AI Automations for Tradespeople to Explode Productivity using Zapier with Tersh Blissett

Speaker A: I haven't quite put my finger on why the hype about AI has made people so much more receptive to saying, oh, let AI help me fix or help me solve this.

Speaker B: I believe wholeheartedly it's fomo. Um, um. Because that's all you hear. If you do one search on ChatGPT, on TikTok, you're going to see the first thing that pops up is like, hey, don't be late to the game. Meet Tersh Blissett, co founder and chief experience officer at, uh, Trade Automation Pros. He is also the chief executive officer at Service Emperor. He helps home service professionals use AI and automation to eliminate repetitive tasks and free up time to grow their businesses.

Speaker A: Just a curious question for you. Are you dabbling yet in AEO or geo?

Speaker B: We've actually done a lot of research. These AI tools, the first thing they want to look for is the price. If they can recommend your price over just random prices that are on the Internet, then they're going to do that.

Speaker A: There's some new protocols that were created toward the middle of last year that will start to be very popular soon, enhancing you from just search engine optimization to AI engine optimization. There. One of the things that they've built is. All right, everybody, welcome back to the AI for Business Owners podcast, sponsored by Stingen AI. Uh, as always, I'm your host, Jeff Torello. Uh, if you're an SMB and you are looking for ways to take advantage of AI and not just buy an indiscriminate tool, head on over to Singin AI and get in touch. We'll see what we can do to help you out. I, um, guess today is Tersh. Go ahead and tell the audience who you are and what you got going on.

Speaker B: Hey, man, first off, I appreciate the invitation and you're welcome to talk to you. So my name is Tersh Blissett and I own two H vac companies, two plumbing companies, one electrical company, and have been building automations for the trades, the skilled trades, since like, um, 08. And I do a podcast service, Business Mastery, which I talk a lot about business in the podcast or in the trades world. And that led to starting another company that strictly is, uh, it builds automations for the trades.

Speaker A: Cool. Yeah. I'm excited to talk to you because I actually think it's a bit of a niche that probably doesn't get a lot of. A lot of traction. I know your podcast is trying to help. I get everything. I'm not, I don't mean it negatively, but, like, I don't think the mainstream press is covering it often and I'd love to, you know, get the audience to see and hear what, what you've got going on and especially if there's anyone out there that can relate, that would be super valuable. How did you get from having a service business to doing an AI? Like what, what was that journey like? Uh, obviously more than a 5 second thing, but uh, that not common. Right? So how did that get there?

Speaker B: Basically? So long story short, I uh, was in the Air Force and went through like processes and procedures and then you leave. I left the Air Force and actually went to school for engineering and so very rigid. But I didn't like the job. Like I didn't like where I was working. I was at a chemical plant. I didn't enjoy it. And I actually had a friend of mine who was doing air conditioning, was a service technician for an, a local H VAC company and I was like, I think I want to do what you're doing because I like tagged along with him a little bit. Uh, we were car buddies so we were always working on cars together and got chatted about it. I got into the trades, like ultimately I came in and I just really enjoyed it and it was a natural niche for me. And after being there for a couple years I was promoted to be a service manager. And one of the things that happened was they're like, here's the Excel spreadsheet. You need to put down all the technicians time for the yesterday and when they arrived on a job versus when they left the job. Like basically how efficient were they? But everything was a manual process with a manual calculator. And it was wild. And it took me about a month of doing this every day for the previous day before I was like, I spend from 8am when the guys leave the shop, finally till 10:30, 11 o' clock every single day, manually going to this website, pulling deep work reports. And then and I was like there has got to be another way of doing this. And so that's whenever I really found how to do Excel spreadsheet automations. And that's where it started. And it was like just a snowball effect of like now I really enjoy this. And ultimately I got to where I was searching for things to automate because I had automated so many things in my life. Little things like bullpay, like bill pay, spend five minutes to set up your bill pay and then let it automatically pay your electric bill every month. Like it's going to be a pretty consistent amount, you know, or your, your cell Phone bill or whatever. So that's kind of how it, it started. And then when I started my podcast in 2016, I was talking about business and naturally I started talking about automations because I, that's where I thrive. And so people started asking me questions like, hey, how did you automate that with this CRM? Because, like a lot of us use, uh, the same CRM or a handful of CRMs in the trade. And so I was talking about how to pull information from one place and put it in the other and, you know, talk back and forth between QuickBooks and stuff like that. And that turned into people like, hey, you need to create a course on how to do this. I was like, oh, that's another task to add to my job, you know. And, uh, so I eventually did. And basically I gathered information from podcasts and that's the, you know, initial courses started that way and it turned into a whole other business. Trade automation pros came from that. Like, it's trade automation pros, you know, like, that actually was created just prior to like the AI. AI to the public boom. Like there was a lot of AI out there before, but the actual general public, um, chat GPT came around. That's whenever things really took off and have taken off. And I'm doing like on sites on a monthly basis now where I'm just going to people's contracting, uh, businesses and saying, hey, can we make this more efficient? How do we do this? And do we use an AI tool? Do we do automations? Or is it purely like, you need to update your processes?

