How to Make Money with AI (as an MSP), with Jimmy Hatzell

[00:00:00] And you know, I wouldn't want to be running an MSP business where no one of my customers have adopted AI in three years because it's a huge threat, right?

[00:00:10] And you know, if we're in agreement or at least in semi-suspicion that AI will have disruption in industries, it's the companies who are going to adopt them three, four years down the line that are going to be the ones that have benefited from it instead of the ones that were disrupted by it.

[00:00:25] Hey guys, Damien Stevens, host of MSP Mindset. Today, we have a special edition where we recorded at Build IT Live 2024 in Orlando, Florida.

[00:00:41] Special thank you to Sonny and to Calm for allowing us to show up and interview such amazing MSPs. And thank you to the whole IT by Design team.

[00:00:53] I want to dig right in. We're at Build IT Live and I just had an MSP asking me, do you know anybody that's actually making money with AI that's an MSP? And how, if so, like, so let me start with that question.

[00:01:07] Sure. I know lots of people making money off it. I've got lots of partners doing it right now. It's a big change, can be a big change.

[00:01:15] There is margin in AI software that MSPs can make money off of through our platform or another platform.

[00:01:25] There is assessments around AI, AI readiness, data governance, that sort of security world of it that MSPs can make money off of.

[00:01:35] And then I actually have MSPs offering managed services around prompt engineering and AI management.

[00:01:42] So if you roll out the technology and someone needs help getting it to work right or to respond better or respond correctly.

[00:01:50] If it's funny.

[00:01:51] Yeah, yeah, yeah.

[00:01:52] Yeah.

[00:01:53] You know, yeah, when for sure.

[00:01:54] Yep.

[00:01:56] When.

[00:01:58] You know, we like on our team, we have prompt engineers that are sort of, you know, in the, in the, in the, that you can escalate up to.

[00:02:08] Uh, but we do office hours with MSPs who are training their own teams and we offer certifications that, you know, they're learning how to do prompt engineering and offer it as a service to their clients.

[00:02:19] So everything from use case ideation, the security, the data governance and data privacy aspect of it, um, to how do I make this work better?

[00:02:29] And I thought it could do it this way, but it responds this funny way.

[00:02:32] Can you help me edit this?

[00:02:34] Um, I've seen MSPs making money in all those areas.

[00:02:38] So if I'm, I want to talk about the other side of this, but if I'm excited about AI and I would say most of them is field talker.

[00:02:44] They're excited, but they don't know what to actually do to help.

[00:02:49] Where should I start?

[00:02:50] So, uh, what we've seen is if you can use generative AI, I'm talking about the co-pilots, the HatsAI secure chat, the, um, chat GBTs of the world.

[00:03:03] If you get yourself or an employee to use it for 10 hours, the use cases, uh, and areas of efficiency start to come out naturally.

[00:03:15] So when you're working from that frame of reference, it's not, okay, here's my big idea.

[00:03:20] I'm going top down.

[00:03:22] I'm going to roll this out to everyone.

[00:03:23] It's how do I go bottom up and how do I get everyone on my team to experience that those 10 hours and start to ideate for themselves?

[00:03:32] And it's different than most technology, uh, macro changes that we've seen in the past from an IT perspective.

[00:03:41] Tip, so typically historically movement to cloud cybersecurity, you know, we're going to, um, like take, take a big IT initiative, like, like move to cloud.

[00:03:53] We're going to increase our efficiencies and cost reduction by X percent.

[00:03:57] Uh, and in order to do that, you know, here's the project timeline and here's when we're going to move these things.

[00:04:03] And here's when this business unit's going to do that.

[00:04:06] And, and that, you know, that's sort of our frame of reference as IT people on like how projects are done.

[00:04:13] And many companies are doing, making this mistake with AI, you know, here's our strategic AI committee.

[00:04:19] Here's our one person who's going to work on rolling AI out to everyone.

[00:04:23] And they'll come up with an amazing idea and an amazing use case for, you know, something very complicated.

[00:04:30] When in reality, the quickest way to gain efficiencies in a business is just getting people to use it and ideate it on themselves.

[00:04:38] So you need that bottom up approach.

[00:04:40] And if you don't have that bottom up approach, people will use AI, whether they tell you they're using it or not.

[00:04:46] So you're going to have employees, whether on your team or in your customer base, you know, you put that Cisco umbrella out or some, you know, SAS monitoring.

[00:04:54] I'm sure chat GBT free is going to, you know, ping up somewhere probably in higher percentages than you'd expect.

[00:05:02] And they're going to be using it without guardrails.

[00:05:04] They're going to be using it themselves.

[00:05:05] You're not going to have visibility and you're going to have those, what we call secret cyborgs.

[00:05:11] People using AI to do their work and not telling everyone because they're scared of it.

[00:05:15] So you need that bottom up approach and that top down.

[00:05:19] Hey guys, today's episode is sponsored by Servocity.

[00:05:24] I created Servocity because I was an MSP that lost data.

[00:05:29] Then I had to go in front of my client and tell them I can't recover.

[00:05:33] If you've ever been there or worried that you would be there, then you may want to take a look at Servocity.

[00:05:39] We do two things that nobody else is crazy enough to do.

[00:05:43] We test your backups maniacally.

[00:05:45] We do that daily, weekly, monthly, and quarterly for you.

[00:05:50] Then we manage the backups for you so you can focus on growing your MSP instead of babysitting backups.

[00:05:57] If you're interested in our unlimited storage or immutable storage or any of the things that I'm talking about, visit servocity.com.

[00:06:06] So you brought up something that I think, I don't know if it's 50-50 or 80-20 or what, but a lot of people have fear of AI.

