Episode 125: Impostor Syndrome

 

Erick and Rich discuss why the rapid emergence of usage-based AI pricing schemes is going to make “token economics” and AI spend management essential skills for MSPs as well as the importance of giving everyone inside an MSP an understanding of big picture business performance. Then they’re joined by John Harden to discuss client-facing AI service opportunities and how his new app Lemhi aims to help MSPs seize them. And finally, one last thing: The most (phones) and least (dentures with two teeth) common items in Uber’s lost and found.

 

Discussed in this episode:

AI Model Spending is a Black Hole

More reasons to get smart about token spending sooner versus later

Lemhi Launches the SaaS Platform MSPs Need to Deliver AI as a Recurring Managed Service

Uber Lost & Found Index includes breast milk, ankle monitor

 

Some guests on this podcast are clients of Channel Mastered. Compensation plays no part in their appearance or the content of the discussion unless the episode they appear on is a “bonus episode” explicitly labeled as sponsored.

 

Transcript:

Rich: [00:00:00] This episode of MSP Chat is brought to you by MSP Mastered. If you like co-host Erick Simpson’s Tip of the Week, you’ll love the comprehensive growth advice Erick and his team provide at MSP Mastered, your go-to resource for overcoming business challenges, improving service efficiencies, selling more profitable MRR agreements, and increasing the value of your MSP business.

From sales and marketing to service delivery, to hiring and retaining high-performing talent, MSP Mastered offers access to over 90 online master classes, 150 on-demand webinars, and 250 advanced MSP tools resources, along with regular group coaching sessions and unlimited strategic email support, all for one all-inclusive membership fee.

Unlock your true MSP [00:01:00] potential by joining MSP Mastered today.

And three, two, one, Blastoff, ladies and gentlemen. Welcome to another episode of the MSP Chat Podcast, your weekly visit with two talking heads talking with you about the services, strategies, and success tips you need to make it big in managed services.

My name is Rich Freeman. I am chief analyst at Channel Mastered, the organization responsible for this show. I am joined as I am every week, virtually this time, by your other co-host, our CEO and chief strategist at Channel Mastered. His name is Erick Simpson. Erick, I think I know the answer to this question, but how you doing?

Erick: I’m fighting off a head cold, Rich. I’m fighting off a head cold, and I think you have a little bit of a cold you’re trying to overcome as well. But as we like to say, the show must go on, so we’re gonna power through, aren’t we, Rich?

Speaker 2: We are. We are. And I j- I just wanna note for regular listeners we began the last episode by noting that the show went dark on you for two weeks because about a month ago I was in New York, came down with [00:02:00] a bad cold.

We had to cancel two recording sessions last minute. Then we came back, and somehow or another, Erick, just a few weeks later, I’ve got a cold again, and even though you and I are 1,000 miles apart, I’ve given it to you somehow.

Erick: I know. We’re simpatico that way somehow, through the ether. You’ve… I feel what you feel.

You feel what I feel.

Rich: Indeed. Indeed. It’s a beautiful thing people, except when what we are both feeling is under the weather. But we’ve got plenty for you on this episode of the show anyway, starting with our story of the week, so let’s dive into that. And this I’ll just preface it by saying I, I’ve got more work for the folks in our audience to do here and they’re not necessarily going to thank us for that.

But I do think this is an important topic, and the topic at hand basically is what I loosely refer to as token economics. And this is a topic that it’s in the news particularly in the last week this issue has been much discussed. I’m sure we’ve touched on it on the podcast. And it’s just the idea that y- [00:03:00] already today, and even more so to an increasing degree AI services are consumption priced.

And I say to an increasing degre- degree, we are this episode airs on June 5th. We’re recording it June 2nd. On June 1st, yesterday Microsoft changed all of the pricing around GitHub to be usage-based, and this is gonna be a big adjustment for a lot of developers out there. I think the big picture story is that over time the companies responsible for these models and these tools are going to have to kinda cover their variable costs by charging on a consumption basis.

And that’s gonna put MSPs in an interesting position, something we’ve talked about a little bit here before, which is just you’re gonna be paying varying rates and in varying quantities upstream to the AI companies that you’re consuming from. Your customers may not be entirely comfortable with paying you for the AI-based services you give them that way.

But to the degree that they’re gonna be billed this way if not by you then by [00:04:00] everyone else the art of understanding and mastering token economics is something that I think MSPs are going to be required basically to have some knowledge of to get better at. I’ll give you three little bits of information just to maybe set this up a little bit.

Uber has been in the news a lot recently for this fact that they burned through their entire AI coding budget in the first four months of the year. So that’s been reported a lot. What’s been reported less, and I found very interesting, is that today 70% of the code that Uber commits originates with AI, and 11% of their updates are created entirely by AI with no human in the loop at all.

So when you hear about Uber saying, “We burned through the year’s budget in four months,” that’s not some token maxing thing where there’s a leaderboard and people are trying to get on top. This is the AI coding tools doing what they’re supposed to do and delivering value. And the complicated question for a company [00:05:00] like Uber or for your clients as an MSP is how do I measure the return on that investment?

How do I optimize my spend so that I’m not sacrificing that return, but I’m also not overspending? I also have some degree of sanity in terms of what I budget and when I’m active. They’re gonna not– Your customers are gonna need help with that. You’re gonna need to be in a position to give it to them.

Recent research results that I came across, a company called Digital Root asked a whole bunch of businesses and found that only 8% of organizations out there are extremely confident they understand what delivering AI services to users and customers costs them. They’re spending money on AI.

They don’t feel terribly confident whether or not they really understand how much they’re spending or why they’re spending that amount, and only 23% of- rate themselves as highly accurate at forecasting AI-related usage cost and revenue fluctuations. And obviously being highly inaccurate at that can have [00:06:00] giant bottom line implications for a business.

So again, another reason why your customers are gonna want and need and probably start asking you for help on predicting these things and managing these things. And I’ll add on one other piece. Part of what makes this whole issue so complicated is it’s not just that it’s very difficult to predict in advance how many tokens you’re going to consume in a given month, and it’s not even that the price of those tokens is like a moving target.

The model makers are adjusting their pricing schemes all the time. They, excuse me, they roll out a new model with different prices attached to those tokens. All of that is true, but then you start thinking about all of the different variables and complexities here. So I’ve got the Claude desktop app on my laptop.

It’s open most of the time. I go in there to to do something or other, and Claude just defaults me to the 4.8 model, which is the newest one. And that’s [00:07:00] fine for me given that my, token consumption rate. But to use the most advanced, most expensive model Anthropic has for relatively simple tasks doesn’t make a ton of sense for end users that are really concerned on their AI spending.

Maybe I could go back a generation or two generations. For this workload, I need 4.8. For this other one, I don’t. For this third workload over here, I’m probably better off using Gemini or ChatGPT. Maybe for this giant workload over there in the corner, the best move to make is to go with something open source.

The ability to calculate all this and to help end users understand predict, forecast, and budget for AI consumption is gonna be part of the whole AI solution and service story for MSPs. And like I said it’s complicated stuff. It’s new to all of us. Very few businesses are confident or feel like they know how to do it [00:08:00] and their trusted advisors I would wager are in the same boat.

But like I said, more work for people to do Erick. I think people need to start educating themselves a little bit around costs and spending and cost management for AI sooner versus later

Erick: I feel your pain, Rich. I feel your pain, ’cause I’m jumping around between models myself, and then, what’s the most frustrating thing because I’m not, and some of these LLMs will, like Claude Cowork, for example, right? We use it at Channel Master and MSP Mastered, and you can– it’ll run out of tokens in the middle of, a coding s- a vibe coding session, and it drives me nuts. It’s “Oh, you’ll be able to log back in in three hours and finish this.”

By that time, the memory’s got to be rebuilt and the whole thing. It’s a nightmare. But I’m not confident yet enough to understand the spend because, you do have the option to toggle on keep going, and it will charge you on, on that [00:09:00] burn, on that token burn.

And I am not comfortable yet doing that, so I’m alternating cons- constantly between two or three different LLMs, including ChatGPT, Copilot, and Claude. And like you, I’m evaluating the different the legacy LLMs to try to bring some of that cost down to see how long I can go. But it is– And I think it’s a challenge that occurs when a user has gone beyond what a flat fee typical subscription Delivers for their workload.

So for instance, Rich, if we position this in a different way, we could say, “Hey, there are gonna be some end users in your customers’ environments that are fine with the, one or two or three subscription models of one of these LLMs.” But then you’re gonna have some power users that, like us, Rich, we will be, h- hitting our head against the ceiling of some of these pre-built kind [00:10:00] of, batches of tokens.

And what makes it even more challenging is it’s a mystery of how to figure out how much token burn you’re going to spend based upon the subscription model you’re in. Because I’ve done a little research, I’m sure you have too, Rich. It’s they don’t really tell you. They just tell you’re gonna get double the amount of tokens than you have in this, but you don’t know, and you don’t know how those tokens are applied, so it is challenging.