Speaker A: It's a similar vein to what we're doing because, and I've said this a few times on the podcast, right, that AI is not the solution to every problem. In the end, I don't necessarily think it matters much if the customer thinks it's AI and it ends up not. Like it's not. It's not like we're not selling them the thing that they need. You know, there's no chicanery involved, right? We're just solving the problem and providing the value. But there's quite a number of instances where it's not actually AI at all, it's just software. And part of what I say almost every podcast now is that realistically, AI is just another explanation or convoluted way of saying software, because in the end that's all we're doing, it's just software. But the idea, I haven't quite put my finger on why the hype about AI has made people so much more Receptive to saying, oh, let AI help me fix or help me solve this, or like, I'm willing to invest in you building an AI solution for I don't know what.

Speaker B: So honestly, for me it was a

Speaker A: lot harder to sell like plumbing company on a custom software. Like that was not happening five years ago. Few and far between.

Speaker B: Yeah, yeah.

Speaker A: Now suddenly they're all like AI and I can't figure out exactly why.

Speaker B: I believe wholeheartedly it's FOMO and like. Cause that's all you hear. If you do one search on chat, GBT on Tick Tock or you know, anywhere on your phone because your phone's monitoring anything you're doing, you're going to see the first thing that pops up is like, hey, don't be late to the game. You have to be here or else you're going to be left behind. And you know, to an extent that's true. But you know, it's the same way with email. Like when email first came about, like there were people that were very hesitant to get on board with it. And then I remember in 05, I think it was 07. Yeah, it was 07 when I became a service manager and we were just getting a website and back then it was a pretty major investment. I remember we spent like a hundred thousand dollars at that company doing a website. And now you can have any of your vibe tools out there that just spin up a website instantly. And granted it might not be the best converting site, but.

Speaker A: Yeah, but if your carrier, a hundred grand on a website makes some sense because you're doing more than. But if you're Dave's Plumbing and Heating and I'm not trying to make that little. I'm trying to make that sound small like it is. You can go to WordPress and have a web like you don't need a major investment.

Speaker B: And so like I built.

Speaker A: But that was in 2007. You're totally. In 2007.

Speaker B: Yeah.

Speaker A: There were no tools. You're out of luck.

Speaker B: Yeah, oh yeah. There was no website builders and stuff like that. But in 2014 is when I started my first H Vac company and I built the website for that company and I have not updated it since 2017 or 18, but it still ranks number one on the page. Now there's even been updates to some of the pages and it's like it says, ah, Lipsum orum or whatever or.

Speaker A: Yeah, creek text.

Speaker B: Yeah. Where it's done the update and then it has that in the metadata on the main screen. So it's like, how is this thing ranking? And it's because I just watched YouTube and was like, how do I build this the most efficiently as possible? And I spent a lot of time doing it for just my website. I learned a ton of. And it still ranks.

Speaker A: And one of the reasons that it ranks just to help you out from the understanding perspective, is the amount of time it's been in the index. Right. So even if you haven't updated the content in a long time, it's still around, which means the business is relevant. The business has a, uh, ten plus year. All of that adds to why you show it, et cetera. The other side of that equation, just to think it through. Right. Unlike a dynamic interactive website that you need to change daily, like a news site or uh, maybe a shopping site, service provider site really needs to get across some very basic information and then, you know, here's how to contact who I am, whatever. And so once that exists, there are not a lot of customers who want to sit on your website for half an hour reading like, it's just not the way the world works. Right. No one has the time. I see the name, the name seems like I've heard it before. Friends recommended it, they're in the area, uh, okay, let me call them. And that's about as far as it goes. Maybe send an email. Right. That's kind of what's going on. They're well reviewed, they're well regarded, those sorts of things.

Speaker B: Yeah, absolutely. Well, the other aspect of that is we make blog posts bi weekly or monthly. I try to stay consistent because Google, it does, but ultimately like sure, it's written to the audience, to the client, but at the same time it's, it's being done ultimately so that Google will keep you at the top of the, you know, the chart. Uh, yeah, for ranking purposes. But just by happenstance, if, if a customer reaches out and does a Google search and then wants to read the article, it still needs to make sense. So.

Speaker A: Indeed. Yeah. It can't be all lorem ipsum or, or it'll, it'll not make sense.

Speaker B: Exactly.

Speaker A: Uh, just a curious question for you. Obviously there's an SEO component towards a search engine optimization component there. Are you dabbling yet in AEO or geo?

Speaker B: Yeah.

Speaker A: So just for the audience, right. Genitive or agentic? Basically, how do you get ChatGPT to respond with here's a plumber in your area? That's what I'm talking about.