[00:06:17] Let's kind of start with even the MSP owner, CEO.

[00:06:22] Yep.

[00:06:23] A lot of them are like, am I going to be out of business next year?

[00:06:26] Is there no need for managed services?

[00:06:27] Like there's these fears.

[00:06:29] What's your take on that?

[00:06:30] So I think that, so if you look at, go back to the 1700s for the Industrial Revolution.

[00:06:39] You're a small brewery in the 1700s and all of a sudden the steam engine is invented and you can get ingredients from different places and a supply chain is forming.

[00:06:51] You sort of have two options, right?

[00:06:53] You can make the same amount of money for less money and you can fire a bunch of people.

[00:06:59] Or you can hire a lot more people and scale globally because all of a sudden now you have this distribution system that didn't exist before.

[00:07:09] And if you are approaching an AI project or an AI rollout from the perspective of, you know, we're going to make the same amount of money for less money, it's going to be hard to get your employees on board with it because where's that cost savings coming from?

[00:07:27] It's coming from them.

[00:07:29] So it's important to sort of drive your North Star of where you want to go and what you want to do and emulate that through your entire business on, you know, we're going to gain efficiencies and here's what we're going to do with those efficiencies.

[00:07:46] Here's what we're looking to do because if you don't, you'll have, you know, the secret cyborgs who are using AI without telling anyone they're using AI because they're scared that they tell their manager that the work they've been sending them the past three months has been done 20% by AI and in, you know, less time than they're known they're going to get more work or their team is going to shrink from three people to two people.

[00:08:13] And it's real fears that people have.

[00:08:16] And I think MSP owners, you know, they love efficiency and process because the way that we think as IT people or we're trained to think.

[00:08:28] So we don't associate that with a negative thing.

[00:08:31] But if you get in the head of a average worker at a company, any company may not be yours, you know, left without a North Star or guidance, there's no actual incentive for them to sort of say why they're using AI.

[00:08:49] And they may be scared to tell anyone they are.

[00:08:51] Yeah.

[00:08:51] Yeah.

[00:08:52] You end up with shadow AI.

[00:08:54] Shadow AI.

[00:08:55] I remember it was over a year ago.

[00:08:57] I interviewed somebody and I asked if they were using AI and they had already used it in the interview process.

[00:09:03] The good news is like it wasn't bad, but like it was the kind of thing you could run something through and detect.

[00:09:11] And the only wrong answer would be lying, like from my perspective.

[00:09:15] It's not that you couldn't use it, but let's just talk about it openly.

[00:09:18] And fortunately they did, but I don't think that's where everybody would land.

[00:09:23] So from an MSP owner or CEO perspective, the mindset isn't all my business or all my employees are going away.

[00:09:32] What about from, I know we talked about this fear, but from an employee perspective, I think maybe half and half of the MSP owners are excited about it and others have some fear.

[00:09:42] Probably not that even, but let's say it's that.

[00:09:44] But it seems more universal that if you're working somewhere, it seems like nothing but downside.

[00:09:50] Yeah.

[00:09:50] At least when I talk to people, they say, you know, I'm concerned it's going to take my job or I have to do 10 times as much or, you know, some, there's variations, but some version of that.

[00:09:59] So what is your take on that?

[00:10:02] I think it's a real fear that should be addressed and you should think of what the reality of it is.

[00:10:10] So take an MSP perspective.

[00:10:17] Doing onboardings quicker is a great use case that people can use right away.

[00:10:22] Communicating with clients better is a great use case you can use right away.

[00:10:27] Spending less time doing your project management is a great sort of way to get going.

[00:10:31] These are things that sort of exist today.

[00:10:33] And as sort of the technology progresses, you're going to see AI integrated in more things you can do.

[00:10:38] So is AI going to crush tickets overnight?

[00:10:43] No.

[00:10:44] Are we going to reach a stage where your employees, your technicians can move through tickets faster because they have AI assisting them in a co-pilot sort of co-intelligence fashion?

[00:10:59] Like, yeah, we're already there.

[00:11:00] Right.

[00:11:01] Is AI going to be able to do your dispatching for you?

[00:11:04] We're pretty close to there.

[00:11:06] In some places we are there.

[00:11:08] Is AI going to be able to handle triaging or high volumes of calls or requests and sort of move them to the right places?

[00:11:16] Like, yeah, we're pretty much there.

[00:11:18] Right.

[00:11:18] Level one.

[00:11:19] How far away are we from AI being able to handle password resets, you know, basic restart your computer, that sort of stuff?

[00:11:30] A year?

[00:11:31] A couple months?

[00:11:33] Two years?

[00:11:34] Like, yeah, that's the reality.

[00:11:38] So, like, it's not going to take your entire business overnight.

[00:11:45] Um, but you need to, if you're not in the position where you're ready to adopt as new technology comes out, it's going to hurt you, um, down the line.

[00:11:54] So, like, the worst thing you can do if you're scared of AI is run from AI.

[00:12:00] You have to embrace it.

[00:12:02] So, if I am an employee doing password resets or only generating proposals or something like that that's very specific that is kind of here today or about here today, what should I do as an employee?

[00:12:16] It's just skill up, you know, and owners of MSPs should be thinking about that and how are we going to retain these employees and reallocate them to resources that can better fit them.

[00:12:28] So, every MSP owner looks at AI projects or automation projects or whatever it is, efficiency-driven projects, is we're going to drive, we're going to automate this whole thing.

[00:12:38] But what are you going to do with the employee afterwards?

[00:12:41] And usually you need that employee on board to automate that whole thing.

[00:12:46] So, how are you going to motivate them for what their next job is?

[00:12:50] And it may look like MSPs grow 50% bigger with the same amount of people.

