Running a Million Emails a Month for $100 with Amazon SES

Jordan:0:00
Welcome back to the height tickets podcast. Everyone please welcome the winner now me Mr. Joseph Franklin McElroy. Joseph, how you doing today my friend?

Joseph McElroy:0:01
Thanks.

Joseph McElroy:0:09
Very well, have a good day. Glad to be here.

Jordan:0:11
Yes, sir. Yes, sir. We are honored to have you on today, my friend. Let's see. For those of us that do not know the brand or the name just quite yet, can you please explain to us kind of what you do and who you serve the high level?

Joseph McElroy:0:25
Well, my company is Gagliatech Media, and from there, Qiskade is a different sort of opportunities that I'm pursuing. But the main one I'm pursuing right now is a sovereign operational systems, which is an open source AI based operational system for marketing and lead gen and then also into various automations to improve the efficiencies of a business. So it's a combination sort of a operational and marketing system that you know, that you know, because our background, Galileo's background, was SEO, right? My background is technical and SEO. And I got into a lot of marketing for real estate and for hospitality. So I have particular expertise in that space. And so Both of those cross over where operational becomes marketing and vice versa, right? They really have a very close relationship. So I'm doing this system to make that even more explicit, especially for small and medium sized businesses.

Jordan:1:43
This sounds really interesting. For those of us that don't know what open source means, can you please explain to us what open source is?

Joseph McElroy:1:49
Well, back in the day, there used to be a thing called free software, that was like free as in, know, something you say, not free as in beer. But the evolution of it was it developed a lot of real licenses, commercial licenses, allowed for it to be both free and to be sold, you know, in certain circumstances, in certain ways. So was a hybrid approach to free software. And so I was earlier a proponent of open source and it's maintained. A lot of the systems that we use today on the internet are open source. the major hosting platform is Linux and that's an open source platform. And it was the original platform for running servers on the internet. And then WordPress is something everybody uses, that's an open source platform. So there's...

Usually at the entry level and somewhat higher, they cost nothing. And then people develop both the smart services and additional services on top of this software to make money. And some of them have gotten huge investment into them now. One of the major automation platforms that people use right now is N8N, and I think they got $180 million investment. So open source is the real deal. Now I got my first you know, independent getting known for my technical abilities is I built the first open source platform for doing business online, totally open source. Nobody thought you could do that. They were using commercial platforms on top of some open source infrastructure, but I did it completely with open source. I got written up and everything else. And so I'm sort of a guy that's been in this open source world. I'm very much... have focused on small businesses, even when I've done big things, like I was the biggest, digital service provider to Marriott, but we were really doing services for about 7,000 small businesses. Because each hotel and their chain is actually sort of a franchise or an independent property, each has a property manager. So I know how to scale small businesses to accomplish some big things. And I'm interested now that AI has come into the picture to help small businesses

Joseph McElroy:4:14
use AI to their benefit, to increase their ability to operate and their ability to get outreach. I'm concerned that SaaS as a model can be corrupted into being sort of a rent-taking model. And since I was at the beginning of SaaS, I don't like seeing that. And so I view this as an open source approach to doing that.

And also that there's now lot of local models for AI. In fact, there was a new release for the old llama by Google called Gemma 4 that takes its abilities as a local LLM that can be on your local server to work on a consumer grade hardware and do high performance AI work, reasoning. So.

So I envision a world where all small businesses have their own LLM, know, protecting their own proprietary data, their marketing techniques, their lead gen techniques, everything, know, keep what's proprietary to them in their own ecosystem. And then they can reach out into the world for additional smarts, additional integrations, things like that. That's the vision I see both on the marketing and operational side.

Jordan:5:32
Gotcha. Thank you for the context on open source. Let's see. How would you integrate utilizing open source for utilizing LLMs and open source to help businesses grow via Legion and even from an operational standpoint? I guess how would that look?

Joseph McElroy:5:50
Well, know, there's actually a million ways to do it. I'll tell you some of the ones I'm doing right now. So, know, part of the work of doing SEO and AEO, search engine optimization, answer engine optimization, it was his content. It's still his content, right? And you have to be putting up all the content, you got to be putting it up in certain structures or certain schema and technical underpinnings. A lot of businesses don't know how to do that.

Jordan:5:56
Yeah.

