody on this call to think about the fact that Forrester does a really good job of calling out— they call prospects candidates. Current customers retention, upsell, and cross-sell through introducing AI to them is a huge opportunity for you that might even be 70%, 75% of the revenue you can generate with AI this year might be revenue you generate inside your current client base. So we want to make sure everybody's thinking about it from that perspective. If you're on this call, and you're an outside sales rep, great, I get it. Your focus is only this 25% over here. If you're on this call and you're a VCIO, you're an account manager, you're an owner, you're a COO, man, we got to think about what we can do for our existing customers, our existing partners. And John and I have talked about it a lot. But also, don't forget, there's a bunch of new business out there for the MSP that doesn't adopt AI fast enough. Man, are they going to be able to— are you going to be able to go in and kind of drive that wedge? And start to get them talking about, hey, well, John's going to do something for me that Alex didn't. I should entertain a conversation with John that I might not have entertained last year.

Downstream in our design partnerships right now, we're actually seeing more revenue, net new revenue be created than expansion. And it's because the assumption is our existing customers want this and will pay for this. The reality is your existing customers assume you're already doing it, as they always have. Like, there's a tale as old as time. But the net new is an absolute rush, right?

Because you can find people who are actively looking for it?

Yeah, because the supply and demand curve is dramatically uneven right now. Like the quote unquote AI experts, and everybody bashes them and I agree with it, nobody's an AI expert. But the reason they're enjoying the spoils of war and rich right now is because they're out there touting themselves as AI experts and the supply for AI experts is so low and the demand for AI expertise is so high. And so I'm not saying go sell vapor. Well, I'm just saying the reason the net new is so strong is because people who would have never considered an MSP prior because it was just like a de-risk thing may consider a growth conversation and the growth conversations. Once you land in like growth and sales and marketing, you're always a bigger wallet than operations, risk, and finance. Like, there's operations and risk in the finance side.

It's, you know, there's a rule here in the MSP space. If you're building a software product for MSPs, like they buy for only a couple of reasons and 2 of them are make money and save money, right? And there's only 5 of them. So half of them revolve around money. And I think that's incredibly interesting and it does carry out to the wider business world. John, We are running up on time here and I'm going to play some rapid-fire Q&A here so that we get through everything. I'm just going to cherry-pick a couple of questions that I want you to answer. You went through and talked to over 100 MSPs as you started building Lemhi. Well, let's take a step back. What is Lemhi? Why don't you take a second and just give us a high-level overview of what you guys are doing at Lemhi, and then I'll pick your brain about what you learned as you've taken this company to market.

Yeah, so Lemhi is a business helping every managed service provider introduce repeatable process programs and platforms for their customers when it comes to creating AI value. Most people don't have time to build something, so they don't know where to start, so we give them programs. What ends up happening is once you have a program, your subject matter expertise gets super constrained when you're trying to sell to your customers, so we give them a platform to where anybody can sound like a SME and have an AI conversation. And then even if they're like having conversations, they're often not actually returning revenue. So we give them a process to actually have a repeatable revenue funnel. So we give them all the things they need to really get started with an AI practice today.

It's like you're touching on all the things that made Lifecycle Insights successful, right? We gave you a process to keep you from swerving too far. We enabled you to go have the conversation. Yeah, it's a great business model. I absolutely love it. Full transparency, I'm a huge fan of what John's doing over at Lemhi. There's no doubt that I'm a fanboy. So my opinions biased. But you went and talked to hundreds of MSPs as you were kind of figuring out what you were going to do next. And I'm just curious, like, what was your biggest surprise and what was something you found that you really weren't expecting?

The biggest surprise to me was probably the thing I shared earlier around— this is talking to the— let me flip it on 2 sides. There's 100 conversations before, and now I've had hundreds. I don't know how many at this point post my research.

Yeah.