[00:12:57] It may look like MSPs stay the same size with, you know, 30% less people.

[00:13:03] It's your decision as a business owner where you want to go and how you want to handle that.

[00:13:08] And if your staff gets, you know, the inkling that you're in that latter category, they may not be very motivated to help you on that goal.

[00:13:18] And you may have a case of, you know, pursuing mediocrity or sort of like avoiding innovation.

[00:13:31] And, like, you know, I think it's important to invest in employees.

[00:13:37] And I think that there's a real opportunity right now for MSPs to adopt these newer technologies and continue to grow during and grow their business.

[00:13:48] And MSP business is a growing market and retain the people and, you know, hopefully pay them more.

[00:13:53] But if your employee figures out how to automate their job 50% and can do twice as much work, do they get twice as much money?

[00:14:03] I don't know.

[00:14:04] I don't know the answer to that.

[00:14:06] Could you give them 20% more if they do that?

[00:14:09] And then your team could see that?

[00:14:11] Like, oh, this person got a big raise and they got, you know, a new title and all this stuff because they figured out how to automate those things.

[00:14:17] I think that everyone wins in that scenario.

[00:14:20] Obviously, it's a very different level, but I feel like even not counting the Industrial Revolution, we all remember where there was generations of computer upgrades that were twice as fast.

[00:14:33] Yeah.

[00:14:33] And we could, I don't know if we got twice as much work done because we could play more games, but the computer was twice as fast.

[00:14:40] Or, you know, SaaS software or RMM or something like this.

[00:14:46] We're already unlocking gains in efficiency and how we do things.

[00:14:52] You know, standing still seems to be the biggest risk.

[00:14:56] But on the other hand, I feel like everybody's either, there's some fear, but also some degree of overwhelm.

[00:15:03] So how do I approach this as MSP with all the shiny things that are AI and all the buzzwords we could throw out to try to not bury my head in the sand, but to try to prepare, to try to keep up.

[00:15:16] Like, how do I even start to upskill myself to even figure out where do I go with my team?

[00:15:21] Yeah.

[00:15:22] I mean, I think adoption is the biggest thing and getting yourself to use it.

[00:15:26] I think if you're an MSP owner and you're not having, you haven't had that 10 hours yet, like you need to start there because that'll help you get a grip on what's real and what can be done and what can't be.

[00:15:39] So that's step one.

[00:15:41] Step two, you know, I think is to sort of do some ideation on your business and figure out where it could be.

[00:15:51] So you need that top down approach and that bottom up approach.

[00:15:55] Picking an improved technology is definitely a big part of it.

[00:15:58] It's part of where I am a lot of the time in the process, you know, as a AI platform that MSPs adopt and adopt for the customers as well.

[00:16:08] And I think that's a big part of it because you need to give people a place to go and avoid that shadow AI, which, you know, we talked about before.

[00:16:18] But you need to get your house in order, too.

[00:16:20] What are the policy changes that you need to make?

[00:16:22] What's an acceptable use policy?

[00:16:24] What does good look like in your organization?

[00:16:27] If someone and then you need to think about employee motivation, right?

[00:16:30] If you're trying to adopt this technology, put yourself in the employee's shoes and think what incentive do they have, if any, to tell you and to work on this sort of adoption stuff besides their boss telling them to.

[00:16:43] So what do you say to the people that say this is just noise and buzzwords?

[00:16:49] And I'm just because it's so noisy because there's so many buzzwords.

[00:16:53] I'm just going to ignore it for a year, three years, five years, something like that.

[00:16:59] I think that there's a big, you know, like I will, first of all, say that there is a lot of noise and there is a lot of buzzwords.

[00:17:05] And overnight, every company became an AI company, right, and slapped AI on whatever they're doing.

[00:17:11] And that's okay, right, because AI is integrated in everything that we do.

[00:17:19] You know, we offer a free training on the fundamentals of AI just to wrap your head around it.

[00:17:26] There's certification, too.

[00:17:27] So basically, the goal of the training is so an MSP could come in and leave with a foundational knowledge to have an intelligent conversation with the customer or with their employees about AI.

[00:17:40] Like, I think that's a great step forward.

[00:17:44] And you should at least do that.

[00:17:46] But, like, invest the 10 hours.

[00:17:49] Do it now.

[00:17:50] Because a risk of being left behind in an industry is huge.

[00:17:54] Like, I wouldn't want to be the person that held out for, you know, against VoIP or, you know, against high-speed Internet or, you know, against whatever new technology.

[00:18:07] And this one is progressing faster than any of those.

[00:18:10] By far, right?

[00:18:11] By far.

[00:18:12] Yeah.

[00:18:12] If you're still holding on to cybersecurity, you're already in trouble, right?

[00:18:15] Right.

[00:18:18] What are some of the interesting use cases?

[00:18:21] Or, I guess, yeah, let's start with that.

[00:18:25] What's some of the interesting use cases of the real world?

[00:18:27] Yeah.

[00:18:27] Because I think people are even having trouble imagining what I do.

[00:18:30] Like, I think it's throwing in a co-pilot or Bing or Chattapiti and, like, it spits out some stuff.

[00:18:36] And if you're listening, my apologies, but you're not very good at prompts.

[00:18:40] So you're throwing in junk.

[00:18:41] You're getting out junk.

[00:18:42] Yeah.

[00:18:43] And so people are like, I can kind of see where it helps me with this email.

[00:18:47] Yeah.

[00:18:47] But they don't know what to do.

[00:18:49] I feel like a lot of folks, they don't, like, I can't really put a price.

[00:18:53] Like, maybe I'm willing to pay the monthly fee to write better emails or something.