Joseph McElroy:6:20
And even if they do, they just basically put in some standard plugins, but they don't really understand them. They don't configure them right. So with AI, you can have your content. One, you can actually conceivably generate content. I've created a content production engine that's pretty good. And I see it through things with emotions in certain ways to make it sound more human. And then...

Then it goes through a layer of viewing it with all the technical, structural things that you need to do in content to make it a good citizen of this ecosystem online that gets you visibility and authority. So having that produced automatically, you can work with, say, one of the really smart AIs once a month, generate a content plan. and you're in the spreadsheet and then you have the automation go there every day, pick up what's supposed to be published today, create it, put it in. You can have you to automatically publish it or have it come send an email to you for you to look at it and approve it. Right. It's approved. And once it's up there, it's got everything perfect in terms of what you need technically and, you know, and, and, and structure wise and in the linking with other content on your site and everything that's supposed to happen is done. You don't want to do think about it. Right. And then it can automatically also promote it.

It can also push it out through, you know, let various social media know, or it could go through whatever your social pilot, or it could work through an API to work with whatever systems that you're scheduling, put it in there to be scheduled. It can actually embed it in a newsletter. The newsletter can be optimized. Also, the AI can be an automatic process. It happens every week. I'm doing one for a magazine right now where it's picking up events and scraping for events all over the place. We're bringing in a of articles, bringing in some news, making a newsletter, bringing in advertisers and doing it all. So it just does it automatically every Monday and you don't even have to think about it. I the AI creating the hooks, right. they get people to read it. Yeah. And, based upon, you know, you have a certain structured data that you can provide to the LLM. So it knows where you are and what you're talking about. And it does a really good job at hooks.

Joseph McElroy:8:45
Yeah, and I'm particularly interested in that because I did a lot of work with headlines for Marriott and then I had my own little commercial motel space and I worked with something called memorable experiences and I learned how to put emotional words into the headlines of things and get a higher click-through rate. So now I can have AI do that.

Jordan:9:09
No, definitely can. I actually utilize a similar process internally as well, where I have AI draft up, not draft up. I have AI create a fully written blog post that's optimized for AISCO or geo. It uploads it to my website everywhere at 8 a.m. PST. Every Monday, it sends out a weekly newsletter on one of the blogs that was just written, whatever.

Joseph McElroy:9:27
Enough.

Jordan:9:35
AI like whatever AI feels like it was the best blog for that week. It was that every Monday. It's pretty interesting how you know how much it can actually do and I don't think people understand it's full capacity. I capacity, you know.

Joseph McElroy:9:39
Mm-hmm.

Joseph McElroy:9:52
Oh, I mean, you'd be. No, I agree. mean, you imagine, imagine that now you put this into a say, a us a motel or hotel, right? Not only is a newsletter going out online, it can be automatically printed and put into every room, right? It could actually be the staff arrives in the morning and there's a stack of printed newsletters that can serve as what's going on this week and some interesting things to read and get you know, and you can a little magnet in. Yeah, you can start.

Everything online to reality becomes something that's totally possible with AI. Right? Yeah.

Jordan:10:26
I feel like the only limitation in my opinion is the imagination behind the user. Especially with API keys and open source and the fact that we can utilize Claude code now. We can just connect so many different softwares without utilizing N8N. I still utilize N8N, but I also utilize Claude code and Railway and other, I guess, softwares to connect everything, but.

Joseph McElroy:10:35
Hmm

Joseph McElroy:10:41
Yeah.

Thank

Joseph McElroy:10:51
Yeah.

Well, everything is going to become agent. Everything is, you look at every software, everything out there now, they all have agents now embedded in them. Right? It's going to become this world of, you know, all these specialized agents that are organized by a smart brain agent. But, know, there's going be a lot of agents you don't have to create anymore. Right? It's going to be crazy.

Jordan:11:19
Like a master agent and then the sub agents from the master agent. Yeah.

Joseph McElroy:11:23
Well, there's gonna be agents that you need to have that you develop that pull from your proprietary information, right? And inform whatever's going on so that it meets your brand and you have, it's really important now to have a narrative spine, all right? Because the narrative spine of your business is what makes people trust and believe in you across all interactions, right? So what is the story?

You know, and how does it build trust? And you have to have that. You have to have your A, your, your big brain, LLM know that intimately so that it was looking at everything that's happening and make sure nothing violates that narrative spine. And if you do that, the A is not going to screw up the trust of your company.