But I think for the majority of MSPs out there that are supporting customers with just their toe in the water of AI, some of the base subscriptions with, Copilot or Claude or ChatGPT can get them s- notable improvement in helping them with their daily jobs, their workloads, efficiency and productivity.

But then I think this now starts bleeding into the vCAIO role, where you’re now assessing the need of different users in an end customer’s organization to see [00:11:00] if, in fact, there, there needs to be a little bit more horsepower for some of these VIP users or somebody else. Or How, h or not? And then how do we then layer our services on top of that?

AI is a great opportunity. It’s changing the world. It’s a… It’s got a lot of promise in helping businesses and MSPs themselves improve their service delivery. But when you think about the cost versus the return and that whole formula, it gets really blurry really quickly.

Rich: I’ll say, Erick, on a hopeful note if you go back in time 10 years when it was all about cloud as opposed to being all about AI, there, there were similar issues that MSPs and their customers were dealing with in terms of anticipating budgeting for optimizing usage-based cloud services like AWS or Azure, something like that.

And over the course of time, tools came along that, cloud optimization, cloud spend [00:12:00] optimization kinds of tools came along. And I’m pretty confident that this will happen in the AI case as well. There are actually some tools of this kind out there now that there’s one I think called Vantage.

There, there aren’t a bunch of them. Most of them aren’t multi-tenant. I don’t know if any of them is really built specifically for MSPs, but I do think you’re gonna see those tools come along over time, and that will take some of the complexity out of this very difficult issue, and arm MSPs to make AI cost management a, a boardroom level service that they can bring to their customers.

Erick: Yeah. Wouldn’t it be great to have some sort of a dashboard? ‘Cause, instead of having the LLM itself determine ChatGPT will say, “Oh, I’m gonna automatically determine based on what you’ve prompted me, what’s the best model to use,” blah, blah, blah. Wouldn’t it be nice to have kind of a dashboard that, that is connected to all of the AI that you use?

And then as you’re prompting in that, then it determines whether not only what out- what version [00:13:00] of an LLM, but also what AI platform you’re using. “Oh, this is better for ChatGPT, this is better for Claude, this is better for Gemini,” and lets you, and I’ll bet there’s some stealth companies out there, building that right now because, ultimately that is probably the best way not only to give us an idea of what the best model to use for what we’re doing across these different platforms, ’cause I’ll tell you straight up, Rich, but also what it might cost us in burn.

But- There’s some things that I will always go to ChatGPT for, and there’s some things that I will always go to Claude for, and there are some things that I’ll do in b- in tandem and feed each other, right? And try to get the right answer. So there isn’t a silver bullet for this, right?

We’re still in a lot of experimental phase. And of course, the cheese is in motion. We- we’re getting new versions every couple of weeks with all of these different platforms, which makes it hard to benchmark what I, Some things that I never [00:14:00] had great success with in ChatGPT in the early days now is oh, it’s killing it on that, and vice versa.

It’s challenging to keep up with all the change and to create some sort of a process or a methodology based on what we know today, because six weeks from now it may all be different

Rich: As you say, it would be nice if an individual inside an organization could go to a dashboard and get the complete picture of their AI spend.

And it might be nice as an MSP if your technicians and your other employees could not only see their AI consumption, but actually see how that figures into the big picture of the company’s consumption and budget and how their individual efforts are impacting that. This is the best transition I could come up with, Erick, for your tip of the week.

Dive right in, please.

Erick: Thanks, Rich. Yeah, and I think it’s a good segue to surfacing [00:15:00] information for the team, your staff and things like that. We talk a lot, Rich, about guardrails around AI and what should be exposed to different team members and what should not be exposed depending upon, the business and the leadership and the vision and things like that.

We certainly don’t want to expose salaries and things like that, or any personally, PHI information, any of that stuff. But when we’re talking about to the team, but when we’re talking about business performance, I think it’s really important to surface key metrics to the team at large, s- especially when their compensation and incentives and rewards are tied to not only their performance, maybe their team’s performance, but maybe overall company performance, right?

So what we can do today to address this challenge that disconnected teams have in terms of feeling like they are part of the organism that is the organization. Connected teams [00:16:00] tend to perform better and tend to collaborate better internally when we’re surfacing performance metrics that everyone is incentivized to achieve.

And this goes all the way down to in the most basic sense, Rich, some MSPs may think, “Oh we don’t really have incentive programs and things like that.” But you do give out some sort of, Christmas bonuses and things like that based on different factors, right? I think that’s the very, entry-level kind of perception of, do I receive some benefit from the hard work I put in beyond my paycheck and what drives the needle for the organization?

Now, Rich, this definitely requires the company to be performing in a profitable manner in order to deliver some of these bonuses and incentives. But let’s assume that, MSPs are moving in that direction. What are some key metrics that an organization can publish and make available to the team- At any time, and certainly [00:17:00] be presenting them during all-hands meetings and things like that to show the organization the path of the organ- that show the team the path that the organization is on, and how their individual or group or team contributions help fuel that.

So start with just a couple gross profit. I think that there’s a huge opportunity to gain buy-in and collaboration when we are setting targets for increasing an organization’s gross growth pr- gross profit over time, and how each team might be contributing to that gross profit number, and incentivizing and rewarding teams and individual team members to help achieve that.

And you can see this happen across different business units, which is not just the service delivery department that helps drive gross profit, it’s also, the sales and sales engineering teams, where we’re making sure that we’re pricing and bundling our services to man- to hit those GP numbers.

And then when we’re delivering these kinds of [00:18:00] metrics and connecting teams together to see how each of them participate as a unit to achieve these performance targets, we’re starting to see teams start to work better together and to not, feel like, “Oh, you’re asking me to do something that’s, outside my wheelhouse,” or, “I’ll put that, down my to-do list.”

If I’m an engineer or a sales engineer and a sales pro- a s- one of the sales team comes to me and says, “Hey, I need help pricing this proposal,” and we’ve got a GP margin hit, I as a sales engineer know that I have maybe some insight that could contribute to assisting that sales team or that sales person to make sure that when we are pricing that proposal, we are at least targeting that GP number for the overall growth goal of the company.

On a quarterly basis, we should measure these at least, right? Not too, not more often than that, I would say. But also when that deal closes now it becomes incumbent on the project management team, the onboarding team, to make sure that we’re hitting those numbers and we’re as efficient as possible, because [00:19:00] we’re also contributing to the realization of the promise of what that proposal reflected in terms of what we expect from a gross profit or gross margin perspective, and we’re measuring along the way.

So once we’re all rowing in the same direction it creates another set of or another level of awareness for the team, right? And when we’re incentivized together and we’re rewarded as a team or individually we can get into some of how to, create rewards and incentive programs on another podcast.

But it just helps to reinforce that teamwork wins. We can overcome these growth challenges and win, and it connects the team together in a way that mature MSPs and mature organizations work. And so help us to achieve these outcomes, and help folks raise their hands, Rich, when they find, hey, this is inefficient, [00:20:00] or we should do it this way.

Let’s become very much more vocal because once we own it, then we feel compelled to make sure that we’re doing everything that we can in a constructive and respectful manner to make sure that we’re hitting these targets

Rich: Yeah. I like it a lot, Erick, and I would say own it and understand it. The only thing that I would add to this, you’re talking about gross profit, for example, and you tell me how many very intelligent MSP founders you’ve had to explain what gross profit is and where…

How do you calculate that number, and why does it matter? By all means, get the whole team on the same page about some critical business metrics, but make sure everybody understands at some level what that metric is and why it matters. And therefore, by extension, they’ll be able to kinda see how they can kinda contribute to improving that metric.

But yeah. The, this is what we as an organization are pulling towards. Here’s what those numbers mean. Here’s why that matters. And that, that creates a whole new incentive structure [00:21:00] independent of the rewards and so on you’re talking about to get everybody pulling in the same direction.

Erick: Yeah, Rich. And I picked gross profit as a number one because that’s k- at the end of the year, this is what, we’re being measured against. Especially when we’re, trying to build equity value in our organization for an eventual exit. A couple of other KPIs would be, like, net revenue retention.

Are we keeping the revenue that we are actually closing? Are we churning out customers? There’s probably four or five really key ones, but I would start with, a GP percentage growth quarter over quarter, right? And certainly not sliding backwards. And the net revenue retention.

There’s, CSAT is important. We’ve talked about some of these metrics in the past, right? So if everybody is pulling together around some of those key metrics that ensure that every that the service that we deliver is as profitable as possible and meeting those minimum requirements, we’re not churning out or losing customers because of not, us delivering service and satisfaction to them.