[00:18:58] But people are having, I think, trouble imagining what can we do now.

[00:19:02] So what are you seeing, like, in the real world?

[00:19:04] Yeah.

[00:19:05] I think focusing from an MSP perspective is probably a good place to start.

[00:19:09] And then, like, in our platform, we have a couple hundred templates and sort of pre-written prompts ready to go where you just fill in the blanks across all businesses and sectors.

[00:19:21] I think that there's a lot of use cases that people have already discovered around, like, marketing, like you said, like, email writing, document summaries, you know, meeting recap, that kind of stuff.

[00:19:32] And it's sort of, you know, it's cool the first time you see it, but it's like, okay, like, is this enough for me to really go forward?

[00:19:39] What else can it do?

[00:19:42] And, you know, I've seen a lot of MSPs.

[00:19:45] They'll take, they'll go through their onboarding procedures of new customers, the entire sales flow.

[00:19:51] So everything from post-sales meeting notes, generating a proposal, checking a proposal against resources, going through pricing scenarios.

[00:20:02] I've seen a lot of that being automated through AI templates or automations in our platform from our partners.

[00:20:10] And they're, you know, uploading examples and they're giving explicit instructions on how to do it to the point where you just, you know, paste in the notes from the meeting or type what the customer wants in plain English.

[00:20:23] I think that there's, too, a lot of opportunity around, in that same onboarding flow, new customer machine, right?

[00:20:33] You dump some software on it, right, and you get back some list of processes that are running, right?

[00:20:39] And, you know, we have a thing that somebody else made it.

[00:20:44] You know, one of our partners made it and shared it with the community, thankfully, where it looks at the processes that are running and then looks at an approved list of technologies that the MSP supports and then pulls out all the processes that are not on the approved list.

[00:20:59] And then gives their, the AI gives its best guess on what that exe file that's running actually is.

[00:21:07] Like, you know, we found QB.exe running on your computer.

[00:21:11] It seems like you're running QuickBooks.

[00:21:13] Do you need a support package on QuickBooks server or whatever?

[00:21:19] Can we take this off the machine?

[00:21:21] You have three antiviruses running.

[00:21:22] You know, we're going to take these other ones off.

[00:21:26] There's a lot, like the highest impact in the lowest amount of time I've seen is RFP and grant proposals.

[00:21:35] Because what's happened is the AI has made the grant proposal process or the RFP response process so short.

[00:21:47] If you have all of your information ready to go, you can do what would previously take a week, like two hours or 20 minutes or something like that.

[00:21:56] So, like you see in, like the people who are doing that, there's a huge influx of applications for these RFPs and those grants.

[00:22:03] And sort of like if you're not doing that, like it's going to be a lot harder to win these types of things.

[00:22:08] A lot around standard operating procedures.

[00:22:11] I think in an MSP, everyone's always trying to clean up their SOPs and to go in and, you know, take the first line, the subject of it, this is the procedure steps, all of that.

[00:22:20] Versus just like, you know, taking a voice note and saying, hey, here's what it is.

[00:22:24] Here's the template for the procedure.

[00:22:26] You know, give me the first draft of it and I'll edit it.

[00:22:28] Like that's an easy win.

[00:22:31] And we also, we have an AI phone agent that, you know, can answer the phone and you can take phone calls and all that.

[00:22:39] MSPs are using that for after hours a lot of the time.

[00:22:42] Hey, what's your name?

[00:22:44] What's the business?

[00:22:45] Okay, what's wrong?

[00:22:47] Is this causing business downtime?

[00:22:49] Okay, it is.

[00:22:50] I'm transferring you to the on-call employee.

[00:22:53] It's not, I'm creating a ticket.

[00:22:55] We'll take care of it for you.

[00:22:57] Thank you very much for calling, blah, blah, blah, blah.

[00:22:59] Like AI is great for that.

[00:23:00] Would I have it take every call right away if your employees are used to, you know, your whole value prop is we answer the phone right away and you talk to a human?

[00:23:12] Maybe not, you know, but you can slowly work it in and we may get there, right?

[00:23:18] In like two years or a year or five years.

[00:23:21] I don't know.

[00:23:21] But it's progressing fast.

[00:23:23] Yeah, yeah.

[00:23:24] Interesting.

[00:23:25] Interesting.

[00:23:26] What are the biggest hurdles you see as MSPs are adopting?

[00:23:31] Like they're coming in like, okay, I have a use case or two and I want to kind of do it, but I'm not quite sure where to start.

[00:23:38] You know, maybe they'll land on your platform, et cetera.

[00:23:40] But what's the biggest challenges you see?

[00:23:43] It's always, almost always trying to boil the ocean and trying to do too much.

[00:23:48] Getting fixated on a very complicated use case that's hard to get everyone on board for instead of, you know, going for the low-hanging fruit first.

[00:23:57] So get the quick wins.

[00:23:59] Get the quick wins.

[00:24:00] Okay.

[00:24:01] Give me an example of overly bloated use case versus a quick win.

[00:24:08] I want to have AI look through every pass ticket in our PSA and then make suggestions and create documentation on how to solve this and give the technician the IT glue link and then the link to the pass tickets that were similar on day one.

[00:24:30] It's going to be hard to roll that out right away.

[00:24:31] Like the technology is getting there that will be easier over time, right?

[00:24:35] Maybe next year, right?

[00:24:37] That stuff will be able to do out of the box, but we're not there yet.

[00:24:42] Versus I want AI to review my ticket and tell me the status priority if it's high or low and get the subcategories correct.

[00:24:51] Save you a couple minutes, right?

[00:24:53] Yeah.

[00:24:54] You don't have to do a ton.

