WEBVTT

00:00:00.000 --> 00:00:31.850
Right now, most videos are focused on Claude Code building your shiny new things, but there is an entire untapped market that is far more lucrative than that and actually a lot easier to manage. In my twelve years of consulting with Fortune five hundred and one hundred companies, this has made me a small fortune because it's always going to be around no matter what tech you are using. So in this video, we're gonna focus specifically on AI maintenance and how you can use it to make a ton of money in your niche. First up, we need to set the scene though with a little bit of doom and gloom. So there are lots of stats floating around online, one of them being from Deloitte that says 40% of AI agent projects are gonna be canceled by 2027.

00:00:32.010 --> 00:01:06.465
There are many of these stats out there, and they're all based on different things mostly around people starting off in the wrong way. There are several reasons for this happening that we'll get into very soon, but one of them was because most of everything that people have right now is built on the wrong foundation. As you can see from our little house over here, that most of this was built on hype and trends and shiny tools that the business didn't actually need but they needed because a whole bunch of kiddies online on YouTube told them it was the second coming of Jesus. But unfortunately for them, it turned out to just be hype because those people had never worked in businesses. So that's what happens when you firstly take advice from the wrong crew,

00:01:06.705 --> 00:01:10.865
but secondly, when you focus on the shiny stuff instead of the strategy, the architecture,

00:01:11.025 --> 00:01:20.170
and the fundamentals behind why you're actually about to do the thing that is gonna change your entire business. One of the reasons for any AI engagement, I always lead with the audit specifically

00:01:20.250 --> 00:01:33.855
for this strategic reason. We wanna make sure that whatever we are giving these people is actually gonna give them some return on investment and that it's an actual problem in their business as opposed to this hype that we're looking at with this little broken house over here. So when we're looking at maintenance,

00:01:34.015 --> 00:01:37.135
we pretty much do the exact same thing. We just do it through a different lens.

00:01:37.375 --> 00:01:46.440
We start off with the audit or readiness. Now historically, whenever I was doing this with cyber security or cloud, whatever the tool was at the time, it would always be focused on readiness.

00:01:46.440 --> 00:02:16.270
When people were all leaving their on prem to go to the cloud, it would be cloud readiness assessment and things like that. So you can change this word from audit because some people don't like it. You call it whatever the hell you want as long as it makes sense toward people being ready or improved or whatever it is they need. But the idea behind this is that you are going to give them the strategy that they need specifically for all of the things that are broken because you don't want to make the same mistake that the guy before you made where he just went in there and yoloed his way and burnt his name. You want to make sure that whatever these people have going on in their environment gets solved properly.

00:02:16.510 --> 00:02:39.805
So we do the audit because that exposes all the avenues or the six pillars that we can use to make money from this situation. And we're gonna dive into each one of these six pillars down here to understand what type of money we can make in them and how we would actually position ourselves so that we come across as something valuable to the business as opposed to somebody who's just reading off of the script. More importantly though, there is money to be had in the audit. There's money to be had upfront in the maintenance,

00:02:39.885 --> 00:02:43.485
but then there is also a retainer option over here. So there are three different opportunities

00:02:43.565 --> 00:02:51.040
that you can use to get more money. And even more so, you don't have to be an expert in every single one of these things. You can pick and choose and add as many as you can possibly cater for.

00:02:51.440 --> 00:03:20.010
The more you have, the more valuable you're obviously going to be and the more money you can charge. But you can also narrow this down and go all in into one thing as you'll soon see. So first up, we have pillar one, and this is migrations and upgrades. And this is really, really lucrative because of how the landscape has drifted over the last year. If we look at where we were a year ago, we were busy doing n n n make that basic plumbing with AI plugged into it to go and do these cool things that now Claude can do in, like, twenty minutes. So the whole architecture behind how businesses are running their operations

00:03:20.250 --> 00:03:32.105
has significantly changed because of AgenTic and MCP and various other things that go into the modern day workflow. So for pillar one here, you would come in, you would assess their environment, and figure out just how old

00:03:32.265 --> 00:03:45.900
their current workflows are, how much they're costing them, as opposed to how much simpler they could be, how much more advanced they could be that would lead to other branches of making their business easier, things like that. It's all about breaking it down into what is still working

00:03:46.060 --> 00:03:59.985
because maybe it's still working for some of the things. You don't want to just rip something out because you need to throw the word agentic on it. That's not what I'm suggesting here. You use the audit, again, the strategy to define what isn't working, what is legacy by modern standards,

00:04:00.225 --> 00:04:02.785
and then the after period is what can be augmented

00:04:02.865 --> 00:04:14.320
to make their business even better, more performant, more cost efficient, to make their staff even better, things like that. That is part of this migration and upgrade plan, not simply going in there and destroying something because it's outdated.