Of course, they’re [00:22:00] gonna churn out because of things we can’t control, like they get acquired and, then another organization now is delivering services to them. We understand that. But how do we maximize the awareness of the team around what the role of the organization is? It is a growing organization that is growing in revenue, but also profit.

I didn’t put a top revenue member in there because that’s typically, owned by a specific team, the sales team. That could be a team member that is, you know an additional KPI that is measured that way. And of course, we’ll have other KPIs for the service team and things like that, utilization, things of that nature.

But from an overall company perspective, to give the entire organization a, a heartbeat of how we’re doing and how we’re performing, and the satisfaction that we’re celebrating these wins together when we’re hit- meeting and exceeding these milestones- gives the team that sense of confidence and that morale boost to hopefully [00:23:00] make prevent them from leaving an organization for another organization.

We want to be a best in class MSP, but we also wanna be one of the best places to work for our team and staff. And so we’re keeping an eye on all those things that incentivize and reward the team and unders- and reflect to them how their activities help us achieve those goals. Now we’ve got a much tighter organization that is working more in lockstep with each other than potentially, opposite of that.

Rich: So I mentioned a little bit earlier on, we’re recording this episode of the show on June 2nd. Turns out to b- me- to be a momentous day for our interview guest this week. He is named John Hardin. You’ll meet him in just a moment. You’ll understand why June 2nd is a big day for him as well. He’s launching a brand new vendor into the managed services landscape.

It is AI related and all about helping MSPs create repeatable, scalable AI services. John is also one of the sharpest [00:24:00] minds I know just about the challenges MSPs face around AI right now. He’s gonna join us right here on the show after this break. We’ll be back in just a moment or two, folks.

Stick around. We’ll be right back

And welcome back to part two of this episode of the MSP Chat Podcast, our spotlight interview segment where we are very pleased to be joined by John Hardin, who I can now officially describe as the CEO of Lemhi. This is actually a new title and a new company for him. John, welcome to the show.

John: Hey, thanks for having me, and thank you for nailing the name.

I’ve heard all sort of variations of it. You got it spot on

Rich: Th- this is a good little prompt on your part. We will get into that actually. I- you have revealed the origins of that name to me in the past, but I’m sure there are gonna be people interested to know that, so we’ll get into that in a bit.

And in fact, we’ll get into Lemhi in a little bit down the road. But until we get there [00:25:00] for the benefit of anyone in the audience who doesn’t know you how you got to where you are here today launching Lemhi give folks a little bit of information about who you are and your background.

John: Yeah. John Hardin, pleasure to meet anybody listening to the show. My background is basically MSP. I don’t know any other way to write that story. 17 years in the space, started as a tier one tech, worked my way up. Left that MSP to build a SaaS company that ultimately was in the contact center voice era that didn’t make it as we went through the current live world we really live in where telecom isn’t the big industry it was when I was out there.

Launched another business off of that one called SaaSlio, which was around shadow IT, and now a raging kind of concept with shadow AI, of course. And then got acquired by Auvic, where I spent a handful of years doing really fun work over there. It’s an amazing organization, grateful for that tenure.

And then ultimately back at it again on the founder side. Building a new business for the third time, glutton for punishment, but who I [00:26:00] am, a double girl dad. Super super lucky to have two young girls in my life right now. I’m a software architect. It’s a weird nuance that most people don’t know, but I’m a builder.

And I’m a avid runner. If you’re ever on the road at any of the shows, you will catch me 7:00 AM with my boo- with my shoes on, running a 5K in the morning before each show. So a little bit about me.

Rich: You’re you’re reminding me the younger of those two daughters is a few weeks old

John: at this point.

Is that correct? We are now at 10 weeks old. That’s gonna date people who wanna do the research when we recorded on this because she actually came the night of CCF awards ceremony at, down at GTIA, so I had to get out of Dodge. I was up there in Chicago, and I’ll tell you you can make it from Chicago to the west side of Indianapolis in two and a half hours if you really want to.

You shouldn’t, but you could do it, and I made it on time. We had plenty of time to spare, actually. But that is a memory I will not forget.

Erick: Congrats, and #CannonballRun,

John: thank you, [00:27:00] Erick. I appreciate it

Rich: We are recording this interview on June 2nd. June 2nd is officially the day that Limai is launching.

And Limai as we’ll get into in a bit, Limai is specifically designed to help MSPs come up with a customer-facing AI offering that they can build around. And that’s the topic that the challenges around that, the strategies that make sense around, that’s why I wanted to have you on the show.

E- Erick and I have spoken on the show recently about this issue when I was at NerdioCon a few weeks ago and I asked the Nerdio executives, “What are you hearing from MSPs about AI?” And their big takeaway was very interested in getting into the market, trying to figure out how. I was at an ASCII Group event here in the Seattle area last week, exact same thing.

Earlier this year, maybe it was last year, you spoke personally and at length with 100 MSPs a- about this question to understand what they’re thinking and what they’re struggling with. [00:28:00] So g- give folks in au- audience a sense right now for h- how common or uncommon it is for an MSP to have what they would describe as a customer-facing AI strategy, and what the, what issues they’re running into around that.

John: Yeah that 100-day, 100-interview sprint, I called it my therapy sessions. I just said, “Tell me all the problems you’ve got.” And a lot of free therapy there. And it was fun. I built a lot of the thesis of things for my business around that. But we ran into a handful of different what I wouldn’t call blockers, but things that are making MSPs struggle.

The first one is a lot of MSPs still have a lot of internal transformation to do. I really highly advise that if you haven’t done that AI transformation yourself, you have to be customer zero before getting to customer one. And so that was something that kind of surprised me but it doesn’t shock me that we haven’t done a lot of the transformation ourselves.

The next area that we got into was this, I call it the 10% and 90% rule. When I started talking to MSPs about what they’re offering, [00:29:00] the two most common buckets out there right now are data readiness and AI readiness and Microsoft coined that term. Chris Ryan, another amazing community member in our space, likes to say, “We’re not getting AI ready.

We need to get AI enabled.” And there’s nothing– I think AI readiness is actually right there on the cusp of what we need to do as an industry to do it all right. I have some opinions at Lemhi here that I can talk about later. But the first space is AI readiness. The challenge with it is it’s unsexy.

It’s not exciting to sell cleaning data up. It’s not, energizing, and the customer doesn’t feel value until after it’s done, and that value after it’s done is when they get Copilot or Claude or whatever tool they get. So that motion is one of them, and the second motion I’m seeing a lot right now is custom agents or out-of-the-box agents.

Let’s go in, let’s find a $200,000 workflow in your business. Let’s charge you $40,000 to solve it, and then we’ll get maybe a $4,000 a month retainer after that to keep maintaining it. And both of those motions are making a lot of good project revenue. The [00:30:00] problem is that it, when I ask MSPs about those two motions I say, “How many of your clients can you solve that problem with next year?”

The answer is normally 10% or 20%, back to that 10 and 90 rule, the Pareto principle, really. We could only do data readiness for 20% of clients. We can only build custom agents for 20% of clients. And I fear if we only s- you know, transform 20% of our clients next year those clients will be looking around.

They want AI change. I think everybody drank something in the water this year that said, “We’re finally gonna do it.” And my focus in that therapy was what’s repeatable and I think there’s a lot of repeatable motion that I can get into next.

Erick: John, I wanna unpack a couple things you said there, but first, this, the idea that MSPs are looking for solutions to deliver to their end customers when they are not yet ready themselves is one thing that we talk about.

It’s we have to get our own dog food first before we can deploy- Yeah … to end customers. But it seems to me like, a business, whether it’s an [00:31:00] MSP business or, a small or medium enterprise of any sort that delivers any kind of services, probably can benefit from some of the same frameworks of AI.

You mentioned agents, you mentioned data readiness and things like that, that MSPs can deliver to, to end customers. But- When you spoke to these MSPs, what are the specific things that you heard from them that their customers want most out of AI? Can you share that? It’s gotta be more than just license management and things like that.

It’s gotta be something uniquely different that drives more of that the growth needle for those end customers. So what are you hearing that end customers are wanting from their MSP specifically?

John: Yeah. I’ll joke. They want the panacea. They want the answer to everything with one solution and two letters.

That, that’s what the conversations are happening right now. That’s part of the problem with the motion right now, is you go to that customer and you ask them, “What do you want?” And you put up a whiteboard in front of them, and they just write out some big, fragile workflow that’s all these [00:32:00] different things and all these different automations.

And it becomes really hard to deliver that when you give the CEO that’s a visionary a whiteboard that says, “What do you want with AI?” But when you get into it, you’re right, it’s not just Copilot licenses. It’s gotta be more but they don’t really want that. They want– They need it, but they don’t want that data readiness and that data governance.

They don’t want the policy and the acceptable use policy and helping and supporting with them with that. But these are things they need. What they actually want is ROI. And what they actually want is to feel confident when they work with you as a service provider, that when you help them with AI rollout, and rollout, not turning on a license or not just cleaning up AI and then turning it on.