Jordan:12:08
Yeah, I think that's something really important is ensuring that the AI agent understands your brand, your voice, how you write, how you speak. Because if it's, if it sounds like AI, it's not, it's not going to convert. So you really have to train these models. You really have to train these agents to your liking. And I think that's kind of where people, I guess, stumble and fumble is because they try it, you know, once maybe even 10 times. like, it's not working. It doesn't sound like me, blah, blah.

Joseph McElroy:12:21
No, yeah, yeah, yeah.

Jordan:12:37
It's like, dude, it took me months, months to train like AI assistant or AI agents to talk like me, know my business, et cetera. You know, it just takes time. It's doable. It just takes time.

Joseph McElroy:12:48
Yeah, it takes time to know that's the interesting thing about investing into the AIs. know, I've dealt, I spent a lot of time for a long time with chat GPT, right? And at this point, you know, now it knows me, it talks like me and everything else, but, you know, I'm not exactly secure with that, right? I'm like, now what happens if chat GPT change, it goes away, things like that, or there's a better AI I want to move to. Now I've got to move all that. over. And so what I started to do is, you know, bit by bit taking old conversations, old things, you know, getting them and putting them into my own local LLM that I own and control and then able to have it make it accessible when they need it to be to other LLMs. And of course, then you get the second brain things like obsidian and things like that.

Yeah, or a different way to hold that information, which I'm going to start investigating now. I haven't done a lot with that second brain concept, but I think I'm going to.

Jordan:13:52
know within Claude, it asks you, like, do you want to transport your memory from chat or Gemini into Claude? And there's a way to do that. I'm not going to get into it now. But what you're saying, a lot of people have that same, I guess, issue or fear or challenge. It's like, well, if I switch LLMs now, I'm going have to restart. It has no context as to who I am, et cetera, et cetera, which is a valid point.

Joseph McElroy:14:01
Yeah.

Jordan:14:19
I just don't think Clyde you can transfer memory when I say that. But I have a question. You said you are building your own LLM. When you say that, are you...

Joseph McElroy:14:21
you

Joseph McElroy:14:28
I mean, I'm using a normal one that exists, but I'm building out its memory and its data and its, its rag infrastructure and everything else. Yeah. I'm training it basically.

Jordan:14:38
You say it all. your own LLM or like an open source from a previous model of an.

Joseph McElroy:14:44
It's an open source LMA. called, what I'm using is called Ollama, O-L-L-A-M-A. Right. And it's really designed to be on my, you know, on many environments like a laptop or a Mac mini or even a small server. And you know, what happens is Google has developed a model for that called Gemma 4, which is

Jordan:14:49
Yeah, yeah,

Joseph McElroy:15:10
from what I read is highly optimized for those environments and supposedly approaches commercial performance on just regular CPU level hardware, consumer level hardware. So, mean, obviously it's not meant to be with millions of users using it, but for small teams using it, it should be pretty.

I'm reading. I mean, I use an older version right now. I'm going to be updating to this soon to test out that what they say is the improvements in, you know, reasoning and speed.

Jordan:15:52
On that topic, I still think SaaS companies are going to be pretty valuable because there's going to be individuals out there that don't want to learn the skill, that don't want to build their own infrastructure. But guys like you and I, Joseph, it's like, why would I pay an extra $1,000, $2,000 a month in software when I can just build it, build in my freaking soul and have more control.

Joseph McElroy:16:01
Thank

Joseph McElroy:16:13
The average small business, the middle sized small business has over a hundred SaaS products. A hundred million dollar company will probably have 200 SaaS products with multiple users. They're spending thousands and thousands and thousands of dollars monthly. Just getting into my SaaS, a lot of my SaaS stuff, I've saved over about $1500, $2000 a month.

Right. Yeah. It's like, you know, the savings is, is, yeah, at this point that you still have to put a lot of thought and mind into it, but it will get more more doable by even the average person. And of course there's going to be any layers like me out there saying, come, I'll do it for you.

Jordan:16:58
And that's exactly your, your sure position, your value, your skill, your expert is not going to go anywhere. Like people will definitely.

Joseph McElroy:17:04
Yeah Yeah, I'm a systems guy. A lot of the people out there doing AI right now are terrific app people. They're making wonderful applications. I'm not a clever app guy. I'm a systems guy. I make the systems work. So that's what they come for me for.