[00:24:55] Yeah.

[00:24:56] I think that hits the nail on the head really well because I think, you know, it went from like, it can't, like, Siri can't get most people's names right.

[00:25:07] Yeah.

[00:25:08] You know, so we're used to like, that's the thing.

[00:25:10] And then chat GPT all of a sudden can answer seemingly every question.

[00:25:13] Now, sometimes it answers them with interesting information.

[00:25:16] Yeah.

[00:25:17] But I think, you know, that created this, well, can't it do everything, right?

[00:25:23] So how do I guard against that?

[00:25:29] How do I make sure I'm not coming in with like a boil the ocean approach?

[00:25:33] You get your 10 hours in.

[00:25:35] Yeah.

[00:25:35] I mean.

[00:25:36] Where do I start with my 10 hours?

[00:25:38] Where do you start?

[00:25:39] Yeah.

[00:25:39] You know, so usually it's getting prompted on one of those low hanging fruit use cases.

[00:25:46] Like, hey, listen, look, you're in charge of, you know, writing, you've been working on these SOPs, right?

[00:25:54] Here's a template.

[00:25:55] Here's a, you know, an app in hats to get started with it.

[00:25:59] You know, mess around with it and try to create a couple.

[00:26:01] And then here's the chat too and see what else you could do with it.

[00:26:04] Like, that's a good way to get started.

[00:26:06] But it, and it doesn't have, like, you can do it in Copilot, like whatever, or ChatGPT.

[00:26:11] Like, go try it out and see if you can get it to do this.

[00:26:15] Or provide a couple examples of our past webinar, the, you know, the marketing materials from them to the AI and, you know, ask it to help you with the new one.

[00:26:26] You're writing a big report or proposal.

[00:26:28] Ask the AI for feedback on it and suggestions for improvement and grammar and spelling and all of that.

[00:26:34] Like, that's a good way to get started.

[00:26:36] Uh, where someone can sort of go in and then, then it's, uh, oh, well, what else could I do?

[00:26:42] You know?

[00:26:43] What's your, um, thoughts on are we going to, it's already progressed fast.

[00:26:50] Should, are we going to expect every week incremental improvements?

[00:26:54] Or do you think it's just nothing, nothing, nothing, bang?

[00:26:57] All of a sudden, you know, we're talking with Jarvis or something.

[00:26:59] So, so the, uh, the underlying technology has been increasing, um, uh, in innovation faster than everything else.

[00:27:10] So, so if the LLMs, so that's, uh, GPT-4, GPT-3.5, uh, Claude 3.5 Sonnet by Anthropic, like those are the LLMs.

[00:27:22] If they didn't get any better today, it would take another five years before our industry caught up to the use cases for actually using them.

[00:27:35] Um, yeah, so it would take, you know, all the vendors in here, you know, another five years for their roadmaps to play out.

[00:27:42] Okay.

[00:27:42] This feature can be done like this, the PSA vendors, all that.

[00:27:45] Like there's a lot that needs to be done and it's not like, we're going to keep seeing incremental improvements on the underlying tech and the LLMs.

[00:27:53] But, um, in terms of like integrating it into your business, there's a lot of work we still need to do on the existing software side, um, to, to build the technology in.

[00:28:06] It'd be like, you know, a new programming language came out that suddenly way faster than everything else or, uh, and then everybody has to rewrite their code that way.

[00:28:15] Or like, uh, you know, uh, PS5 came out and you've got all the games for PS4 and you need to remake them for PS5.

[00:28:23] So like we're still catching up and there's a lot of catching up that has to happen.

[00:28:28] Hmm.

[00:28:29] What's, uh, what about things like scripts or code and coding?

[00:28:36] Yeah.

[00:28:37] Obviously there's been some pretty big improvements in what AI can build.

[00:28:40] Um, so as an MSP, is that something that I can use today?

[00:28:45] Uh, yes, absolutely.

[00:28:47] I mean, uh, AI is very good at writing PowerShell scripts or looking at code.

[00:28:51] It will not, um, be 100% correct.

[00:28:55] And if you don't know what you're doing, blindly committing PowerShell scripts and running them is not a good thing to do.

[00:29:01] Um, some examples of how you can improve that is you have an existing script that you need to change to go another way.

[00:29:08] Um, and you, you feed it, you say, Hey, listen, I have this PowerShell script, uh, for offboarding a user.

[00:29:13] Uh, it currently, um, uh, deactivates their Microsoft account.

[00:29:20] Uh, and then, you know, whatever, uninstalls this and does that.

[00:29:24] I also want it to remove the license in Microsoft.

[00:29:28] I don't even know if that's possible, but like, I also wanted to do this.

[00:29:30] Can you edit it and explain why, uh, those changes are made?

[00:29:34] You're going to get a lot better off than write me a script that does, you know, X, Y, and Z from scratch.

[00:29:41] I can tell you speak prompt language the way you were being very specific there.

[00:29:44] Yeah.

[00:29:45] Because I feel like the normal one would be like, I want a script that does onboarding.

[00:29:48] What does that mean?

[00:29:50] Onboarding what?

[00:29:51] Onboarding to what service?

[00:29:53] What users?

[00:29:53] Where?

[00:29:54] Which, yeah.

[00:29:55] Um, and I find that it's interesting because, you know, we grasp that context as humans and, uh, we have to, that's learning its own language, right?

[00:30:04] Prompt engineering.

[00:30:05] Um, is that something that you think the average MSP needs to spend more time on?

[00:30:10] Yeah, absolutely.

[00:30:11] I mean, we have prompt engineering certifications.

[00:30:13] Uh, they're free for any MSP if you want to check them out.