Like rollout, like enablement helping them adopt the tools, like helping them get access to, some use cases and workflows and some early quick wins. What they want is that. And if you just sell them data readiness, you just sell them policy, they’re gonna be like they’re gonna look at you with the deer in the headlights.

It’s easier to just go sign [00:33:00] up for Claude and ignore the MSP in that case. So you have to make sure that you’re pulling ROI into the front of the discussion. And my one piece of advice to service providers out there around ROI is ditch the whiteboard and come in with repeatable. There’s so much value already in y- just basic gen AI.

Everybody has access to Copilot, everybody has access to Claude. The delta between somebody getting 10X out of Claude and Copilot and 1X is often enablement. It’s the delta between I know about this use case and I know how to run it in my business, and I know how to, for instance, get email drafts in my inbox autonomously without having to do it, and the user who puts in a prompt and gets nothing out of it, is really that enablement and that training layer.

So I, I don’t hear a lot of customers on the inside saying, “I want it.” But when you present it to them that way, they go, “Actually, that’s what I want. I don’t need a license. I need activation of the licenses. I need somebody to guide me through how to get value out of AI.” The data supports [00:34:00] that most AI rollouts are in the murky middle right now according to Microsoft, so

Erick: Quick follow-up on that though.

I think, Everybody’s trying to figure out how to deliver that promise of ROI to end customers as an MSP. Are there some e- low-hanging fruit opportunities that, are the hooks that get the end customer to think, “Oh, that’s exactly ROI that I need”? Is it like a sales, focus?

Is it a financial focus? Have you gotten any of that kind of data? Can you give our audience some specific areas where the conversations tend to be more sticky?

John: Yeah. The general leniency when you think about it is some custom bespoke flow for their business. That’s what you think is what they want, and it is ultimately what you should deliver at some point in the relationship.

But we have a survey component of our platform, and when we ask employees, not the stakeholder at the, that we’re talking with on our VCIO engagement or the stakeholder at the C-suite. When you ask the employees what they want some really boring stuff, but it’s really valuable. They wanna know how to be able to do insights and analytics inside of Excel quicker, [00:35:00] or they wanna be able to, draft communications quicker to their clients, or they wanna be able to get those first iterations on documents simpler.

And while that sounds so rudimentary if you’re a power user of AI, you have to put empathy into that and understanding that our small businesses aren’t power users of AI. They don’t attend six conferences a year to learn about the newest, coolest thing. In a lot of cases, they hired us as an MSP because they maybe don’t even value that technology to a degree.

And so you have to remember, one of my most impactful stories during research is I did a ride-along with a VCIO down to a 30-person uniform production facility. They just make uniforms. And when I sat in that room and I said, “What do you want?” They didn’t know what they wanted.

When I put the things in front of them, I said, “Wouldn’t it be great if we could, auto-draft marketing or when you could get emails drafted, or you could have a chief of staff agent remind you every morning what needs to get done?” Their eyes lit up. So we just have to remember that, we are, as an industry, always the blind [00:36:00] leading the blind, and we can’t just assume that our customers want what we want.

We’re on the other side of transformation in a lot of cases, but we have to remember where we started, and we started with the basics. We didn’t start with autonomous agents and automations. We started with activating every employee to get some basic value to AI to tip that domino effect internally at our business, and that’s where you really gotta start if you wanna help that customer get ROI.

Rich: You’ve been circling around something I wanted to ask you about. You were talking there about the basics, which are relatively universal. You were talking a little bit earlier about the repeatability issue to, to keep in mind when developing an AI service. I mentioned I was at this ASCII Group event last week.

I, I had a chance to interview Jerry Cotavas, the CEO of ASCII Group, and one of the things that he pointed out is MSPs are struggling with the scalability issue in AI because Every customer is different.

Therefore, theoretically, every customer has different needs. What do I build?

What do I [00:37:00] offer that is repeatable and therefore scalable? And you clearly have some thoughts about that. So what are the the right techniques for MSPs to use to kinda overcome that scalability issue?

John: Yeah I’ll give you an exercise versus the answer. What does every desk worker do?

And if you take that question, you peel back the layer you start to realize there’s a lot of things that desk workers do. They often have calendar coordination. They often have data that they need to present to team members internally. They have drafts and documents they need to actually work against.

They have lightweight, basic research and intel that they need to pull together. They’ve got to put plot charts together in spreadsheets and pull that data together, and I know that’s not some beautiful again, that panacea that I started talking with. You will get them there.

But what it boils down to is helping them get the fundamentals first, and then moving into the agentic stuff. So the exercise I would ask if I’m [00:38:00] an MSP is, what do my clients do, and what are the basics that we can help them do with just an out-of-the-box AI tool? Everybody has access to tooling, but everybody doesn’t have access to enablement, and if you even ask that question or even enable your end clients to ask the questions themselves to Copilot, for instance.

One common exercise we’re coaching around is when you’re onboarding the clients, have every one of those people put into Copilot, “Analyze my last week and make 10 recommendations on how I could have used AI to get my life a little bit easier.” I believe deeply in the fact that there is no agent that you can build for any small/medium business today that will outweigh the value created if you activate the entire workforce with AI.

And the question to ask is what are the things that every one of those workers need to do, and how do I help them understand how to use these tools? Because we take for granted AI, we take for granted how much it’s usable. We still have to step back to those basics. Now, don’t get me wrong, [00:39:00] by the way Rich and Erick, I’m not saying we sit at the basics for the next, three years as MSPs or customers are gonna leave us.

But I propose setting that foundation, and that foundation is more than rolling out a tool. It’s going through the strategy and under- sorry and Erick, I’ll let you go. I could have gone on for a little bit there, but I have some thoughts on what that foundation looks like if you’re interested.

Erick: Oh, no, I wasn’t.

Please continue. This is good stuff.

John: Yeah. The foundation isn’t some big custom bespoke workflow. The foundation is what’s the business strategy and what are they trying to achieve? What are those numbers that move them up to the right, and how can we activate em- employees to be more productive individually to achieve those numbers?

An area that is a really uncomfortable muscle for our industry that we’re gonna have to build is that enablement of people. I know we don’t like being trainers but fundamentally There isn’t gonna be some magic trap, I don’t think, out there in the industry that we’re gonna roll out a tool and they’re gonna thank us.

The hyperscalers are outpacing anything else that gets built out there. Claude’s [00:40:00] innovating at breakneck speeds. Copilot’s br- innovating at breakneck speeds. We can just enable them to adopt the new stuff that’s coming in there. There’s a lot of value. Quite frankly, there’s a lot of value in just being a SME in the room to where that CEO doesn’t get stressed when they go golfing and their buddy tells them they activated some massive workflow, and they get freaked out that they’ve done nothing.

Just being the SME there is a lot of value. So the people, the policy and the governance, the technical readiness, the processes, the data security, and the configuration, like those are pillars we’ve mapped in a framework here that I’m happy to share after the show. It’s an open source framework. It’s not branded or anything.

But we often lead to technical readiness and data governance and then check the boxes and step away. We gotta do that whole roadmap to do it right

Erick: Yeah, John, we talk a lot on the show about the need for governance and security as it relates to AI. It’s a never-ending conversation. What is it that MSPs are [00:41:00] missing-

John: Yeah

Erick: when they’re trying to design something for their customers in terms of governance and security? And talk a little bit about the opportunity for MSPs to finally get these clients that have been reticent to, increase their cybersecurity posture. MSPs do seem to have this challenge all the time of trying to get the, their customers to, to get to the right level of cybersecurity compliance for the MSP not against any framework, but yeah, there’s frameworks too.

So you understand the challenges. The MSPs have been talking to these customers for years now, trying to get them to, “Hey, we need to, upgrade you to our next package of cybersecurity for your own good.” The MS- and the cl- customer’s going, “Yeah, we’re…” The typical, objections.

“We’re too small. Nobody wants our data.” Ridiculous, right?

John: Yeah.

Erick: But what is it that MSPs are missing from a governance perspective when they’re putting together these programs or services for AI, and how can they leverage that to pull through some of [00:42:00] these, I would say, difficult customers to to invest in, in stronger cybersecurity?

John: Yeah. I wish I had a graphic here, and maybe I can post-show note it. It doesn’t matter if not. I can represent it with my hands. I’m pretty creative. But-

Erick: Well …

John: you talked about data governance and security. And I said earlier, AI readiness is so close, so you’ve teed me up to giving that answer.

If you think about data governance and AI readiness, what it ultimately is maybe a 30, 60 or 90-day project where you’re cleaning up my online so that they can activate Copilot, Claw and ChatGPT safely and responsibly at their business. We know we have to do this. We don’t wanna put their customers’ data at risk.