00:04:14.320 --> 00:04:19.040
But one thing I can guarantee you is a lot of these original processes over here are massively

00:04:19.040 --> 00:04:24.960
non secure. They have massive compliance problems. They're missing a ton of context for these systems to actually be beneficial.

00:04:25.545 --> 00:04:32.345
So when you look at the after picture over here, where we're kind of building on this AI operating system model that we've been talking about on this channel,

00:04:32.665 --> 00:05:14.690
it's way better for a business because we can stash skills, we can have knowledge, we can essentially capture everything for each member of the team and have it augmenting the role of the person who is using it. Because sure, this entire landscape has changed, but there are a lot of other things that form part of this landscape or any environment that we're working in. And that's what brings us into pillar two, which is performance and cost. And this is pretty straightforward. Right? People got these original systems in there. They cost x amount of money. Are they running the best way that they could? Probably not. Have they ever come in again to fine tune the model to change any of the prompts that were originally in there, pull things out? Probably not. So many businesses are probably still paying for things that they don't need or things that are massively overpriced or inefficient by today's standards.

00:05:14.930 --> 00:05:31.025
So your goal would be to come in there and sort that problem out. But again, it's not just about the tools here. It's about the strategy behind the entire scenario of the business in order to make it more performant and more cost effective at what it's doing. So when you look at the after state, even if it is using modern agentic workflows and things like that,

00:05:31.425 --> 00:05:51.165
you've seen how Anthropic is cutting down those usage limits. AI is going to get more expensive. So part of your role here is to come in and make sure that any skill or any workflow that is running is using the correct model. It's not just all thrown on Opus running the most expensive thing possible with a really long skill chain that's also inefficient. Your idea would be to shine your expertise

00:05:51.405 --> 00:06:19.135
on the floors in their system as a whole. Figure out how you can make it cheaper, how you can make it run faster with lesser models, more efficiently with lesser models, things like that. All ties in with pillar one as well. And just to note something on this for any developers out there, you guys have an absolute gold mine in front of you because everyone who's been vibe coding their apps and throwing in all these agentic things without knowing a single thing about coding, you can obviously go in there as the developer who actually knows what he's talking about and fixes up all of their code, all their broken workflows,

00:06:19.215 --> 00:06:42.870
and that ties in with the performance and the cost and the upgrades and the maintenance of this web app or whatever it is that they built for them because you guys actually know what you're talking about and you know what efficiency looks like. So if you are a developer and you're looking to branch out into something like that, this is definitely one of the ways to go. Few of my friends are doing this right now and making a small fortune doing it. Next up, have pillar three, which is monitoring and observability, and for some reason, this is a very non talked about thing on YouTube,

00:06:43.110 --> 00:06:54.665
but it's vital because all those things that have been built, if no one was ever monitoring them, how the hell would you know if this thing is working? How would you know if your customers are getting the experience that they need? But more importantly, how do you know that whatever you built

00:06:54.985 --> 00:07:01.920
is doing what it's supposed to be doing in the best possible way? You don't know what's failing, when it's failing, or how it's failing, or why it's failing.

00:07:02.640 --> 00:07:30.430
Those things are, again, how we feed into the upper layers of these pillars. Pillar two, cost efficiency. If you had observability, you could see how many skills run, how often, what models they're using, how often they fail, how often they hit the cache, things like that. That all helps you understand it. So it's not just about your client understanding this. Most of the time, they don't really care about this. It would be whoever is in charge of the tech and the monitoring, of course. But having this layer helps you understand those sorts of things. And when you understand those sorts of things, you can make your life a hell of a lot easier.

00:07:30.590 --> 00:07:51.625
So this panel that you're looking at right now is the command sense that I've built specifically for my AI operating system. The video is on the screen now if you wanna check out how to do this for yourself. But with this, we have all of the observability that we need for any Claude environment so I can make decisions based on actual data and not some form of YOLO. I know exactly what my skills are doing when they're doing it. I know the security of my entire AI operating system environment,

00:07:51.785 --> 00:08:14.915
same for how much context I'm always using. I understand my MCP service, everything that goes into it. So when I need to make a decision on whether I need to change something or optimize something, I have the data to prove it over the last thirty, sixty, ninety days, whatever I might need. It all lives inside here. So for you having this as the consultant who goes in there to try and fish out all this information to be the maintenance guy, you need to set up some sort of observability,

00:08:14.915 --> 00:08:32.460
whether it's taking this dashboard from the video that I just spoke about, or if you're in a much larger business, of course, you're gonna use something more established. You wouldn't be working with someone's personal AI operating system here. You'll be using something like Braintrust or Langfuse or Helicone, although I think they just got bought by someone. But the point is you would have a more formal point of observability