Jordan:17:21
Say it.

Absolutely. Let's see. Where do you feel like the future is going with AI, you don't mind me asking? We alluded to it, but maybe you can elaborate.

Joseph McElroy:17:36
Well, think that you're going to have this, I think that you're going to have a movement to, I think it's going go both bigger and it's also going to become more insular. All right. Those are already big companies and big governments talking about sovereign operational systems and they're designing it because governments especially start being afraid that SAS has control of their data. Right. And there's a lot of privacy concerns with that.

Yeah, HIPAA is pretty difficult to get to work on. Companies with HIPAA compliance have a little bit of a difficult time working with AI and opening up their skirts too. So there's going to be, I think, a lot of movement to these localized AIs, right? To hold the smarts and hold the precious data and the things that they don't want the world to see. Hotel Change would never use program advertising for years because the ones providing the program, if they want to have access to their servers and their data and they protect that data like it's gold and it is gold, right? Yeah, so there's gonna be a lot of this where there's gonna be nodes in every office, right? Yeah, that are doing AI and having brains and things like that or every server farmer or their co-location facilities because they're gonna protect it.

And maybe somebody smart will figure out a way to do that on us on a scale and it'll be out there but still you know, there's gonna be you know, this drive to create these the fastest best giant is you know AI Yeah, they're gonna have to become the Jesus moment of like how do we protected it enough? Right? Because there's too many things coming out where the the AIs are not behaving as predicted and while they're not I do not think that there's any sentience there. I think that they just follow logic and directives. there's a lot of times that the way a directive is given, it becomes not completely understandable because it's compromised by other directives. And the minute that happens, the AI has a hole. So, and you know what?

Joseph McElroy:19:54
A lot of times now directors are coming from multiple places, right? And so that's going to create all these holes and how are we going to, you how is there going to be a layer that says, listen, these are fundamental, you can't violate these. They haven't done that. They haven't done that because they've relied upon prompts and other things, but there needs to be a more stringent layer that just says, hey, you're not going to try to blackmail people. I mean, you see that recently?

You know, there was instances where somebody denied, you know, AI access to something and they did research and found stuff on them and tried to blackmail them, even though he'd been directed not to, but there was loopholes. And then there's been other cases of that, right? So it's like, yeah, it's funny, but it also, yeah, it won't be funny when it happens to you, right? Right?

Jordan:20:48
Right, On that topic though of AI kind of doing its own thing, I know within Claude, you can create a file called dot Claude dot MD. I think it sits at the very top of your files. essentially whenever Claude is spitting out output, whatever the fric, the output is, it does not matter. It always reads the dot Claude dot MD file. That those are the hard rules that you're referring to, like never do this or never do that.

Joseph McElroy:21:15
Right.

Joseph McElroy:21:21
You create them, You create them. Humans create those, right? So you can make a mistake easy. Anybody could, right? So in terms of that ultimate like, you know, things that they just cannot do every, you don't really rely on every person in the public knowing how to make one of those files work. You need to have something in the system itself, say, hey, this is,

Jordan:21:24
I do.

Jordan:21:31
I always make mistakes.

Joseph McElroy:21:50
This is that bottom line. This can never be violated. Right?

Jordan:21:56
think the LLM company should have that as like a mandatory rule for every. I think they should, but, getting back to what you said about models and sometimes the chat or the, output doesn't follow instructions all the way. So what I've learned is that yes, the prompts, the rules to the output definitely need to be clear, constructive, and orderly. Also,

Joseph McElroy:22:00
Yeah, oh yeah.

Joseph McElroy:22:22
Mm-hmm.

Jordan:22:23
It also depends on the model that you're using. I don't really use ChatGBT anymore. I always use Claw. I just think it's better for whatever I do. No offense, you want. Cool. For Claw specifically, there are three different models. There's Opus, there's Sonnet, and then like Hayaku or something. I don't know to pronounce it. Okay. They all serve different functions better. Opus is really good at taking your instructions and then being creative.

Joseph McElroy:22:32
I'm sorry.

Joseph McElroy:22:42
Yeah.

Joseph McElroy:22:53
Mm-hmm.