[00:30:16] But it's, it's important to do it yourself to learn how to get better.

[00:30:22] And, um, like I wouldn't just go do prompt engineering courses.

[00:30:26] I'll try it first.

[00:30:27] Because then you're like, oh, wait, I was doing it this way and this response was so much better because I learned how to do it this way.

[00:30:35] And, like, concepts like chain of thought prompting, um, that, uh, you know, can perform better where you sort of ask the AI first break it into the steps that you think this should happen and then perform those steps.

[00:30:47] Um, you can get better results than just saying do this.

[00:30:51] Um, so, yeah, I think it's, I think it's important.

[00:30:54] I think it's going to be increasingly important because everyone sort of doesn't know what they're doing.

[00:30:59] And when small businesses and medium-sized businesses need help with a new technology, they turn to their MSPs.

[00:31:07] So, I think, um, offering that service, whether or not your clients instinctively turn to you or not, is important and could be a big moneymaker for, um, people.

[00:31:17] And also just knowing it to adopt internally.

[00:31:21] What if I'm more bullish on it or I'm already doing these things, like, after the 10 hours?

[00:31:26] Like, what, if I want to be the, uh, the leader, you know, one of the leaders, uh, in the MSP space, what do I do?

[00:31:33] Uh, identify, so, work on adoption your own organization first.

[00:31:37] Uh, I found that the MSPs successful selling AI to their end customers, uh, have adopted it internally the most.

[00:31:44] There's a direct correlation between sales to end customers and internal adoption.

[00:31:48] Um, so, to do that, what you're going to do is, uh, you're going to, uh, identify the leaders in your organization.

[00:31:56] Um, you're going to identify use cases that are the low-hanging fruit, maybe five to ten, that everyone can get started with easily.

[00:32:03] And, you know, maybe half of them everyone can use and some of them only half the people can use.

[00:32:09] Um, and you're going to pick and approve technology and you're going to roll it out to everyone.

[00:32:13] And you're going to say this, you know, we have this, you can use it, here's how we suggest getting started.

[00:32:18] Um, and then, uh, in addition to that, you, as you move through the course of maybe one month, two months, three months, six months, um, you move to the next phase.

[00:32:29] Which is what are the sort of templated or niche use cases that maybe work for one department, but not everyone else.

[00:32:36] Let's build those out.

[00:32:37] In addition to all of this stuff, you need to get your house in order.

[00:32:40] You need to get your policy straight, your acceptable use, uh, like your terms and conditions as an MSP, all of that stuff done.

[00:32:49] Um, so you move on to the, the templated.

[00:32:51] Um, and then finally, once, you know, that sort of phase and you have a good handle on it, that's when you can work on those more ambitious projects or, uh, automate an entire area of business just for one specific, uh, person and focusing on them.

[00:33:06] Um, but I think if you try to jump to the end, you're, you're going to fail, uh, or maybe you won't fail, but it'll be a lot harder than if you bring the whole org along.

[00:33:15] What about getting your data ready?

[00:33:18] Yeah.

[00:33:19] I mean, that's a big thing.

[00:33:20] I think a lot of people, um, uh, a lot of MSPs are sensitive to AI data readiness because of AI in the Microsoft ecosystem.

[00:33:29] So you look at Copilot, uh, it's AI that's integrated into, uh, office 365 and it uses the Microsoft craft API, um, to query, uh, files that it can have access to on SharePoint.

[00:33:42] Um, so if you are an employee and you might have access to some shared drive, but you've never accessed it before, like, yes, you know, you need to fix that as an MSP.

[00:33:55] But if all of a sudden the AI can answer questions about it, might accidentally pull information from those files that employees didn't know existed.

[00:34:01] That's a huge problem.

[00:34:03] So, um, I think from data ready.

[00:34:05] Last thing we want is we roll out AI and then people are like, Hey, what's the C, what's Bob's salary?

[00:34:11] Yes.

[00:34:12] And all of a sudden you get an answer.

[00:34:13] Yeah.

[00:34:13] That you didn't set the permissions.

[00:34:14] Well, now you've given away the whole comp data for the whole company.

[00:34:18] Yeah.

[00:34:18] And, you know, you can turn a lot of that stuff off.

[00:34:21] Uh, and I would suggest when getting started, you know, in our platform, we follow like a least privileged AI model where you sort of know what's going into it and you know who you're sharing it with.

[00:34:30] Uh, which I think is important and sort of the way a lot of people want to operate.

[00:34:36] Um, I think that there's a misconception that more data is better for AI because in many cases it is right.

[00:34:43] When training, uh, uh, uh, large language model or, uh, something like that, but garbage in, garbage out, right?

[00:34:51] How many copies of the same PowerPoint presentation do you have where the first six copies are like the drafts, right?

[00:34:59] Like you don't need all of that in AI doesn't need access to your six drafts.

[00:35:03] Right.

[00:35:04] So being, being more, um, I think if you're going to roll out copilot, uh, with the, um, sort of graph API enabled and, and SharePoint and all that stuff enabled, you need to do a lot of work on data readiness.

[00:35:18] And it's important to do that.

[00:35:19] And I think MSPs are great to do.

[00:35:20] I think it's good work.

[00:35:21] Um, just, you know, build project hour for it.

[00:35:24] So you're not getting crushed, but.

[00:35:26] What role do you think MSP should play in helping the owner or the employees of the company train, um, put in policies, maybe deal with fear?

[00:35:39] Yeah.

[00:35:39] That's one of the things that they're probably, you know, dealing with right now.

[00:35:42] So, cause we've talked about the MSP owner and maybe their employees worried, but obviously the, your customers, cause you know, that, that, that team and your, your, your, your, the CEO of that company, if that's who you're working with.