We don’t want the intern finding confidential data. We don’t want the secretary getting everybody’s salary because of unshared files or overshared files. But when we think of that 30, 60, 90 days, what we look at, if we plot it on a chart, is this spike in work that must be done before they turn on AI.

And what MSPs often do is they spike up, they do a [00:43:00] lot of work, and then they step away and they turn on Copilot and they say they’re gonna monetize the licensing if we go back to the foundations, turning on tools isn’t the delta right now. Everybody has access to tools. If we know that everybody has access to tools and the difference between a 1X and a 10X is enablement and ongoing training and ongoing management of the data governance and things like that, what we can actually do is say, “Okay, we have a 30 to 60, 90-day sprint of work to clean up, then we’re gonna help you with an adoption sprint.”

So I’m plotting time going down on the MSP, but it’s more time. They didn’t just leave them once they did data readiness. They said, “Hey, you’re activated. We turned on Copilot, and now come join us.” I’m recommending right now at least a couple to three or four adoption sessions. That’s one of the best practices I did at my last AI rollout when I was at Auvik is like a four-session training.

And so again, data readiness being this really big block, and then a nice little slowdown curve on effort as you’re training and adopting, and then a [00:44:00] heartbeat. And this heartbeat represents a new part of that managed service where you’re coming in and you’re offering a monthly or quarterly session dedicated to AI and strategy with their business.

It’s I recommend if they’re less than 150 employees, you get a small stakeholder group together, maybe one or two. Bigger than that, maybe it’s an AI council, and you plug into that council as the MSP, bringing subject matter expertise, bringing observability into the adoption, bringing best practices into the adoption.

And what happens if you actually plot that 30, 60, 90 days of big work, a slow down slope during an a- adoption sprint, and an ongoing heartbeat It starts to look a heck of a lot like a managed IT contract or a managed cybersecurity contract. You are in the red for the first 90 days, and then you do a little bit more work as you onboard them, and then you are there and you keep them maintained and sustained.

And so you asked what the big barrier is. I really fundamentally think it’s a lot of framing. If you do an [00:45:00] adopt- if I do AI adoption for 30, 60, 90 days and I give you the tool, Microsoft’s data says there’s like an 81% chance that’s their le- latest work index data, only 19% of companies became frontier.

There’s probably about an 81% chance you’re not going to become frontier if you do it on your own. Those numbers go up into the right when you read the data. If you read Proskauer’s data, it says three and a half times more likely to be successful if this, if a C-suite member kicks off AI.

That’s really easy to help. If they do an adoption spread, the success rate goes through the roof. If they do ongoing council meetings, the success rate goes through the roof. And so what you’re proposing to your customer is, I can either spend 90 days cleaning this up and turning on licenses and saying goodbye, and there’s an 81% chance you’re going to plateau on ROI out of AI, and you’re going to be calling and asking what’s the value, or I can sustain this relationship for another year or two with you.

We’re going to turn it on. We’re going to do a lot of work. We’re going to peanut butter the contract. We’re not going to charge you 15, 20K for readiness. We’re going to [00:46:00] charge you 3 or $4,000 a month. And what we’re going to look like here is not 15K for an AI readiness project, but, would you give me $3,000 a month if I could help activate nine hours a month for every one of your employees?

The problem that gets missed and the little adjustment here is with AI readiness, you’re leaving right when the value gets created because you are turning on Copilot and in their head or Claude or whatever, and in their head as a client, they’re going, “Oh, my MSP did nothing. Copilot’s the one that gave me all the value.

Yeah they charged me for this stupid project.” But if you hijack onto the value being created by the tooling, what you ultimately are doing is saying, “I’m gonna be there, I’m gonna enable you, I’m gonna adopt you, I’m gonna be part of this heartbeat.” So when the client thinks about AI, they don’t go, “Oh, Copilot gave me all this tool.”

Like when I did AI rollout, nobody went and said, “Oh, ChatGPT gave me all this.” They go, “Oh, John taught me this, John taught me that, John helped me how to do this.” So the client is gonna be able to look at that value [00:47:00] created and a- append at least a part of that journey to you as the MSP. And so I think the biggest mistake right now is just doing the day of readiness, stepping away.

That adoption sprint’s a light touch engagement. The ongoing heart rate or heartbeat is an ongoing engagement, and what you could do is convert a big one-time project to a nice service layer that you can repeat over and over, and you can do that for every customer as you build up the muscle. And so I know that was like a long tirade.

I have a graphic that represents it a little better, but I think it’s just a small change that you don’t even need a tool to do. You don’t need an AI, you don’t need anything to do that. Just start changing the way that you’re framing your conversations today, and they’re gonna go a heck of a lot better.

Speaker 2: Let’s talk tools anyway. I’ve been teasing the audience for a while now about Lemhi. The time has arrived- Yeah … to part the curtains and let folks know a little bit about what it is. And before I forget, as I promised before, tell folks what that name is, where it comes from, why it’s appropriate.

But then more importantly, what does Lemhi do for MSPs? [00:48:00]

John: Yeah. Kudos to my marketing guy for coming up with Lemhi. Tim Mickle, if you’re listening I give you many compliments. Here’s another. But Lemhi is a name picked with intentionality. When… Everybody’s brain starts flashing back to eighth grade US history when I say this.

But, … when Lewis and Clark began their adventure through the West, they were given maps by the authority at that time that said, “Follow the rivers west and you’ll get to what is ultimately we now know as California.” If the brain starts firing off US history, you remember they didn’t just follow the rivers west.

The maps were wrong. There was a mountain range up in Idaho called the Bitterroot Mountains, and the Bitterroot Mountains for Lewis and Clark when they were on their journey, represented a time where they decided they had to make a decision. They could either traverse the mountains moving forward and p- chart a new path, or they could turn back and go home.

I ultimately think as an industry, we are at that moment ourselves. And the the pass that precedes the Bitterroot [00:49:00] Mountains is called Lemhi Pass, and it’s what’s represented here as a graphical representation of the Lemhi Range. So as Lewis and Clark stood in Lemhi Pass, they had a decision to make, and as an industry, we have a decision to make.

Do we turn and pack our bags and go home, or do we chart a new path forward with our customers? And so Lemhi was built on that concept that nobody knows what the map is. Look, I’m– I’ve studied it I have opinions on it. I’m probably not perfect. When we flash forward in five years, or rather I know I’m not perfect.

But if we flash forward in five years, we’re ultimately gonna know what the answer is, but we’re all in the crystal ball phase right now. Lemhi is trying to be that navigator to help you through that crystal ball phase. We’re building a software solution paired with a bunch of blueprints to monetize this transformation as a service layer in your client space but the software right now that we’re launching next week at Pax8 Beyond, as we’ve come out of stealth today on June 2nd, is around the engagement.

It’s called Lemhi Engage, and it gives your account managers, your [00:50:00] account executives, and your VCIOs, and your VCSOs the ability to begin engaging your clients down the journey that I just proposed. It’s a repeatable motion to discover ROI that you can send out to the clients and find a constrained set of use cases that you can deliver.

It’s a tool that helps you build a plan around those pillar points earlier: strategy, people, culture, budget. It gets all of those questions in one one-hour session with the client that ultimately helps you understand where they are and where they need to go. It’s a tenant readiness tool that helps you begin the data governance strategy with your client.

We have AI governance visual visualizations to give you an understanding of how much work needs to be cleaned in that 30, 60, 90-day roadmap. And ultimately, it’s a proposal to the client in a nice branded format that says, “Here’s where we are. Here’s where we wanna go. Here’s the, we’re in Lemhi mode.

Here’s the mountains in front of us we need to traverse and here’s what we can do to help you [00:51:00] get through it.” And it allows any AE or AM or VCIO to do that motion, not somebody who studied the topic and gone to every Microsoft AI tour. It allows you to pull that pressure, what we’re hearing internally, off the CEO or off that AI specialist in your team that can have the conversation.

It allows that release valve to open so that you can talk to 100% of your clients next month, instead of only 10% of them when that specialist is available. So that’s just the beginning. Lemhi is ultimately building a compass module that is designed to take you through the mountains with your client, and that’s that ongoing VCIO motion.

We are building a tool to facilitate that, helping you measure how much adoption is actually having in the tools once they deploy and roll it out. Are those ROI use cases you talked about actually becoming a reality? Is there ongoing data governance that needs cleaned up from your technology team?

So it takes that one-time product and turns it into a journey that any account manager or VCIO that wants to [00:52:00] get skilled up can run with their client and deliver AI as a transformative service over a multi-month journey. And I’ll say one last thing, and this isn’t about Lemhi, but it’s an important piece.

AI, when you read into the studies of how it b- gets done successfully or not, it’s not a boom. It’s not a 90-day sprint and you’re gone. It’s iterative and repeatable reinforcement just like any other org change. This isn’t rocket science. I didn’t reinvent the J curve here. It’s just every other org change that happens the, this is why I’m so passionate because if you think about that statement that it’s a repeated iterative momentum-driven change, there is no better business in your client’s relationship than the MSP suited to deliver that.