Speaker B: Yeah. So we, we've actually done a lot of research. We actually put on an event, uh, October of last year in Savannah. Uh, by the way, I don't think I mentioned that. I actually I'm in Savannah from Savannah, Georgia and that's where home base is. But we put on an event called Tap into the Future in October of last year and part of that event we had a guy named Marcus Sheridan on the come present. He was our keynote and his whole premise is put your pricing on your website. He had a pool. Ah, so it used to be put your prices on your website, create a blog post like a best of blog post but don't put your name like don't put you on the list. And and that worked amazingly for SEO rankings. But whenever it came to AI rankings and search, uh, what he found was that his company was no longer being listed in the list but they're taking his list off of his website as the recommendation. So he was like oh, I gotta change this up. So then they put in there like their company was the best but then here the other, you know, non companies do a great job as well. And so then they found that these AI tools, the first thing they want to look for is the price. Because if you can put your, if they can recommend your price over just random prices that are on the Internet then they're going to do that. Uh, especially if they're, if it's somebody in your local market area. And then the other thing is putting those best of list together and the FAQ sections and for a while there we were being told to put FAQs all over Google business profile but then I guess they did away with that. I don't know, they're bringing it back or whatever.

Speaker A: Google changes the game if you will.

Speaker B: Often they do so much because, because

Speaker A: there's so many companies out there that are basically gaming it and offering it as a service. I know you know this because you've probably been pummeled by it. Another tidbit I can give you that'll probably help make sense with three respect to enhancing you from just search engine optimization to you know uh, AI or agentic or generative engine optimization. There's a bunch of different names, they're all talking about the same thing. There's some new protocols that were created sort of the middle of last year that will start to be very popular soon. They're all about agentic shopping.

Speaker B: Yeah.

Speaker A: And one of the things that they've built is a secure transaction model so that the agent can actually make purchases on behalf of the user or you when you're chatting to whether It's Claude or ChatGPT or Gemini, it doesn't matter. Right.

Speaker B: You're not giving them access to your entire wallet.

Speaker A: You're not. But you're doing this, you're saying, here's the ability for you to make a payment. And, and it comes with this vetted back and forth structure where it's almost like the padlock on a website when you go to Amazon or whatever. Right. Same structure behind it for the actual transaction. And what happens behind the scenes is there's this protocol that the agent, so, uh, when you're having the conversation, you can say inside of, you know, your chat interface, show me the top three plumbers in Georgia. I don't know, I don't know how that's a rerele and no one would search that way, but you know what I mean? Yeah, we'll pick Atlanta, whatever, it's fine, show me the top three. Now, I don't know what that would really do anybody, because what does that mean, top three? But it's going to answer you no matter what, right? And it's going to use your website information to answer. And if your website includes prices, that's going to be included in the answer. And it may not necessarily say it will cost, especially if your website's intelligent. It says things like, you know, consultation starting at, you know, $99 or whatever. I don't, you know, I don't care about the details. But, but the idea is you've got to put something out there because your competitors will. Right. And from a service business perspective, there's still a bunch of discussion going on and the protocols are still evolving about how this would work because it'll probably be, the agent will reach out to the contact on your website or an email address you provided or whatever that looks like. But if you were selling product, the actual agent will attempt to purchase the product via the protocol so that the end user, and this is where things start to get a little interesting from my perspective. So the end user can say, you know, show me the top three hubcaps for my car. You know, I don't know, whatever. And they say, okay, great, buy the first one and then you move on to the next thing you're trying to do. And you don't, you don't say the agent knows your address. The agent has your ability and authorization to make a, a purchase via, uh, secure payment gateway and it's done.

Speaker B: You said go how that's going to go into effect, like Amazon, because Amazon's thing, it used to be pricing and then it was like next day shipping but now they've moved our shipping out to three and four days a week sometimes. And then now it's purely convenience. Like for me it's not pricing anymore because more often than not I'll find that the prices are more on Amazon than they are at a local shop, um, or at least another online retailer. But for me it's convenience right now and the ease of looking up the part again in the future. So like if I need to reorder it or something or warranty issue but that's going to go away with, with that sort of agent.

Speaker A: Right. So uh, you know, a couple, couple comments, right? First one is if you can do that, why are you going to the Amazon website as often? Because you're just going to be in your chat doing your thing anyway A, B, Amazon refused The offer from OpenAI to be part of the commerce model for ChatGPT and Walmart said we'll do it. So Amazon is currently seemingly expecting to keep everybody inside the Amazon garden and uh, not allow Amazon to be listed on a uh, competitor's chat site from a product perspective, which sounds weird to me.

Speaker B: No, that makes sense though because I've had conversations. I was rebuilding one of my PCs. So like I built a super computer about 10 years ago. We all that even a supercomputer 10 years ago is basically what you can buy off the shelf at this, you know, today. But I knew that I wanted a new PC for using Open Claw because I didn't want I do it online on my PC. Yeah.

Speaker A: But also careful because boy oh boy, is that a nightmare.

Speaker B: It is.

Speaker A: We can talk about it, it's fine.