00:08:32.620 --> 00:08:49.235
that would link into their larger infrastructure. Because as you see over here, when we understand the quality of the output that we're getting from the AI, it means that we can make the right decisions based on that objective data. On to pillar four, which is security and threat patching. Again, ties into the whole thing because every little system out there needs security,

00:08:49.635 --> 00:09:28.980
and my god, this is gonna be so valuable as a niche to be in, AI security, anything related to it, especially when mythos and all of the higher models get released. Cyber attacks are just gonna come up more and more often. More importantly though, the vulnerabilities that get exposed are just gonna explode and obviously people out there who aren't technical, they're not gonna be reading this stuff, they don't stay up to date. If they don't have anyone in house who is taking care of this sort of thing, you would be the person to come and do that for them. And I can guarantee you that every single system that has been built in the last year has some form of security issue in it right now because everyone again who was building this sort of thing didn't factor this kind of stuff in. More importantly, people who went out there and yolo their way into OpenClaw and other such solutions,

00:09:29.300 --> 00:10:14.085
the amount of poisoned MCP servers and skill repositories out there is ridiculous. So everyone who's going out there and just bringing in all these skills to their environment, all they're doing is opening themselves up for exposure. If you've seen my video on skills and how to build them and the other two skill videos that I have, you'll know that that's one of the worst practices you can do again specifically because you should be only putting skills in your environment that you actually need and that's usually built off of the strategy, the foundation that we talk about. Don't just go out there and connect to repos and things like that because someone told you it's a cool skill. You might not even need it. But your role would be pretty simple here. You would come in and do a security assessment on whatever their environment is currently set up, find any loopholes or problems with compliance and vulnerabilities and things like that that can be exploited, and then you just patch them. It's really that straightforward.

00:10:14.405 --> 00:10:28.700
Thankfully, with the correct skill, in terms of their own personal environment, you can order that in less than five minutes. I've built a skill that's on one of the videos that I'll link down below. More importantly, you can add to the expertise by making sure that they understand what good looks like when it comes down to security,

00:10:28.700 --> 00:10:52.900
data privacy, and things like that. So there are many different consulting layers that you can also add on top of the basic technical work here because a lot of businesses really don't have that capability, and most people don't even think about this kind of stuff. Next up, pillar five, and this one kind of ties in with the first pillar, but it's specifically for knowledge and skill hygiene. So these two tie in with one another because the skills are built off of the business knowledge, and ultimately it lives next to the skills that it can go out there and do this repeatable workflow.

00:10:53.060 --> 00:11:07.465
But what a lot of people are doing is they're working off of knowledge that was accurate six months ago but is no longer accurate. So one of the things you could do here is obviously automate this entire process for them so that their skills and their knowledge gets automatically updated based off of the latest context.

00:11:07.705 --> 00:11:26.620
For instance, if somebody has a call, they get off a sales call, whatever it is that they were speaking about on that call updates the context or knowledge about that specific client. That's just one working example, but the idea behind a much larger system for this would be for every person who is a part of this business, you would find out where their global knowledge base is, and that would be automatically maintained

00:11:26.700 --> 00:11:40.155
and kept well over time. So for instance, if they're creating content and nobody went back after the first time somebody built the system to update a voice, to change any words that they didn't want in there or any styles of content, to update their brand and what it looks like, what it sounds like,

00:11:40.395 --> 00:12:46.390
all of those things need to regularly be addressed because AI evolves over time, so does the brand image. And a company changes faster than it ever has, which means systems and knowledge and all of those moving pieces need to change along with it. So our after screenshot of what this would look like, you would have a lot of structure behind where everything is stored, why it is stored there, what is stored in an actual knowledge base with versus what lives in the actual skills. You would also have a skills library because most people just have stuff thrown in their environment because, again, someone told them that they needed it. That's ridiculous. You need to get rid of all of that stuff that they're not using. You would do audits on this regularly, get reports. As you're the person managing this, you would understand exactly what they need and when they need it. And as a result of this, you're not just fixing their skill hygiene and their knowledge, you're also making them more secure because they no longer have things that could be exploited that don't need to be there or systems that don't need to be there. So these layers all tie in with one another. Then we have compliance and governance, and this is obviously quite niche because you can't just walk in there as somebody who's never done anything with either of those things and be like, yes. I will make you compliant. That's not gonna work at all. So this, I realize, is a little bit more niche, but if you are in the GRC space or in some form of compliance,

00:12:46.630 --> 00:12:54.070
you can make a ton of money doing this. This market is literally going to do this because as more AI models come out and things change over time,

00:12:54.390 --> 00:13:26.435
nobody really has any idea how to manage any of this stuff. If you can go in there and not only make it faster for them but also easier, you've pretty much sold yourself. This is also really important from a data privacy perspective because a lot of people using these systems right now, they don't understand the problems that they're creating for themselves with data privacy, GDPR, other things like that depending on whatever region in the world that they're in. If they are handling data or putting data through systems, they don't know what they don't know. So your role would be to come in there and expose this kind of information to them, show them what they are doing wrong and the trouble that it can cause them and how they can become compliant.