Now it’s new muscle, it’s like an uncomfortable muscle, and we had to build uncomfortable muscle in the cloud, we had to build uncomfortable muscle in the cybersecurity. I’m extremely confident we’ll build uncomfortable muscle again here in the AI era. But there’s no other [00:53:00] person in that client’s relationship that they’re gonna have that they can trust to do long iterative improvement.

You’re already there for multiple years with them. You might as well help them drive this next big change

Erick: John, we talk a lot on the show about the need to move and the opportunity to leverage AI to move the conversations from the server room-

…

Erick: To the boardroom, right? Because, we’re not selling technology solutions anymore in this new era of AI.

We’re selling, business outcomes and moving the needle. And it’s been a challenge historically, for MSPs to p- realize and demonstrate that they’ve delivered on the vision or the promise that they’ve laid out for a client at the beginning of a, a scope of work or a project or s- or things like that, right?

And one of the things that you emphasize in Lemhi is that measuring and reporting of that value promised and how– and the realization of that value, which to me is a [00:54:00] boardroom conversation all day long, and strikes me as being, I get a seat at the, at the, in the conference room when we’re having these meetings and being much more strategic as a vCAIO, right?

Talk about how those two metrics really help move the needle for MSPs and what… I guess the follow-on question would be what that forces MSPs to do as they look at the rest of their client base and try to determine what’s their ideal customer profile moving forward look like.

John: Yeah, my opinion on an ICP is a lot like they better be ready to adopt.

I remember my COO at my MSP a decade and a half ago, he said, “Get real comfortable with being uncomfortable.” And I think you can’t put that in an ICP, but you really need to find those customers who wanna make this investment. I have a deep fear for SMBs out there not adopting. It’s why our mission is no SMB left behind with AI.

I really wanna help these SMBs get adoption of it, and I think the MSPs are the right path. So you asked the question though, [00:55:00] like how do you defend those statements and how do you get yourself to the boardroom? Ultimately, we all know that every board has pressure downstream right now on their clients or most boards have pressure downstream on their clients for AI.

This is a lot of how we made money back in the cyber era, by the way. The board had pressure to be cybersecurity, and that’s why the infinite budget came up because every Forbes article was talking about another ransomware here or there. Now every Forbes article’s talking about AI here or there, and so now it’s just time to keep up with the next motion.

So when there’s pressure already coming down, what you need to do is take your customer and make them look like a hero internally. And so we do that by plotting a few ROI metrics. I started earlier with those low-hanging touch points like the Excel analysis and documents, like the really boring stuff, but does level up giving hours back to every employee So we measure a few different things.

In the platform, we do a pulse survey on a regular interval right natively inside of the ways that they’re working, so right in Teams or right in email, where you can pulse the customer. [00:56:00] This is something where you can say, “How comfortable are you with it? How much are you using it, and how many times does it give you value back?”

If you read into Microsoft’s data, those are data points that they’re collecting to understand frontier status. So we’re asking on a regular interval. You’re not gonna get everybody to take that survey, but if you can get 20% of your base, you can get some understanding of what’s going on. So what is the anecdotal data from the base?

What is the quantitative data that ultimately comes in from our observaber… can’t get that word out, observability module. How much is they, are they adopting AI? This is a sticky wicket. Just because prompts are going up to the right, that doesn’t mean they’re, That means they’re adopting it, is what Microsoft calls it.

That’s adoption. That’s not absorption. Absorption is a term that is used, meaning that they change the way that they work because of AI. That’s the second metric. So pulse surveys adoption metrics, and then ultimately how many of those people are taking the trainings and being part of those adoption sprints?

How many people… We have in our platform a prompt [00:57:00] library that’s custom to the way that they work. Like, how many people are engaging in the learnings and the trainings there? And I believe between the employee sentiment, the quantitative observability data, and the training and enablement completion data, we can do this and say there’s an ROI coming back, because if the employees are saying they’re feeling ROI, if the employees are saying they’re getting quality inputs of ROI and they’re saying they’re more comfortable with it, then we know in general that they’re getting adoption and ROI out of it.

Now, that’s why we start our engagement with that pulse to get… to be fair. We wanna know where they are so that you can show those numbers moving up to the right. And we all know we love telling a good up and to the right story to the board and that’s how you earn the r- the right to the board.

You say, “Hey, we’re not just gonna roll out Copilot. We’re gonna, we’re gonna quantify some key metrics as we go, and we can all agree that these three key metrics probably mean we’re doing the right thing. And if we can all agree on that, then we’ll bring it to you on any regular interval you want to defend how we’re moving forward in [00:58:00] AI.”

Rich: There, there’s so much reporting going on right now about businesses adjusting their AI budget, Yeah … due to disappointment around results and so I’m just thinking about that. Y- yeah, if you want to deliver AI services to your customers and make that something that is sustainable and that they’re willing to invest in over time, you are gonna need to help them appreciate what they’re getting back, what the return on their investment is.

I wanna pick your brain about something a little bit here, though, while we’ve got you. Because you were talking earlier on in the conversation about how there are some MSPs out there who they’ll go find a $200,000 workflow, and they’ll automate it, and they’ll price X dollars a month to, to save companies a bunch of of money around that workflow and so on.

And it gets to this whole thing that keeps coming up online, at events I go to, in conversations about outcome pricing. And the anticipation that the world we are headed to is one in which end users are going to insist upon paying for [00:59:00] outcomes as opposed to just a monthly service fee for for whatever.

And my frustration is I personally don’t know what that would look like yet for an MSP. And that’s the piece I most wanna know. I agree. I certainly appreciate the logic of outcome-based pr- pricing, and my gut tells me that is where we’re heading too. I’m trying to understand what that might look like for an MSP.

Do you have any thoughts around that at this point?

John: Man that, that is one thing that I really struggle to get behind. I know it, it makes sense. I wanna meet the CFOs that are willing to pay like that. I get it. In general, it makes sense, especially in a utility area where I’m gonna come in and I’m gonna cut the amount of like paper used in printers, or I’m gonna cut the amount of AWS spend, and you’re gonna give me the outcome based pricing back.

Like that model makes sense in a lot of utility things. And so maybe if we believe the MSP is becoming a utility, it does make sense. But you have to display cost savings in those conversations. And displaying a [01:00:00] cost savings is really difficult with like general AI rollout. Now, with these use case based scenarios, I think there’s a lot of ways to get like st- table stakes in front of that conversation to validate it and say, “Hey, you agree it’s 200K, and we agree it’s 200K.

If we do it, it works out.” But I’ll be honest, like my biggest fear is the innovation around it. Last year, ChatGPT 4.5 came out, and we all remember how blown away we were with the reasoning models that came out. We’ve got Claude Cohort doing our job for us. We’ve got in Microsoft, inside of Microsoft, there’s a f- a group of employees called Frontier employees inside of Microsoft.

They have access to ClawPilot. It’s an open claw variant of Copilot. I struggle that we’re not gonna be in an era here in a year or two where all of that work we just built is gonna be moot because now ClawPilot can do it for me. And so I really struggle with the agents. I see it and I’m not tr- I’m not challenging it.

There’s a lot of money in it. Make sure you price it for what it is. It’s a custom software [01:01:00] development engagement. Do not price it differently. You need to have a monthly retainer. You need to have ongoing services to keep it afloat. So I’ve deviated from your outcome-based pricing for a couple of reasons, but the core reason is I just, I struggle to see a world in the near term, I believe we’ll probably ultimately get there, but I struggle to see a world in the near term, especially when the McKinseys can’t even justify ROI of AI to find our saves in our industries with our small, medium business customers saying we can defensively position ROI.

And I wanna hit one last point, Rich. I’m sorry. I know I’m a verbose speaker. But earlier you said people are pulling back their budgets from AI, and I wanted to highlight and talk about that for a minute because the reason they’re doing that is ’cause they can’t really ju- justify the ROI because it is intangible.

But the reason I think a lot of them are pulling back is ’cause if you look at the 2026 Work Index data by Microsoft, they say that 19% of companies became frontier firms. I m- I mentioned that stat earlier. 50% in, are in this [01:02:00] murky middle. So 19% of companies with big enterprise supports and budgets have absorbed AI and changed the way they work.

Those companies aren’t pulling back their AI spend, I guarantee you. It’s the 50% murky middle And therein lies again my flag as the opportunity. Whether you outcome price, whether you do a managed service price, whether it’s a project pri- or however you price it, the opportunity is there to help them do it right.

And then if you help them do that right, you can justify, whatever model you and the CFO can shake hands on. But I don’t know. I don’t… I struggle from a CapEx and OpEx side of the finances to really believe that I can even get comfortable paying outcome pricing, and I’m pretty, pretty forward-thinking in the way that I’d like to do things with AI.