Speaker B: But no, I've, I've been learning that um, yeah, so basically I put all of my stuff inside of a project inside of chat GPT and so I already have a project for like my office because I have so many different PCs in my office routers. I have a NAS system. So I have just a full project with each part and everything else in it. So I started another thread inside the project and it was like, all right, so order, uh, you want to change your GPU over to this gpu because the one I had was for graphics. It was a massive graphics GPU for video games and stuff, uh, and 4K video editing. But I was like, okay, cool, give me an Amazon link. It would not give me an Amazon link. It gave me every link but Amazon. And I was like, you are bound to determine for me not to purchase this on Amazon, which was Actually, fine, because I found the GPU for $75 cheaper from this other website. The problem with the other website is just the trust, the lack of trust. And like with Amazon, if the GPU didn't show up in five days, I just message Amazon and say, hey, send me another one, because it never showed up. But even that's not accurate anymore. It's not like the way it used to be.

Speaker A: No, it's not. One of the things that I know OpenAI is doing in this particular case, I'm sure the others will do the same. Thing I only know about the OpenAI one is they're vetting the individual vendors for the. For products that ChatGPT will show.

Speaker B: Oh, well, that makes. That makes me feel better.

Speaker A: Right. So what would happen in this context is you would ask, you know, chatgpt, hey, um, I need a. I need a graphics card that I can use to run local AI models. Just. If we just make that simple conversation, it's print what you asked, and if it comes back and says, okay, here's three choices and three prices, and you pick. There's a component here where you don't even need to know the vendor.

Speaker B: Yeah.

Speaker A: Like, you don't actually necessarily need to care most of the time. And, uh, you know, like, if weird, but I mean, like, let's say you're buying a brand name and it's. And it's already a brand you know you want.

Speaker B: Yeah.

Speaker A: You know, Dawn, Dis. I don't know. I don't know. Whatever. Right. It doesn't matter where. It doesn't matter if it was CVS or Walmart or Best Buy. Who cares if it's the price you wanted to pay? And it'll drive when you want, buy and off you go. And if it came from Amazon or it came some other way, you really don't care in the end.

Speaker B: Exactly.

Speaker A: I mean, if you're an Amazon stockholder, maybe you care. But the point is, I think that structurally the shopping model is going to change wildly as this happens, because it looks like AI and CHAT is here to stay. And so, you know, we're early days now. Right.

Speaker B: It's wild to think that, because literally two years ago it was like, just night and day different the way it is right now.

Speaker A: Uh, completely. Uh, the analogy I like to make for folks is in 2007, Apple comes out with the iPhone.

Speaker B: Dang, I can't believe it was 2007.

Speaker A: 2007. The biggest thing about the iPhone at that point in time was not that it was a cell Phone was not that, it was email. Blackberries had been out for a while and were very popular.

Speaker B: Yep, I had one, but there were

Speaker A: others like Palm and there were other ways to do it. And it was not that you could text, it was the fact that it actually ended up being a platform. But if you look at what we're. And this accelerated like mad for us now. But if you look back in 2007 era, you're looking at three technologies that are, were ancient, that were combined in order to make a platform to allow uh. So cell phones had come out in the 70s. If you go really back to Motorola. Right.

Speaker B: Old bag phones.

Speaker A: Exactly right. Those, those existed um, in limited form and a foolish expense. Right. But that technology was not. 2007 was a long ways away from its inception. Over a couple decades. Right. Same thing with the idea behind, you know, email. That was an 80s kind of thing. DARPA, net first, then ARPANET, then Internet. But by the time you get to 2007, email is how business is done at that point. Well into the commonality. Now I want to do it mobile. Okay, I get you right now. WI fi shoots out around 2003, give or take. So by 2007 your phone can do WI fi and that's great. But when you're driving down the road, having Internet over cellular was uh, insanity. You know, I'm just bringing back all these crazy things, but I remember that is it's the convergen of all these. So then Apple creates this platform. Android also does the same thing. Google buys Android, blah blah, blah. Fine. My point is simply this. The technological innovation is the LLM and the AI and the chat. They're trying to make them all be platforms.

Speaker B: Do you think that's going to happen?

Speaker A: What a prognostication question to ask, right? I mean, yeah, of course you want that answer. I think it's this. I think I'll keep the analogy going. Nokia had a platform. Oh yeah, right. There were all kinds of apps you could do on a Nokia phone. AT&T, BlackBerry, they all had their own gardens, their own. But it took the app Store where everybody could participate and all these different apps could show up. What made the iPhone amazing was the apps, not it's a cell phone. Because in the very beginning, iPhones kind of sucked at a lot of things. Right. Even the first iPhone didn't even have copy paste. Like you couldn't do that. But you know, nobody needed it at the time or if you needed it, tough, you couldn't until they grew. Right. But the point is this uh, Gemini is going to try and make a platform. Microsoft's making Copilot be a platform everywhere they possibly can and shoving it down people's throats. Even if you don't want it. OpenAI is going to do the same thing. So we'll anthropic with Claude and so on and so on.

Speaker B: Perplexity browser thing that sucked. Uh, by the way, I don't know if it's any better now but like last year when I tried it I was like uh, oh no thanks.