[00:35:53] Yeah.

[00:35:54] Yeah.

[00:35:54] Yeah.

[00:35:54] They, they all have their own fears.

[00:35:56] I mean, I bet my, you know, future on MSPs being able to do this, right?

[00:36:01] Like I started a company with that thesis that MSPs would be, um, the most well-suited group of people to serve small and medium businesses for this emerging need.

[00:36:10] Um, and I think that you can do it.

[00:36:12] Um, I think that it's the natural progression of things.

[00:36:16] And, you know, I wouldn't want to be running, um, an MSP business where no one of my customers,

[00:36:24] have adopted AI in three years because it's a huge threat, right?

[00:36:28] And, and, you know, if, if we're, if we're in agreement or at least in semi, uh, suspicion that, um, uh, AI will have disruption in industries, it's the companies who are going to adopt them three, four years down the line that are going to be the ones that have benefited from it, uh, instead of the ones that were disrupted by it.

[00:36:49] So how would you start that with your customer?

[00:36:52] I mean, do I, do I start having conversations about fear or policy or because at a certain point, then you're talking about maybe we can help you build something, but there's, it seems like there's things before that.

[00:37:03] Cause it seems like, you know, if you have this concern, if you're, imagine you have significantly less knowledge.

[00:37:10] Yeah. I mean, I think, um, you know, like, like maybe it's time to turn your, you know, 10th cybersecurity webinar into, you know, the topic of AI and address this stuff and, and become a thought leader in it, uh, to your customers.

[00:37:25] Uh, a lot of the MSPs I work with are having a lot of success prospecting new customers and, and, um, you know, marketing themselves by, um, being a thought leader in AI inside their community, the same way that they did with cyber, right.

[00:37:39] And the same way they did with it and, you know, different it trends that have happened over the, you know, past 20, 30 years as MSP business model has emerged.

[00:37:48] Right. Um, so I think that's a big step forward.

[00:37:51] I think the other is just, you know, being ready, like we can give you a quick assessment.

[00:37:56] Um, we can give you approved tools and we can help you with your, uh, AI, uh, policies or at least tell you where, what you need and, and what you need to look for.

[00:38:04] Um, those are, that's like a great, uh, sort of, uh, first go to market for, uh, MSPs.

[00:38:11] Uh, I think like the, like, Hey, maybe we'll turn, you know, AI on for a couple of users in every company or whatever.

[00:38:21] Like I think you're going to be chasing yourself and it's going to be the side thing that never happens.

[00:38:26] If you get methodical about it, this is our AI offering.

[00:38:28] This is how we go to market.

[00:38:30] Um, you'll have a lot, a lot more success and you can address those fears, um, and sort of guide business leaders along the way.

[00:38:37] But you need that, um, sort of talk track with them.

[00:38:40] Interesting.

[00:38:41] So let's say I, I buy in, this is going to be a big change.

[00:38:45] I put in the 10 hours.

[00:38:46] How much of my time or of my staff should I say, let's allocate some time because it's obviously fairly early.

[00:38:55] Yep.

[00:38:56] And, uh, it's a very shiny object.

[00:39:01] And MSPs, we love shiny objects.

[00:39:03] Yeah.

[00:39:03] So how much time should I be devoting on this versus all the other competing priorities?

[00:39:09] Should I put five, 10 people?

[00:39:11] Should I put one person on it full time?

[00:39:13] I mean, like just from an allocation perspective, what, what are your thoughts?

[00:39:16] Yeah.

[00:39:16] I think it's something that happens naturally as you do other work too.

[00:39:19] So if you're working on a, um, uh, PSA migration or something like that, having AI support you along the way is going to help you do that faster and better.

[00:39:29] Right.

[00:39:30] You're feeding the documents in from one to another.

[00:39:32] You're looking at error logs, why it's not working.

[00:39:35] You're asking how you should set up, um, you know, the, the field mapping here, my fields in this system, here are my options in this one.

[00:39:43] Uh, how should they match up?

[00:39:44] Uh, and why, like, I, I, I don't think it's like, okay, we're going to set aside, you know, three hours away today to work on AI and, you know, like go back and forth with it.

[00:39:54] Uh, I think it's more like, like, uh, start using it, uh, in other projects and, and how you do things.

[00:40:01] Um, think of it like, you know, almost like Google search.

[00:40:04] Like, you know, you wouldn't tell someone, okay, go spend, you know, two hours learning Google search, right?

[00:40:09] Like you're, you're gonna, you're gonna, uh, use it to, uh, as a tool, um, and start putting it in your toolbox.

[00:40:17] Well, on that note, it's been a little while, right?

[00:40:20] But since Google was a new thing, but when that was new or when certain versions of Windows were new or whatever it is, there were the techs that couldn't help themselves.

[00:40:28] They spent their nights and weekends, right?

[00:40:30] Yeah.

[00:40:31] So do I need to be looking for that person since this is such a new paradigm and new technology?

[00:40:37] Uh, I think that they exist and that they're important for adoption, but I think we over-index on that person and leave the rest of the people in companies behind.

[00:40:48] Um, and it causes AI, uh, adoption to fail because you have that person go from level one to level 10 while everyone else is still at level zero.

[00:40:56] Yeah, that makes a ton sense.

[00:40:58] Don't invest only in your, in your star employee.

[00:41:01] It's, it's, that's like how it happens though.

[00:41:04] Like people teach themselves AI from using it and learning it.

[00:41:08] And then it's like, okay, now I've, you know, I've done this for a while.

[00:41:11] Let me actually take this prompt engineering thing and, and learn a little bit about it just to level up how I'm doing it.