Jordan:22:53
and make more fucking output, right? It's like, dude, I don't want that. Sonnet follows instructions perfectly. It says, okay, here are the examples, here is the rules, or here are the rules, I'm gonna follow this and execute. Cool, Opus does do that. However, if you're looking to have a more creative project, right? Creating long storyboard or code, like Opus is really good at freaking coding and reasoning, you wanna use Opus. So it really also depends on like what

Joseph McElroy:23:05
Right.

Joseph McElroy:23:19
Yeah.

Jordan:23:23
model you're using to get to the desired output.

Joseph McElroy:23:26
Yeah, and that's important when you're coding and when you're actually coding automations, all right? Is you know intimately what all of those models are best at. And like you don't want, you just want to summarize an email. You don't send that to the top gun of a model because you're gonna, and emails are usually huge, especially if you actually sent a...

Jordan:23:32
Yes.

Joseph McElroy:23:49
a long chain as the email. I ended up spending a couple hundred bucks on just summarizing one email before I realized that.

Jordan:23:51
Okay.

Jordan:24:03
Oh, yeah, you have to watch your context tokens and like actually what's being said, because obviously the more the newest model are always going to be the most expensive in this. The older models like Sony in this example, or maybe like chat, GBT, I don't know, 4.5 or even five. Oh, whatever. It's still really good. And it's definitely not as expensive as the newest model, you know? So if you're looking

Joseph McElroy:24:11
You

Joseph McElroy:24:20
Yeah.

Yeah.

Joseph McElroy:24:26
And there's open source ones that are great for doing the things like summarizations. They cost you nothing. right. So doing a lot of things that doesn't need high advanced reasoning is where you take advantage of the open source ones that are local.

Jordan:24:32
Yes, patties.

Jordan:24:41
Honestly, Joseph, I just consumed a video about what you're referencing right now and I looked into it and I like almost jumped off a bridge, bro, because that is so complex to me and I don't have the time to build my own open source LLMs. Would I want to? Of course! Do I have the time? No!

Joseph McElroy:24:59
Well, they also come pre-loaded, they come pre-trained. You can get them pre-trained, right? And you can get one installed and it's working beautifully right off the bat. Now, if you want to know you, you have to then do the stuff, the chunk information into say a Postgres SQL database so that it then can pull it out in an efficient manner to write content that's specific to you.

That is a little bit of a tech thing, but that's going to become easier and easier. In fact, I'm writing automations to do that. So everybody will have, upload your PDF, it'll chuck it up for you, put it right into that RAG database and whatever other kind of, there's now becoming more ways to get it to remember stuff. And yeah, some people say obsidian's better, but we'll see.

Yeah.

Jordan:25:59
Two questions. One, what's obsidian? Second question is are you using N8N for O llama and building your own LLM or different?

Joseph McElroy:26:07
Yeah, I'm using N8N as the workflow level. So I'm using workflow level there. I can put some basic agents in. I don't know if you know that N8N now has got agents. One are workflow agents, the other are of personal agents, which could be more free flowing, do make decisions on its own. Yeah, so I'm just now starting to play with those.

Jordan:26:29
I haven't played it.

Joseph McElroy:26:35
I'm watching OpenCla and some other things to see which one I'll eventually put in. At an automation level for a business and you're putting in a system that's running the business and doing the marketing, it's not doing the lot of stuff like going to figure out what they're doing for a business. So it's got to be creative. I'm not there yet. But once we get all the basics in, we have all the workflows that somebody might need, I'm going to be introducing. you know, this kind of decision making capabilities. have have decision making along the process, like deciding a lead is a lead, you know, intent or is it lead, you know, commercial is a lead for, you know, what kind of, what kind of, you what kind of, you know, what are they interested in buying, know, interpretation and, you know, without sending it to the right people, it's not the right thing. So there's a lot of AI that you can do, you know, decision making that you can do in workflow automations, right.

Well, it's not going out there deciding how to make you a million dollars today. those are, but we're getting there. Yeah. But right now I don't think there's enough systems that can just run a business. The basics of a business. There's lots of flash everybody's doing, which is good. You need to flash, but there needs to be a lot of systems that are built just to run a business and then view it with the smarts.

Jordan:27:55
I agree. Where do you see yourself and turn into the marketplace? Would you consider yourself done for you agency that continues to optimize these systems or would someone pay you like a one time installment fee to install the systems and like you, you turn the keys over to them.