00:13:26.595 --> 00:14:39.410
And again, you can automate this all super easily. If they have several workflow skills, whatever it is that they've got in their environment, you can have a skill that runs through all of that, understands what each workflow does, and then shows you a report of where this could be breaking compliance or governance or security and privacy and things like that quite easily. In just a few minutes, you would have that report. So audits might be boring, but you can definitely make a lot of money from them. Now something that's really important is you don't wanna walk into a business and say something like, I'll migrate your stack for you. They don't give a shit that kind of thing. You need to frame every single one of these things that I just spoke about about pain prevention. Focus on the things they're afraid of or the things that they don't know about because in there is the pain that they're trying to avoid. People will always do anything to avoid pain. So instead of talking about optimizing their costs, you talk about how much money they're wasting in their systems because they haven't bothered looking at it. Instead of saying that you'll set up observability for them, you tell them that their AI is making bad calls and nobody knows about it. More importantly, they don't know how this could be affecting their clients or their sales or their team. Focus on the layers of pain within that because observability is boring and really no one cares about it. Same for security incidents. You don't want to focus on the security audit. You focus on the fact that even one incident can cost them millions of dollars and destroy their entire business and reputation.

00:14:39.810 --> 00:15:13.260
Things like that. Always focus on the pain prevention behind your marketing when you're trying to sell this sort of thing. In terms of what to charge for this, there are obviously different methods that I spoke about earlier in this video, so you could charge for the audit. You could also lead with a free audit for specific thing like a quick security review that would take you a few minutes to do that. One of the things we used to do in cloud is we used to do a cost optimization checker, and that was simple. You just go into an AWS panel and check all the services that they switched on and map that to whether they actually used any of them. Often, would turn out that they weren't using a bunch of them. One of the places we went to, there was, like, $45,000

00:15:13.260 --> 00:15:26.085
of services that were being charged every month and nobody was actually using it. So obviously, it depends on the size of the business. I'm working in enterprise most of my life, but for smaller clients, money is money. They are still wasting money in accordance

00:15:26.245 --> 00:15:41.580
with their level of the business. Point is find out what you can cater for, and then you price according to that. I've got a whole video on value based pricing, which I'll also link down in the description below. But once you've come up with your number for perhaps doing the initial migration or just this whole package upfront,

00:15:41.900 --> 00:16:56.665
you would then be able to track in this retainer model so that you have recurring revenue from these people because clearly if their system's been abandoned for the last year, they don't have anyone internally who's championing this, so you come in as the person who augments this for them. Obviously, prices here are pretty arbitrary. It depends on the size of the client, how much revenue they're bringing in, how valuable you are to them as in how many of those pillars you can stack behind you and your skill and expertise and things like that. So what I wanted to do was just give you a breakdown of how this might look based on tiers. It's a very popular approach at the moment to have these different tiers that people can choose what they want from you. Your goal would obviously be to get as close as you can to tier three because then even if you've got like five clients in a small enough business, you would have 50 k a month, and believe me, this stuff stacks because even something like this is way cheaper than having a full time AI employee right now who knows what he's talking about. So this whole fractional approach can really work for you. Again, comes down to the clients that you're working with and how experienced you are. You're probably going be starting off with tier one if you're brand new to this sort of thing. But don't ever sell yourself short because your value is in the fact that you know these six pillars exist as opposed to people who are just going out there and not doing anything with them in the first place. Then one final thing for this video, those of you who do want to make more money than this, obviously, your easiest way to do that because you're not gonna get in front of enterprise clients very easily as a solo consultant

00:16:56.825 --> 00:16:57.945
without partnerships.

00:16:58.080 --> 00:17:39.680
So if you're starting out, the best thing you can do right now is find partner consultancies within your area that need augmentation from you. They have access to larger clients that ultimately you want to get to, and believe me, partnerships have served me over the last twelve years. They're the reason that I made most of my money in enterprise because they opened those doors. But I also made friends, I built relationships, and those all opened up extra doors for me as well. So that's probably the best thing that you can do. You don't always need to be the lone wolf here. Oftentimes, it's the worst play because it takes you much longer to get to where you wanna go. So I hope this video was helpful to you. If you have any questions, leave them in the comments below. Otherwise, check out the videos on the screen now. They'll definitely help you in your journey. If you want any more guidance, I do have a community around this and building your own AI operating system for yourself or your clients. Thanks for watching. I'll see you guys in the next one.