[00:41:17] Um, so like employees will naturally, uh, their skillset and knowledge on interacting with AI will, will happen over time.

[00:41:25] Uh, I don't think it's something that you can really push down from, from the top.

[00:41:30] So what's, uh, what's one thing you're looking forward to?

[00:41:34] I'll answer two, two ways then.

[00:41:36] I think for AI in general, I'm really excited about the advancements in, uh, medicine that are sort of happening, uh, very quickly behind the scenes.

[00:41:45] New drugs are getting invented, uh, novel drugs to treat specific people and specific things, uh, automated testing of things.

[00:41:52] I think it's going to be really great and really positive for humanity.

[00:41:54] Uh, I also like, I know there's going to be a lot of pain to get there, but like, um, uh, uh, autonomous vehicles.

[00:42:01] I think like there's a lot of vehicle car crashes and there's a lot of people who, uh, especially in the U S, uh, can't drive and they sort of get, uh, you know, their quality of life goes way down.

[00:42:12] I think that that, those are two really positive things that, um, uh, AI, uh, can help.

[00:42:17] Um, and then like in my own product, you know, we're, uh, working a lot to, um, integrate external systems into, uh, our platform right now and being able to run, uh, AI and workflows and being able to have it sort of do, uh, chains of prompts and chains of commands.

[00:42:36] Uh, and it's a big change for us, uh, but something that we've always anticipated and always wanted to do and something that I'm very excited about.

[00:42:42] When do you think the next big jump or leap might be?

[00:42:46] Obviously it hasn't occurred to setting you up for failure here, but what's your wildest guess?

[00:42:51] Um, I think this year was the year of experimentation and next year is the year of implementation for AI.

[00:43:00] I think we're going to see mass adoption in maybe like, you know, maybe it's the top 20, maybe it's the top 50, I don't know, percent of, uh, clients, software and services.

[00:43:12] Um, but I think it's going to, I think it's going to come on pretty strong because there is just so much happening and it's, it's real innovation.

[00:43:20] Um, and there's like the top 1%, right.

[00:43:24] In terms of technology and adoption are all doing this stuff.

[00:43:27] Um, and everyone else, like if you look at the fortune 500, they're all running AI pilots with like 10 different companies trying to figure out how they're going to integrate it in their own business and their own software.

[00:43:38] Um, and sort of next year is the time to make the real decisions for a lot of these companies.

[00:43:43] Any, uh, on the other end of the spectrum, any bets when we're going to see AGI?

[00:43:47] Uh, well, so depends.

[00:43:50] So first of all, as anytime anyone's asked this, they always say, well, what's the definition of AGI?

[00:43:55] Uh, a common, um, definition is, uh, AI that is smarter than humans.

[00:44:03] Um, and generally speaking like 99% of times, right?

[00:44:09] So, so super intelligence is smarter than the smartest human.

[00:44:12] Uh, artificial general intelligence or AGI is smarter than, uh, most humans.

[00:44:18] Um, uh, pretty much all the time.

[00:44:21] Uh, I think we already have, um, elements of it.

[00:44:25] Like what, like there's, there's things that AI can do better and faster than, um, humans already.

[00:44:32] Um, but like, I don't know.

[00:44:38] I, I really don't know.

[00:44:40] I, I think it could be a couple of years.

[00:44:41] I don't think AGI is, is, um, like, I think we still have a lot more years of, of, you know, using and building software and, and there's always the people element of it.

[00:44:52] And so, um, you know, I think it's this, this thing that nobody really knows what it is and we're scared of when it's coming.

[00:45:00] Interesting.

[00:45:01] I wouldn't bet against the model developers from innovating, uh, faster though, because they seem to keep doing that.

[00:45:07] Yes.

[00:45:08] They seem to have a lot at stake.

[00:45:10] Yeah.

[00:45:10] And there's billions of dollars being poured in.

[00:45:12] You mentioned chain of thought prompting and other prompting techniques.

[00:45:16] Is, is it prompting techniques?

[00:45:18] Is it the rise of agents?

[00:45:20] So what do you, what's the trend now or what's the thing to be looking at?

[00:45:24] The kind of like, it's the new, new cool thing, new, interesting capability.

[00:45:28] So, um, or assistance.

[00:45:31] Yeah.

[00:45:31] I think like, uh, a year ago, AI was really good at, uh, responding in just text, um, and sort of not in computer language, like, like a JSON format or like something that a, uh, uh, uh, system could parse and sort of take instructions from.

[00:45:45] Uh, now it's a lot better at that.

[00:45:47] So we're seeing this next wave of like, uh, AI being able to make decisions in, and, and send API requests and do that.

[00:45:54] So I think that's the next wave of things.

[00:45:56] You can call it agents.

[00:45:58] You can call it workflows.

[00:45:59] Um, but that's definitely like where a lot of time in software is being, uh, developed right now and where we're going to see the biggest changes soon.

[00:46:09] Awesome.

[00:46:10] Jimmy, uh, let everybody know how to find you or hats AI.

[00:46:14] Yeah.

[00:46:15] Um, our website's hats.ai, H-A-T-Z.ai.

[00:46:19] And you can connect with me on LinkedIn.

[00:46:20] Just search Jimmy Hatzel, uh, H-A-T-Z-E-L-L, two L's in Hatzel.

[00:46:26] Um, and, uh, yeah, reach out to me.

[00:46:30] Love to connect.

[00:46:30] Well, thanks for making time to speak at Build IT Live.

[00:46:33] I know Pack's schedule presenting.

[00:46:35] So I appreciate it.

[00:46:36] And thank you for being on MSP Mindset.

[00:46:37] Yeah, my pleasure.

[00:46:38] Thank you for having me.