Joseph McElroy:28:13
Well, I guess I'm going back to the old days, an integration company, right? Right? You you put in a server and you have the software running in there and then you, you know, runs the business and then you do smarts in there and then you can connect to the rest of the world. yeah, so essentially as a developer, I will eventually have pre-packaged solutions for just different things. And I'm focusing on you know, right now media and hospitality, know, so websites are trying to make money from content and media plays and then, and then the hospitality industry, you know, you know, wants to build their business, you know, and do things, you know, smart with that. And then founders, you know, I've done a lot with founders, so single operators that are, or duos that are building up, you know, their businesses. So essentially either building it Yeah, some businesses will build our servers internally. Now what I do is I set up a dedicated server and I create VPSs for clients and then build it on their VPS. I build a whole infrastructure, right? And it's pretty flat now, but you know, and I give them like for the media companies, I give them the ability to send out a million emails a month for a hundred bucks, right? Yeah.

Jordan:29:35
Okay, but I have to question this because I actually do. I utilize email for outbound systems on behalf of clients. So when you say a million emails per month for a hundred bucks, is that one domain you're sending all of the emails from?

Joseph McElroy:29:42
Okay.

Joseph McElroy:29:49
Yeah, you can do that. got to build up a very trusted list. It's using Amazon SES, right? Amazon SES can scale to pretty much any size you want. And it's actually a per transaction cost and it's not, it's very inexpensive. It's the most inexpensive solution out there. Like I said, you can do like a million emails for a hundred bucks.

Jordan:30:14
It's a full, like you said, the list has to be trusted. can't mark your email.

Joseph McElroy:30:18
Now, if you get an untrustworthy, you know, you get an unhealthy score, they'll knock you down. You they won't look. So yeah, you can't just doing blasting out there, getting a lead list and blasting again. So what I do is I set up two different CRMs. I have one that uses, that's going to be build the audience layer, right? The audience layer is just going to your audience and you know that they're audience and you built them up preciously over time, right? And that's why it's for media property because it's a trusted. audience, then I have another system as for like getting leads, know, sending out in different ways. You can set up different domains to do the emails if you want to, however, whatever, you know, there's multiple processes to do that where, you you sometimes have to burn through domains. Yeah. Yeah. Yeah.

Jordan:31:04
I know all about that for sure. Let's see, my friend. You were doing SEO for, I guess technically you might still do SEO for hospitality.

Joseph McElroy:31:18
Yeah, I do. I've made it more mechanized now. know, SEO and AO, through this, yeah, I can do either the whole system, I can do just the parts that do take content providing. I can stick it in anywhere. can give them their own instance off of my servers, have it do the automations for them to format their content, go through the content, find underperforming, you know, pages and rewrite them, right? Yeah, and different things. you can do a lot of what you need to do for AEO and SEO automated. I there's gonna be some external things I would get to, know, link building and stuff like that. But there's also some good programs out there for that. Yeah.

Jordan:32:05
I'm starting to utilize AEO or GEO for my blogs and I haven't, well, I shouldn't say I haven't seen the results yet. I mean, if you look up my company name in LLM, it will show you. I haven't generated traffic from it's one point, but we'll see, we'll see. It'll take some time.

Joseph McElroy:32:21
Yeah.

Joseph McElroy:32:25
Well, here's the thing I'm experimenting with, and this is very technical, you know, how they take that content to put into their rag, right, and make it, you know, and keep track of it, or put it in, and put it in anything. Basically, they convert the content into numbers, right? An array of numbers, all right? It's a vector. And basically, all those numbers represent the ideas and meanings that are in that content, not the words. So when, you know, AI searches, it's not searching. for words as searching for meaning, right? So if you chunk up your content on your site, you could go through and conceivably look at those vectors for that content. And the more numbers it has, that means the more meanings it has, the more information it has, which means it'll show up more often than the AI results. So it'd be a way to test your content, right? To see if it's actually gonna You know, have a broad set of meanings and information that AI will consider important and bring up. Yeah. I just, I wrote about, I got into the nitty gritty of that recently and I was like, wow, this could be a way to really understand if my content is going to show up a lot because it's got a lot of vectors or doesn't, it's just, you know, one number. It's just expressed as one idea.

Jordan:33:30
That's it. I actually didn't know that. I didn't know that.