Erick: That’s very interesting, John. You alluded to, when the services become a utility, then it might make more sense, and I think that’s… I was thinking about that as I was listening to you expand on it, and it’s like managed services didn’t just come along and become the way that we deliver IT services.

[01:03:00] We had a years and y- decades of delivering services on a reactive break-fix or project-based model until some smarter folks decided, “Wait a minute, maybe we can create a subscription for this and guarantee some SLAs and guarantee some outcomes.” And now it became, I think, expected that outsourced IT services are more of a flat fee managed services type of a model.

I’m aligning that with where we are with this, with AI. It’s we have a… It’s AI at, in and of itself, we’re still trying to figure out, the different value propositions for agents and workflows and things like that. I… And so I appreciate your perspective that says, yeah, we’re early days here.

It’s risky if you try to build something and then charge for it based on an outcome where the cheese might be moved later. A different, LLM model comes in and you have to recreate everything, or the client, it’s not sustainable. It’s not a one-and-done thing like you said, which leads me to my question.[01:04:00]

Knowing all this and knowing what the landscape, the potential landscape looks like here that no one can predict in three to five years. Remember he’s talking about five years out, seven years out? We’re seeing this stuff happen in, warp speed, right? How does an MSP qualify to be a vCISO?

What is it that, gives an MSP the confidence and purpose to stand up and say, “We are your vCISO, and here’s what qualifies us to deliver these services for you, and this is why you should trust us”? I think LEMHI helps in that regard, but what beyond that? Because I think that for typical MSPs that are engineers like me We think, oh, we’re thinking in sprints, we’re thinking in projects, we’re thinking, like you said earlier, and we’ve heard other examples of models with AI where you don’t just deliver the, the licensing and the [01:05:00] co-pilot, and then turn the keys to the Cadillac over to the kids, right?

You have to enable them, deliver some prompts, and work with them in that long-term, evergreen kind of a vision like you described, right? That’s where you monetize over time. So back to my initial question. I said a lot, so be- No, I- I’m trying to match your verbosity here. So what are your thoughts on, any of those that I asked, but specifically around this kind of is it an imposter syndrome for a while to be in a vCIO?

And if so, where’s that light… when does that light at the end of the tunnel get bigger for the MSPs?

John: Yeah. I’ll bury the lead. It’s imposter syndrome. Fun- fundamentally this is no think back five years ago or six years ago. We had imposter syndrome during the cybersecurity era.

We had imposter syndrome when it came to cloud. I still remember reading my Azure docs. You get those core memories built in your brain. I still have a core memory of reading an Azure doc and going, “I have no idea how I’m ever gonna do this.” Like I, [01:06:00] it’s just there’s no way. I’m used to plugging in 1950s and bonding T1s into DS3s and plugging them in and that’s how we do things.

There’s no way I’m gonna move this to the cloud. Here we are 10 years later or 15 years later and it’s done, and I had the same level of pain when I looked at the cybersecurity space, and I know we as an industry did. So like tap into that core memory for a moment. You had that feeling multiple times in your life if you’ve been in this industry for more than a decade.

And every time, on the other side of that core pain you felt, you figured it out, you delivered it. So I steal a quote from Wes Spencer here ’cause he’s so smart and I just love his point here. We are an industry that is the blind leading the blind. We have always only been a step ahead of our customer.

We’ve never been a mountain ahead of our customer in the… If I’m gonna go Lemhi analogies here. We’ve only been just a step ahead of them on the trail. And so fundamentally, we have to knock that imposter syndrome out. When our SMBs at the 30-person uniform [01:07:00] company need help with Excel and Insights, like surely we can deliver that when we’ve done a lot of AI transformation internally.

So I think it’s a lot of imposter syndrome personally and I think vCIOs and account managers can build that muscle pretty quickly with some intentionality. I wanna hit one other point though that you said preceding your question around the vCIO, and you said, this lack of confidence in building something because the cheese is gonna move.

In software engineering, we learn some, a lot of principles, but one principle we learn is called the steel thread. And it’s this concept that if I’m standing on a building here and I’m standing on a building here and I wanna get to this building over here, what is the bare minimum thing I do to get from point A to point B?

And in engineering it’s called the steel thread. You build the minimum viable feature to get the customer the value they want I think the only thing that we could probably write in a contract today that we can guarantee without any lack of confidence is that there’s gonna be change in the next year in AI.

And we know for a fact that if we went from ChatGPT 4.5 last year, and the [01:08:00] predecessors to that were us playing around with the s- the previous models, and we now find ourselves here with Claude Cowork and Claude Pilot and all those innovations happening. We know for a fact that next year we will likely see Claude Pilot in the stream.

We’ll see Copi- or not Copilot, but Cowork in the stream of all workers, and we know that we can talk to our customers and say, “Dear Mr. or Mrs. Customer, haven’t you felt the change in the last year of all of the things happening? We have. We’ve innovated inside of our business,” which is why you gotta be customer zero to say this.

But we’ve innovated our business and we know how much is gonna change. We have dedicated staff who are in charge of following all this change and being there to help you guide the future of where we’re headed, and we w- we know that, these big innovations are coming. We know we’ve got Claude Pilot on the pipeline.

We’ve got… You probably don’t get technical, but a digital employee is an agentic AI on the pipeline for you. But we know it’s gonna change, and we want to put a vCIO in your business today, so when your CEO goes [01:09:00] golfing and hears from his buddy that another 40% efficiency was driven at their manufacturing firm and not th- not mine they know who they can talk to.

And what an amazing opportunity, whether you build agents or whether you don’t build agents. If you are that trusted advisor that they come to when they get off the golf course and they go, “Ugh, I meet with Eric next week I gotta bring up this thing that, that, yeah, Phil told me.” And I f- And he’s sitting here telling you all the things that Phil told you, and you go, “Great, we can help you with a project just like that.”

You will spin so much opportunity out of this real heartbeat layer that you’re establishing with all your customers if they look at you as the answer to that panacea they’re se- seeking for. And that’s why I argue we gotta start today and planting those conversations. Whether you build it into the managed service contract, whether you build it as a new layer on top of it, you gotta make sure your customers know that you are the trusted advisor.

We have to knock that imposter syndrome out, because w- we can empathize with a decade and a half of experience, at least I can, [01:10:00] of feeling like I was inferior to this conversation. And look, nobody’s AI experts. I just started this journey January of last year, so that means what? I got 18 months on everybody?

That’s nothing. So just break it off and just get out there and start those conversations. There’s no way to build muscle unless you start doing reps.

Rich: All right. John fascinating stuff. Thank you so much particularly on a very busy day, launch day for Lemhi. For folks in our audience who want to get in touch with you, learn more about you, learn more about Lemhi, where should they go?

John: Made it real simple. Wore the shirt today. Lemhi. You can reach us at lemhi.com. I say all this I’ll be doing education over the next year, but we’re only taking in a limited cohort of design partners between now and November. And so if you, if what I just talked about sounds like it resonates to your business feel free to join.

Go to our website, hit our wait list. We’re reaching out to people on a weekly cadence with- to those people on the wait list. If this is something you’re serious about, reach out to me directly. You can hit me up on LinkedIn, [01:11:00] [email protected]. I’m always happy to have a conversation.

And Rich, you said it’s a busy day, but like I’ve said this to you before, I read your news higher priority than the majority of the r- news out in our MSP space. I will always take the time to talk with you, Rich. So thank you for having me on a busy day.

Rich: Thank you very much for those kind words, John.

And and congratulations on the launch of Lemhi. For folks in our audio only audience who can’t see the shirt, that’s L-E-M-H-I.com if you wanna check the product out and learn more about it. Folks at this point, Erick and I are going to take a quick break. When we come back and rejoin you on the other side of that, we’re gonna share some final thoughts about this very interesting conversation with John Hardin, have a little fun, wrap up the show.

Stick around, we’re gonna be right back

And welcome back to part three of this episode of the MSP Chat Podcast. And one last thank you to John Hardin on a big [01:12:00] day, taking some time to speak with us about AI and managed services, as well as his new tool Lemak. So many dimensions of that conversation we could go into Erick. I told you John is somebody who thinks at a very deep level about the challenges that MSPs face around monetizing AI services for their customers right now.

He had all sorts of interesting thoughts. One thing I’ll just call out that really stuck with me is at a time when all anyone wants to talk about is agentic this and agentic that, and building agents for customers, the idea that the real near and medium term opportunity is around people.

A- activating the workforce, as he put it. Enabling them to get value and be more productive and get more done using AI. That, that’s the thing that’s maybe gonna motivate management most to continue and make deeper investments, gonna get the workforce on board with being a part of an AI initiative.