Speaker A: But again this is all everybody trying to make a play for. I want to be more than the thing. So Perplexity being, you know, semantic search, which is really neat and I actually like Perplexity in a lot of ways. Claude is super powerful and does some really, really awesome things. I'm actually a big fan of Claude Gemini 2. I actually think if people played with different models and this is one of the things that I tend to do, especially with clients. I like to take clients and have them look at platform providers that allow you to change models rather than ah, like a magi. That's one. Right? There's one that I use often called Sim Theory. AI happens to be a name like Simulation Theory. Um, and there's plenty of other providers that aren't providing their own LLM. They're providing like a gateway with a chat. And do we allow you to have

Speaker B: a local also or.

Speaker A: So the problem with local AI is most people don't have the correct hardware. Let's, let's get into that for a sec. It's a good question. Right?

Speaker B: That's exactly why I took my PC to do that, because I wanted local LLM.

Speaker A: Um, a lot of folks would first say why do I want to run a local LLM ip? So the very big driver for most folks is privacy security. But if you say openclaw, you've destroyed both privacy and security because it's neither of those.

Speaker B: You've gotten completely rid of them.

Speaker A: But there are a couple interesting applications. Some are uh, open source, some are subscription that let you run models locally as well as use cloud models. And this I think is more in a long term approach. Here's what I mean. There's a thing I think for Macs called, I think it's called Osaurus. It's like O S A U R U S. There's another one, um, there's an open source one that runs on all platforms whose name I can't remember at the moment. So if I remember the beginning of the podcast, I'LL tell you again. But you install the thing locally and you can put in credentials to use your ChatGPT, your Claude, your whatever, you know, if you've got an account there and then it talks via the API and you have your chats, you can also download and install local models like an Ollama instance would let you do. So Llama four, you know, Mistral, any of those open source guys and there's a whole bunch of them. I'm not naming them all on purpose.

Speaker B: Well, that's the two popular ones. That's the two that I have.

Speaker A: I'm sure it is. Yeah, yeah, yeah, it's common but. But the way it works is you have the chats that you don't think you need a lot of privacy and security force. You're just asking generic questions or you're asking for research or whatever. And then you have the chats where you are want to talk about local documents and do manipulation and you don't want to share that with any of the vendors because all these vendors are turning into the same thing as social platforms where they just take everything they can get from you to try and sell it for ads. Right. Okay, so what you do then is you just have these chats are done with the local model, these chats are done with the cloud models. And the vast majority of the time the cloud models are way, way, way, way faster.

Speaker B: If you're running an agent to do this stuff, it doesn't necessarily matter to speed as much. It doesn't like, especially if you're saving money from it.

Speaker A: Because that was uh, the cost, the execution cost, especially from a cloud perspective. Totally agree. The problem is, I mean you've done it because you went and built a machine to specifically run AI stuff, Right? But most people run the laptop. They have.

Speaker B: Yeah, no, that you'd be.

Speaker A: Which isn't going to work. Right. There's a little. The Macs have a little bit of an advantage. A uh, small one here.

Speaker B: Yeah, that's true.

Speaker A: Because of the way the shared memory model works on a Mac. I don't want to get into a hardware conversation too deep, but if you buy a Mac with a decent amount of memory, you can run models that are. Because the models aren't small and you need a lot of memory to run models. I don't want to waste too many people's time here, but you know, those Macs are not inexpensive. You're going to spend a couple thousand dollars by the time you're done. And if you're trying to run Like a background agent or whatever. You probably don't want it on your Mac that you might be closing and m. Like, that's not the right answer. The answer is have a desktop or whatever. Exactly. Right. By the way, for the future, if you're going to build another one.

Speaker B: PCpartpicker.com oh, I need to write that down.

Speaker A: It will let you find the best price for the different parts anywhere across all vendors on the Internet. It will also not let you pick incompatible parts. So if this CPU and this motherboard don't go together, it won't let you do it.

Speaker B: Really? A cooler for my, uh.

Speaker A: Right, exactly. PC Part Picker will do all that for you.

Speaker B: I ordered the one for my new tower. Not thinking about the fact the motherboards are completely different.

Speaker A: Yeah, it didn't work. Yeah, yeah. PC part picker will help. Level 5 nerd. I don't know what the right answer is here, but like I've built, you know, hundreds of computers. PC Part Picker is your friend. Also, there's a bunch of existing builds you can just look at and go, oh look, someone ditter of this. And here's all the compatible pieces. Click. Let me just have that as a shopping list.

Speaker B: Yeah, so.

Speaker A: And it'll, it'll find what rebates going on at this vendor versus that vendor. It'll end up showing you buy it here, buy it there, buy it there.

Speaker B: That's what's uh, up right there.

Speaker A: It's a really cool tool. I've sent them money many times. I've used them for years and they helped a lot. Diverts like mad there. But I think a lot of folks don't like. I think running a local model is probably off the beaten path for most users.

Speaker B: Yeah, I agree with that. I think that. I 100% agree with that. The other thing is, is making sure it doesn't turn off because like I'm in Seattle trying like, and then we have a power surge here or something like that, like so I had to invest in battery backups and everything else.