The before surprise to me was, um, it goes back to the same analogy earlier. If you're on the edge and you're trying to chase the edge, you feel like you're really far from the edge. But like when you get down to everybody in the world, like it's not moving that fast. There's not a lot of people really making money on AI. Don't let like all the FOMO and FUD like scare you into building a program and a practice. Like you really should build one, but like you're not gonna like collapse next year if you didn't. You're just going to have a lot of work to catch up. So that's what kind of shocked me was like there's not really a lot of money being created, but I think it goes to 3 of the core reasons I established the business. Why is preventing it? Then I'll go into what I found most surprising post working with partners, which is what I said earlier. It's the net new conversation is an absolute tear. Like there's so much revenue on the net new side to be made, and if you're AEs or your BDRs are out there smiling and dialing and trying to get the local dentist to show up to a meeting and you're not talking about AI, the other MSP that is, is just crushing you. So it's just switching the MSP pitch has to happen. So those are the 2 different sides I saw with the most surprise, I suppose.

That's awesome and I love that. And MSPs have been, you know, for decades trying to find the thing that will help them make the next dollar and that will make sales easier. And we've said for a long time, this is the, this is the thing that is the, you know, the pull motion that they're going to come to you. So we talked about shadow AI, we mentioned kind of, you know, you kind of built your reputation and your brand here in the industry on shadow IT and how to get away from that. I hate the phrase shadow AI, but I know there's some similarities to the problem you were solving at SaaSlio. And as you're kind of building Lemhi, just talk about the similarities that you see And where you think the opportunity is there?

Yeah, the—

I think you already kind of touched on this, right?

No, this is an introspective question. I want to draw some parallels here. Like, well, I don't think I've coined a new term for our industry that nobody's heard yet at Lemhi. I think I've coined some practices. And I think what some of the new— some of the things that I think I'm preaching today will ultimately become what everybody does in a few years. I firmly believe we're within that 80% barometer of right right now. And so I'll give that piece of advice away for everybody. Like, one of the easiest spines of revenue creation when it comes to AI is 4 steps. It's coming into your client and helping them make a tool of choice. So when I say tool of choice, Copilot, Claude, OpenAI, SMEs get another tool or whatever it is, but help them choose their predominant generative AI tool. Secure the tool of choice. So if you're going to go Copilot, you need to do data governance and Microsoft 365 readiness. If your Claude's going to integrate with Microsoft, then you need to do the same task. What the work that we're actually really good at, the thing we're actually trying to sell, by the way, this is where everybody's making this. I don't even know. I can just— that's not the middle finger, but just try to put that finger up like people are coming over here and skipping like, let's understand what values you need to create. So they're skipping tool choice and they're saying, let's just secure. The reason you do tool of choice first is because if you go into tool of choice, you get to the use cases and you get to the ROI, and ROI is in the front of the sale. Then you go tool secure, then you go tool integrate. This is the lowest hanging AI revenue you can create right now. If your customers are using Claude or Copilot, what SaaS softwares are in their business that you could just set up MCP connectivity to? That sounds rocket science. I promise it's not. It's setting it up. So simple like that. Yeah, it's an API integration. Just keep it simple. But it's called MCP. And then tool enable. And I think this is the biggest opportunity in our space right now. You'll hear me say it this whole year. It'll probably change next year. Tool enable. The difference between me getting 100x out of AI and selling accounting getting 1x out of AI is not tools, security, or integrations. It's capability. I am highly enabled when it comes to AI. And so I get 100 times ROI out of AI. If you as the MSP can help your customer understand tool of choice, secure it, integrate it, then teach them how to use it and create value for it, you will, going back to our point earlier, create such an out-leveled value of an outweighed value creation for your customer that they will have forever loyalty to you because you help them through the transformation. So that spine is repeatable. It's not rocket science. I think everybody should be doing that right now. It's just you need to learn how to have the sales conversation and have a repeatable process around it.

I love that. And so with that, we're gonna skip a couple of slides because I want to get to the end of this. No, no, you're good. You're good. We've covered a lot and it's been really good and I can't wait to post this in Empath and drive some folks to it. But the 2 questions that I gotta ask because, you know, I've gotta be, gotta be. What is, what is John's favorite AI? What's your AI of choice?

It's Claude, 100%.

Okay.

Though I will say I've been looking into Descript a lot lately.

Really?

So add that into your new stack of tools.

It's one I haven't tinkered with.

Okay.

All right. And Then the last question I want to ask is, what's the most interesting use case that you've, that you've come up with with an MSP? What's the most interesting actual piece of enablement that an MSP has taken to their customers and been really, really successful with?