I kinda like that a lot. And [01:13:00] then later on in the conversation, he was talking about getting into an iterative motion with customers. Once you’ve kinda done that readiness and governance work, once you’ve started actually rolling out solutions, the idea that you’re gonna be meeting with them on a regular, a quarterly or recurring basis and improving, iterating, adding optimizing tweaking and so on.

I think it’s really important because as other people sometimes observe we’re at an early stage in a big technological pivot right now, and it’s a little bit like the big technological pivot when PCs first appeared, and all of a sudden PCs were showing up in offices and there was this idea initially that these are amazing productivity tools.

If I just put one on everyone’s desk, everyone’s gonna get more productive. And it took years before people actually did get productive. Years of learning and practicing and researching and trying things, iterating. That’s exactly how AI is going to work. It’s gonna be a process. It’s gonna take time.

You’re gonna need to get executive [01:14:00] leadership bought in and so on. And so having that awareness and, and accepting that this is how it’s gonna work and making sure your customers understand as well that th- this is how it’s gonna work for them and this is the value that you’re bringing to them, that you’re gonna guide them through that process, I think it’s a smart and valuable idea.

Erick: Yeah, and I think in so many things in my head after having that conversation with John and I’ve… i’ll just pull one, and it was this idea that, when I asked the question about, the imposter syndrome, right? It’s how do we become a vCISO and not feel that?

And, his response, Rich, was, “That’s what we’ve been doing all along.” MSPs, like we’re… When cloud came, when cybersecurity came, we’ve figured it out along the way, and I thought that was refreshing because that is exactly how change happens. It’s like nobody has the magic bullet.

Nobody knows what’s going to happen. And AI is just such a complex and fast-moving technology, [01:15:00] probably the fastest adoption of any technology in the history of the planet, right? How can we not feel like we’re a little bit of imposters when we’re trying to figure this thing out?

But I think what I really walked away from from John’s interview, and, we… On The Chill, we’ve had lots of AI folks, really bright folks give us some of this insight, and it’s e- every time we have a conversation, I, That’s another puzzle piece I’m putting into the puzzle now.

Now I’m trying to get this full picture, but it’s just this sense that AI is more hands-on than what MSPs typically feel their role is with clients, and I think that’s the caution that I want to shine a light on for MSPs. If we treat AI as just another project and hand it over it’s not going to succeed in the way that we want, and, John mentioned some very s- clear statistics around that.

It, for… [01:16:00] Because it’s such a- A critical technology for any business, we have to think much differently as MSPs. And I hope that MSPs are preparing for this because it is not a one and done solution. We have to adjust our relationship with our clients, and we have to become that next level of VC, AIO, or strategic partner to the organization and have much more different conversations than we might have had in the past.

I know when I started as an MSP, Rich, we were selling solutions, technology solutions, and we weren’t selling business outcomes. We weren’t having boardroom meetings or any of that. Now today, mature MSPs are having those conversations. Of the majority of MSPs as we’ve seen from industry reports and statistics, they’re taking a little bit of a wait and see attitude toward this.

And, I think you can’t wait to be pushed out of the airplane door for the parachute jump, right? You [01:17:00] have to prepare, you have to train for it, and you have to start asking these questions of your clients and really understand that your relationship with them has to change. And if you’re an, like me, an engineer, you either have to build those muscles, like John said, like these uncomfortable muscles yourself, or you have to find somebody in your organization that can lead that charge.

Having an AI champion within your client’s teams, like he mentioned, requires us to have AI champions in our teams. We can’t, expect to be reactive to this. We have to build that internally so that we can represent to the clients and our prospects our journey through it, which is another thing that I liked that can be used as a case study.

This is what we’re doing. This is what we would like to do for you, and guide those clients along the way. And then that last, or one of the last questions I asked was about changing up your ideal customer profile, and his answer was very, immediate. It’s yes, you have to be working [01:18:00] with clients that want to go on this journey and want to grow AI in this direction.

And so my question to you, Rich, is Do you think that MSPs will start evaluating their customer lists in a different way than what we have traditionally been looking at from an ABC perspective now because of the opportunity for AI and how much additional work it’s going to take to do it, but how much additional revenue and value we’ll be able to deliver receive on our end, revenue and value we’ll be able to deliver to our clients?

Rich: So eventually but I think there are a lot of other things that they’re gonna be focused on figuring out before they kinda get to that point in terms of perfecting a a set of services and a kind of relationship with all the stuff that we’re talking about here. And I think at some point, it, in the list of difficult challenges that people are gonna have to think their way through, I think eventually they will get to the point where they begin to understand which [01:19:00] customers, what kinds of customers are gonna be the best fit for their business going forward and which aren’t.

I, they’re gonna have to… They’ll get there Erick, but it’s gonna take a little bit. And the last thing I’ll say before we move on is simply this is a lot to think through and you’re gonna make mistakes, and that is totally okay. You don’t have to figure it all out overnight. You don’t have to perfect any of this within the next two weeks, but you do need to kinda get started and just get that iterative process in terms of what you deliver and what your customers get from you sooner versus later.

The sooner you start the better, the sooner you’re gonna get good at it and folks, with that we are left with time for just one last thing. And this is a little timely for me. About five weeks ago, Erick, I left a pair of prescription sunglasses behind in an Uber. A very expensive mistake for me, and that is why this headline from today’s news jumped out at me.

Uber has published its [01:20:00] 10th Annual Lost & Found Index where they quantify what it is that people like me and you leave behind most and least often in Ubers. And most often, I don’t think any of this is gonna surprise you, phones, wallets, luggage, right? You put it in the trunk, they drop you off somewhere, and you walk away.

Keys, headphones. Where things start to get interesting is when you get into some of the least most often, the le- least common lost and found items, say, for example, dentures with only two teeth. This has happened. Breast milk. I’m guessing that would be a, an unpleasant surprise to realize you, you left that behind.

You might need it. Something identified only in this news article I’m looking at as 420 donuts. I don’t know exactly what those are, Erick, but I can guess. And they have been left behind in Ubers before. So have live fish live butterflies, and in one case at least, [01:21:00] two trees, and that one I’m really trying to figure out.

Was that in the backseat of the car? Was that in the trunk? How do you bring two trees with you to an Uber ride and then just forget about it?

Erick: Rich, I remember one time we were podcasting at an event, and I left a table throw in the back of an Uber, if you’ll remember, and I had to go back and try to figure out what happened to that table throw.

It w- all worked out in the end for us, though. But so what happens with this this list? Was there any guidance? Were there any tips from Uber to say, “Hey, if you think you left something, here’s what you do”? And did they give you anything there, or was it just “Here’s our report, guys.

Keep an eye on your stuff”?

Rich: I was focused mostly on the report, the results, so I don’t know actually if they if they provided guidance on that. It is something I was thinking about a little bit because it honestly didn’t even really occur to me is there a way for me to get in touch with Uber and ask, “Hey, did somebody report sunglasses that are useless to me because they’re in Rich Freeman’s prescription?”

Yeah … ’cause if there’s a way to get that back, it might be useful. I- that there must be some [01:22:00] place you can go for that. It’s probably a low percentage probability that it’ll all work out for you, but, Yeah … yeah, I don’t know.

Erick: Yeah, in the table throw scenario I, pinged the driver, right?

Because you still have the ability, I think you still have the ability, and then I also pinged Uber. Uber was less than helpful. They were like nope.” The driver was like, “Oh yeah, let me, let me… Yep, here it is. Let’s connect and I’ll get it back to you.” So that was nice.

Rich: That and that is an actionable tip for our audience.

That’s what I should’ve done with the sunglasses. Keep that in mind, folks. You drop something behind leave something behind in the Uber, don’t contact Uber. They’re getting approximately a million such requests an minute. You wanna contact the driver, say, “I made a mistake,” see if you can work something out.

Folks, that is all the time we’ve got for you this week on the show. We’ll be back in a week’s time with more for you. Until then, I’ll just remind you this is both a video and an audio podcast. That means that if you are listening to us right now but you’d like to check us out on video, you can go to YouTube.

You’re gonna find us there. If you’re watching us on YouTube but you’re [01:23:00] into audio podcasts, you can go to A- Apple, Google, Spotify, you name it. Wherever you get your audio podcasts, you’re probably gonna find us there, too. And wherever it is you find us, please subscribe, rate, review. It’s gonna help other people find and enjoy the program just like you do.

This show is produced by the great Riley Simpson, part of the team with us here at Channel Mastered where we have an end-to-end set of services for helping vendors build, grow, optimize thriving MSP channels. You can learn all about those services at www.channelmastered.com. Channel Mastered has a sister organization called MSP Mastered.

That’s Erick and his team working with MSPs to help them grow and optimize their business. You can learn more about that at www.mspmastered.com. So once again, we thank you for joining us. We’ll see you in a week. Until then, please remember, as we always urge you to, you simply can’t spell channel without [01:24:00] MSP.