Speaker A: And that's why the cloud. Yeah, there's a trade off. Right. Because it doesn't matter if you're in Seattle. The same access to the cloud that you get from home, blah, blah, blah. You could still run your local model in the cloud. You would, um, there's.

Speaker B: You can rent some AWS space.

Speaker A: Yeah, I wouldn't do that. The hyperscalers are foolishly expensive. I'll give you a little secret. There's a company called RunPod IO and they basically let you rent cloud based GPUs.

Speaker B: Oh really?

Speaker A: Yeah. So think of it this way. Tersh, you write your agent to run on your local whatever but you make a call to a llama running a run pod on a 48 gig GPU and you only pay for the API use to run that. And then it spins down when your agent comes back and just does local calculation. And like I know you've outfitted yourself. I, I'm not trying to, I'm not trying to change your whole world here. I'm just telling you there's other ways.

Speaker B: I'm always down for better ways of doing things and like, uh, for me the like. So I wanted some security aspect of it. That's why I ran it versus on my own PC. I also wanted to make sure that man, I have one um, hundred twenty eight terabyte NAS system that is halfway full right now. And majority of that, I would say at least half of that is duplicates and triplicates game stuff. I'm like, I'm not going to do this. I'm not going to pay a VA to do this. So if I can build a bot that can somehow manage that. And not only that, I have podcast episodes from four years ago that might be 60 gigs each between all of the video recording and audio stuff and then duplicate files inside of there and then editor files and all this other crap that can be compressed and like I've never gone through and compressed any of, of the documents and stuff like that so.

Speaker A: Have you ever heard of Dupe Guru?

Speaker B: Ah, no,

Speaker A: I'm just throwing all kinds of stuff today.

Speaker B: I love this.

Speaker A: Just search for, search for Dupe Guru. It's open source software, runs on every platform and it will help you find duplicates and it will work across your NAS drive from your machine and Google Drive. That's interesting.

Speaker B: Um, drive on my Windows. And the other thing is, is I'm not a huge Linux user but the server that I built out, I wanted it to be on of course.

Speaker A: Yeah, I'm sure, yeah. To do all the right things. I think if you had Google Drive installed locally.

Speaker B: Yeah.

Speaker A: As a, as a local resource in like Windows, let's say, might be possible for dupe. I've never done it so I can't say. But it might be possible for you to look at it and search. Yeah, yeah, it might be possible. Um, but that tool will definitely help you figure that out and uh, it'll automate as opposed to you having to do anything manual. Um, it'll save you Some amount of time. I don't know how much, but it doesn't matter.

Speaker B: Even the duplicate ones, they're like, you need to go in and manually review every one of these duplicate files. And I'm like, oh, I don't have time for that.

Speaker A: Not at all.

Speaker B: This is just as bad as if I would have done it manually.

Speaker A: Well, the other thing that you could do, right from an IT perspective, there's a mantra that's not easy for people to grasp, but it kind of works like this. Three is two, two is one, one is none. And it's the backup mantra. And it basically means, I agree with that. If you don't have three copies, you don't have a copy.

Speaker B: Yeah.

Speaker A: And then if you get into, you know, high grade protection, what you're looking at is you have two copies and the third copy must be somewhere physically different.

Speaker B: Like on an external hard drive or something like that.

Speaker A: No, like in another building, in another. Somewhere physically completely different. Because if you've got two copies but they're both in the same house and there's a fire, you have no copies.

Speaker B: Well, the other thing is, is majority of what I have are past episodes or events or keynotes. So they're not that important. It's not like we have securities of clients information or anything like that on our NAS system. So more is more of a digital asset management tool. And that's very gotten out of control. The more people we've hired on for the podcast production company, we're like, at first we didn't have a systematized file naming system.

Speaker A: Yeah.

Speaker B: Uh, everybody just did it their own way. And where I was like, no, if I have to click more than three times to find anything, like I don't want to go six levels deep in one file folder to get one image create and then find, uh, the images after the event's over because they all got put in that event folder instead of like a, uh, an actual m. You're chuckling. You know my pain point.

Speaker A: Oh, absolutely. I've gotten to the point where I've had to create specifically with, you know, large enterprise, but I've had to create server and client naming conventions in order to understand what's out there and be able to look and go, oh, I understand what that is.

Speaker B: Yeah.

Speaker A: Which doesn't seem like something you would ever need to do. But again, at an enterprise level, you're talking about hundreds of systems easily.

Speaker B: Not even at enterprise level, like I go into a building and um, let's call it a 15 or 18 million dollars a year revenue home service company. And they may have 15, 20 people in the office. And if you have 20 people have with 20 different file naming ideas or

Speaker A: whatever, then you never find anything.

Speaker B: You never find.

Speaker A: You don't know what's what.

Speaker B: Yeah, made it. It's like, oh yeah, it's right here. Duh. Like this is where we go every day. I'm like, no, not duh, because you stay in it every day. It doesn't make sense to me.

Speaker A: Exactly. It's funny, the more hands in the pot, the more process you need, right? Yeah, hilarious.