It's going to be boring because you know me, I love boring.

I love it. There's a lot of money. There's a lot of money in picking up trash cans, man. You don't have to do those crazy complicated things.

It's not even interesting. Can I do a quick stump speech or do I have time for that?

Do it. Yeah, we'll make time.

I'll answer your question. It's the chief of staff agent. It's the world's most simple, universal, applicable thing that everybody in the world can use.

Even I can do it and I, and I love it and it helps. But yesterday morning it went, hey, it looks like you're double booked on your calendar. And I went, sure as hell I am. Like, and I realized that at 7:45 instead of at 11:30.

Sit, sit on a sales call with your client, help them set up a chief of staff agent. It's not even that hard. It sounds hard. Just go find a YouTube video. You can learn in 10 minutes.

Yep.

You will be considered an AI expert to them, and they will open their wallets and let you sell them what they need. With that said, I'm going to do one last stump speech. The chief of staff agent is probably the most universally easy, applicable, value-creating agent that every client in the world can create. Yet we as an industry, for some reason, are going in and choosing some 30-step fragile automation and agentic workflow that we want to sell them.

Because we're nerds, John. We like to do fun nerdy stuff.

The last thing is I'm going to put it on a scale is training and enablement. It's agentic. I made an MSP have a debate the other day. I said, you get to choose one of the two. You get agentic triage or you get training and enablement. And I count training and enablement as teaching them how to build a chief of staff agent. I say you only get one. And they said, I want both. And I said, why do you want both? He said, they're equal. I said, if they're equal, why are you spending 40 hours building one when this could take 4 hours to train a whole company and it's repeatable by anybody and you're the only person that can do this one? One is repeatable, scalable, resellable, and creates immediate value. And the other one's fragile and going to fall apart in 3 months. Why would you choose the left-hand side?

I'm going to give you the one that came out of nowhere that I didn't never thought about until Kyle built it here at Empath. He built a marketing brand book and handed it out to the whole company. And now when Claude ties into Outlook or into Word or Excel or PowerPoint, we just go brand this like Empath and it goes and grabs our brand book, uses our fonts, our colors, you know, looks at the full brand book. It puts things in exactly the right places and it can do all of that for us. And now it doesn't matter if I make a piece of content Or if Joey, who's incredibly talented and actually understands what a brand is, does the piece of talent, we can both make them not look shitty. And they've got my company colors and my company branding and all of that on them. And it was super simple.

We have a, we have a business process here internally. It's a SharePoint drive called Product Context, but we basically refer to it as golden documents. We have a SharePoint site where no AI is allowed to get dropped into it. It's all hand-rolled assets that we've created that are core to our brand. cornerstones of the way that we talk to businesses. And every time that anything gets authored, it references that SharePoint site to pull context in. So it always comes from the voice of us and not AI eating its own tail.

That's a really interesting thought. I'm ashamed that that made it to the last comment that's going to be made on this conversation because I would like to dive into that. But we'll have to save that one for another time, John, because that's a really great piece of advice for everybody. So if you guys aren't connected with me, please make sure that you do. But I think that you all are. September 1st, we'll be back and doing this again. September 1st, I'm going to have— actually, that is not the next one of these. That is the launch for our product called Grow. And if you know anything about Empath, you know that we're all about education. John has railed on education today and I have stayed away from touting this, but on September 1st, Empath launches AI enablement training for your end customer that you can take to them, sell them, and make money at it. So if you could make money at the one thing that makes your job easy, that is really what we're talking about. September 1st, we will launch that. September 10th, we're going to have another conversation about it. And then on September 29th, Wes Spencer will be joining me here to have some more conversations about building your MSP's AI go-to-market. We'll bring you some concepts and ideas and plans around some of those cool ideas that you can bring to your customer. We'll have a couple of easy ideas that you can build and take and turn into repeatable process for your customer. So, lots of good, cool, exciting things coming. If you haven't already connected with John and checked out LimHi, please do. You know, guys like him, I appreciate that he will come here, have this conversation with me today, not really sales pitch, and just lift everybody up and make everybody smarter. So, John, thanks for hanging out today, buddy. I always appreciate you.

Cheers. Thanks for the time.

Take care.