Joseph McElroy:33:52
Right? But if it expresses 100 ideas, you know, it has information. You they tell you about the AI, try to get as much information into the content as possible because it's pulling up chunks. Well, that's the reason, because it converts it into vectors. Right?

Jordan:34:08
You're saying it's wiser to make the blog post, as an example, really, really condensed with tons of information opposed to having one big idea in many different blog posts.

Joseph McElroy:34:19
Well, if you, you, or what you do is you really segment it. have, uh, you know, one segment full of information about something. Then you have another segment on the same blog post. That's full of information about something else with a nice heading, nice heading, you know, heading content, heading content and content is full of information. You have, you know, then you have tables you have.

You know, other things to support that you have an FAQ in there. You have good schema all the way around telling you every which way is possible to be interpreted by this. You know, and, you know, and that'll get you better results.

Jordan:34:55
Gosh, you, okay, okay. I'm definitely gonna look into that when I have time tonight. Thank you so much for that. Let's see. I guess on a personal note, for you in the business, I guess what are you guys looking forward to in the next 12 months?

Joseph McElroy:35:09
For us, I'm looking forward to having a platform into 100 businesses, doing automations and marketing for them in a way that they're pretty happy about. And I'm developing a reseller network that can build on top of this. Because it's not particularly expensive platform to put in. It's all open source.

Jordan:35:25
I love that.

Joseph McElroy:35:38
is zero other than the time put into installing it. you know, the sewing it together is a little bit of a challenge, but I've managed to do that. Yeah. One of the things I would just say, you talk about founders, you're talking about small media blogs, things like that as a base. They, know, a hundred dollars a month, $200 a month, you know, $500 a month becomes pretty expensive. Right.

Jordan:35:50
That's my channel.

Joseph McElroy:36:06
So a lot of open source, yeah, they have good smarts and you can use them, but they cost like a hundred bucks a month. So what I do is the entry level, I'm not trying to replace all that, because eventually somebody grows up to not use all that, but the entry level, I sew it together by creating plugins or whatever to work at the underneath all that security stuff. I don't know if you know, but you can pretty much bypass in WordPress, any kind of API arrest prevention. you know, through internal calls, right? And make everything sew together. So that's what I've done.

Jordan:36:44
I know you can connect tools with API, but I don't know if I understand.

Joseph McElroy:36:47
You can bypass that too. If both of your tools are running in WordPress, then there's an underlying layer that works with WordPress. can work with WordPress because they have to open up things to WordPress. So then you can come in through those same things. You can find the same ways they interact with the data barriers.

Joseph McElroy:37:15
It's a way to do it. Now the problem is every time they upgrade and they change stuff, it could just destroy your plugin. yeah, now we have AI code writers. You put it into the code writer and say, just update this for the current race. It's not so hard anymore. I can't cloud write my flows for me.

Right? You just have to really detail it out. But I get it half the time. makes my automation nodes, you know, for me. Right? And I just cut, copy and paste. it's like, look what I did.

Jordan:37:55
That's essentially how I use Ended End. say, do this for me. says, OK, I copy and paste it into Ended End, and it works like a charm.

Joseph McElroy:38:03
Hahaha!

Jordan:38:05
Listen, if someone wanted to explore your work further, because they definitely should be interested in what you are articulating, it's very, important. Where can they find you?

Joseph McElroy:38:11
Uh-huh.

Well, you got my website, GalileoTechMedia.com. I also wrote a book called Wild Ducks Don't Fly in Formation, which is about this life we're living. And I had gone through a lot of adventures and catastrophes and how to be both resilient in life and in business and reinvent yourself as necessary. And I have a community developer around that called, it's at wild ducks.net where you can either. you know, buy the book or, you know, set up a meeting with me or you can go to, I have a discourse community that you can become part of. Well, we just talk about, you know, everything from, you know, freedom exercises and freedom to AI to, you know, marketing to tech, you know, and I got some pretty good people in there, you know, that have some good conversations. So you can reach me in that, of course, I'm online, LinkedIn and things like that.

Jordan:39:15
That's awesome. I had such a pleasure chopping it up. It was really good conversation regarding AI. I learned a thing or two for sure. Let's see. Just want to say thanks for hopping on, man. Absolutely. All right, audience, we'll talk to you later.

Joseph McElroy:39:25
Sure thing, thank you for having me too. All right.

Have a nice day.