Speaker B: I guess that's the profound prior to processes. But that's one of when I do an on site, that's one like optimize, automate, delegate. That's it. Like that's what I, I tell everybody. Like before we automate a process, we got to make sure this process is optimized. And if we can't automate this process and it's something that doesn't add value to your day, like let's delegate it and. But it can't, we can't delegate it until we've optimized it. Because if we delegate a, a poorly optimize process, then you're going to get a lot of crap back.

Speaker A: There's an engineering adage that I would throw at you just to help you with this along the same veins, but it's beware of early optimization. Now the problem is I think optimize is being used in a couple different ways.

Speaker B: Yes.

Speaker A: But for example, you don't want to automate a process that you can delete that too.

Speaker B: Do we need the process that's like, uh, is it Elon Musk that's got

Speaker A: the five steps that's part of the problem? And what Elon's dealing with most of the time when he's talking about this, it's, it's physical hardware. Right. Sometimes he's talking about like the Tesla building process where you know, there's a part here we have to find a seven step thing to put in place. Do we need the part? No one asks that question. Right. That's what he means by early optimization. The first part, don't optimize it yet. Figure out if we need it, uh,

Speaker B: just because we've done it for 20 years.

Speaker A: Exactly. Yeah. And there's like, I can tell you firsthand, there are any company struggles with this, especially the longer they're around. Right? That's the way we've always done it. Oh man, that will make a deadly phrase

Speaker B: like that's. When I got here. Or that's how we've always done it.

Speaker A: Well, I don't know. Bob's the only one that knows anything about that, and Bob left three years ago or whatever. Right. That sort of stuff. And it takes someone who is willing to ask the questions that no one asks. Because for a. People are very resistant to change for a lot of good reasons. There's a lot of psychological reasons. I don't. You know, I'm not. I'm not suggesting changes. The only way, uh, the only thing

Speaker B: that's consistent with me is change.

Speaker A: Yes. And that's the way the world works. But a lot of people resist. Um, the second piece of that, from my perspective is not a lot of people question. Why enough.

Speaker B: You're right. I think just ask it five times

Speaker A: and things like, uh, okay, you know, I work with a client. What are your problems? Problems? Okay. Why is this a problem?

Speaker B: Yeah.

Speaker A: Well, here's what you're not telling me.

Speaker B: Why really a problem.

Speaker A: You're telling me what the problem is. Yeah, but tell me why. Well, uh, it takes so long, and in it, okay, there are plenty of ways to do that. Why aren't you doing it a different way? Uh, that's just the way we do it. This is the question you haven't asked yourself. Why don't you do it a different way? Et cetera, et cetera. And, you know, maybe you don't have the skill. Maybe there's a technical reason, like, it's okay. That's why we exist, to try and help out. It's fine. But I don't see people asking that question as much. And a lot of it, a lot of the time, especially when we're talking about very deep technical stuff. Right. They don't even know that they could. Like cloud GPU was a new thing for you. Didn't know was an option you hadn't exported. Okay, so you didn't know to ask. How could you? I mean, you can't know everything. Neither can I.

Speaker B: Know what you don't know until you know it.

Speaker A: Right. Um, and that's why sometimes it's worth the value of getting a hold of that expert and getting, you know, it's a little bit of money, but when people know what they're talking about in the service business, you know the exact same thing. How many times have you had to go fix a homeowner, I'll do it myself project that turns out, no, I needed a pro from the beginning. And, you know, at a certain point in time, you just Learn. I'm not good at that. I'm just calling right.

Speaker B: You hope that somebody picks up on that. Sometimes they just are, uh, a glutton for punishment.

Speaker A: I mean, I applaud folks who want to try and fix their own stuff and learn certain things. And I think that a lot of things are homeowner capable to a certain level. And then, uh, no, not so much. We'll end this on a really funny note. I have a toilet that's been around for a long time and I've replaced the innards of the toilet many times. The seal, it's a two piece toilet. The seal between the tank and the bowl is gone. It's just right and it's coming apart. And I have looked for three weeks to try and find a replacement seal. But this thing, this unit is so old.

Speaker B: Oh really?

Speaker A: I can't seem to find it. And I have ordered replacement parts like a new tank just to replace the tank, etc. Uh, they don't line up with the bolt pattern on this. Like I can't do it. I cannot. It, it's too old. So I was like, I just have to replace it. Like I don't have a choice.

Speaker B: I've looked or anything like that. Do you know? Like it's.

Speaker A: I did a ton of research. I used some AI chatting, I did a bunch of independent research. I even ordered one off of Amazon that was like, you know, fits most, blah, blah, blah. Just to try.

Speaker B: Yeah, what brand is it?

Speaker A: Oh, I don't remember.

Speaker B: Okay, yeah, I can't remember what it is, like a photo or something like that.

Speaker A: But the way I've resolved this problem, which will resolve in about four days, is I ordered a completely new ah, unit from Lowe's and the Plumbers come in Monday. Because I'm just not bothering with the. I'm not replacing the toilet on my own because I have other things to do.

Speaker B: Yeah, at some points you just like, all right, is it worth the ROI of me doing this?

Speaker A: This is the point, right? And so like I was willing to change the tank. I put a gasket in. Like I got all that. We got to change the whole toilet. Goodbye.

Speaker B: I had enough, like me going into a business and same with you, like, is it worth these business owners and, and also at what point in the stages of business is it worth just hiring someone else to do it? But then if you hire someone else to do it. So like, I have this uh, friend of mine who hired a quote unquote Zach Zapier expert. They were going to be their automation guru internally. And that person, while they were pretty good with Zapier, they didn't know the CRM at all. So my friend paid to have them trained up and trained really well.

Speaker A: Yeah, uh, now there is Zapier and CRM expert.

Speaker B: Exactly. And then they marketed themselves, started their own business. Of course they did other companies. This guy paid for that to happen. And so like, now he's gun shy. He's like, I will never hire an internal person again. I'll only hire people like you to come in and do these things for me. And I'm like, yeah.

Speaker A: I mean, the pendulum swung the other way and at some point he might get burned with a vendor and he'll maybe come back to the middle. It's the way the world works. Right? But.

Speaker B: But he doesn't want to learn it himself. Like, he's like, I, uh, just. It's a really cool concept. I'm just really good at financials and I don't want to learn automation. I'm like, okay, fair enough.

Speaker A: I honestly think vast majority of business owners, I don't think, I mean, depends, you know, if you're a solopreneur, maybe this is the right answer. But if you're running a business with decent headcount and decent revenue, you have other things to focus on. Learning how to be an AI engineer is probably not the right place to go. And maybe long term hiring your own internal AI nerd makes sense. But I think you start with using consultants, finding good strategic wins that are relatively small in scope so you can get a feel for the value in the ROI and decide to like. I just don't think, I don't think you should buy a big six figure AI renovation of your whole company. I think that's foolish. And I don't think you should hire someone immediately and be like, you're our AI dude forever. Because you don't know if you need AI that much, you know, like, right. So it's a fun.

Speaker B: Like we were talking about that earlier in the episode. Like it. I believe that we have two opposite ends of the pendulum. We have those who are like, hey, we're just gonna wait until this stuff gets figured out and then we're just gonna figure it out at that point. Because I don't want to learn seven different platforms because one's gotten better than the other. And then you have the other end of the pendulum where it's like, hey, we've already hired three AI specialists and they're going to learn while they work for me. And I'm like, okay, all right. Like, I get it. But when people, when I talk to people and they're like, hey, I'm, I'm checking out this Claude for the first time. Um, have you heard of Claude before? And I'm like, yeah, I've heard of Claude. Like, I have it on my phone and Gemini and Grock and Chat GPT and they're like, you, you have all of those on your phone? I was like, yeah, they all do different things. Like, if I want an image created really quickly, I can just go to Nano Banana inside of Gemini on my phone and just pop it up real fast. And like, if I want analysis done on an Excel spreadsheet or a document, I'm throwing that in the cloud. And like, if I just, like, Chat GPT knows me personally. And that's the other problem with not having a local LLM is that, like, I'm conversing with Chat GPT all day, every day, or might be quad, and they're learning my personality or as best they can, learn the person out.

Speaker A: Yeah.

Speaker B: Now I'm having to go back and redo the entire thing with my local LLM and like, basically Chat GPT either owns or has access to my intellectual property. And I don't know if you've seen this or not, but I haven't personally seen it, but I've read and watched that. Chat GPT has now, now has a health tab where you upload all your health documents.

Speaker A: Yeah, it does.

Speaker B: And it does, like, analysis for you,

Speaker A: which is great, which is convenient, but

Speaker B: also, yeah, like hipaa, like, how are we. You're giving it information, technically, you're letting it.

Speaker A: I'm a security nerd, so this, uh, I'll never use any of these things this way. I'll, I'll run local stuff or I won't use them for that purpose. You know, I understand why it's convenient and I understand why people are like, why do I care? And I, I, I get it. Some people don't. And it's a choice. I don't, I don't begrudge the choice either way, but I do think it is a slope. And I'm not, not a happy one. Yeah. Ah, not one I want to jump

Speaker B: on when I do presentations. I tell people all the time, like, hey, we really need regulations in this space. But I just wanted to wait until after I'm finished building my AI tools.

Speaker A: Just think of it this way, though. Whatever you build in six months will be out of date.

Speaker B: It is. And I mean, I started building this out 18 months ago. Over 18 months ago now. Well, actually 20 months ago. And the first 18 months, like slow moving, slow. And then all of a sudden, December, January, February, Hockey stick like crazy how fast things have gone since then. So. Yeah.

Speaker A: And what I was saying, be good. Sure. We're still early days. This technology is still nascent. It's just going to continue to get better. Today is the worst day. AI Ah is going to perform for you. You can say that every day.

Speaker B: That's crazy.

Speaker A: Maybe that's where we end it for now.

Speaker B: Yeah.

Speaker A: I mean, it's been great talking to you. Thanks very much.

Speaker B: Absolutely.