UnHacked - Cybersecurity Made Simple for Small Businesses

Justin Shelley (00:15)
Welcome everybody to episode 88 of unhacked guys. There is something that I love and hate about our new topic of AI ⁓ I love it because it seems to have breathed new life into this group we've been talking about cyber security for two years plus and Listen, it gets old. We're talking about the same thing in different versions ⁓ Now we come on here to record We're 27 minutes past our intended start of record time and I can't you guys just shut up so we can start recording

Joshua Holloway (00:41)
you

Mario Zaki (00:44)
Haha!

Justin Shelley (00:44)
And

you said some things I wanted to have been recording, but I missed it because as I mentioned, I can't get you to shut up so we can start recording because I was so excited about AI. And two of you said, I hate AI. think somebody said it. I'm like, I'm like, wait, wait, wait, I want to record that shut up. Um, and then somebody mentioned something about an employee getting fired and that went super sideways.

Bryan Lachapelle (00:54)
You

It's a love-hate relationship.

Joshua Holloway (01:09)
Thanks.

Justin Shelley (01:09)
I'm going

to do a basic intro and then I'm just going to turn that loose. I want you guys to talk through what we should have been recording, but I couldn't get you to stop so we could record if you're tracking me. Anyways, Justin Shelley here, CEO of Phoenix IT Advisors. And since 1997, guys, I'm old. I've been working with technology. I used to be a computer repair guy. Now I'm a cybersecurity guy turned AI, whatever the hell we are these days. But what I love to do, my passion is to help businesses grow, build their wealth, their business.

earn money and then keep that money from them dirty rotten Russian hackers, greedy attorneys who come around and sue you and the government who can't stay out of it either. That's what I do. Josh, you're up next. Tell everybody who you are, what you do and who you do it for.

Joshua Holloway (01:52)
I'm not going to say with the same amount of enthusiasm, but Joshua Holloway, CEO for 7th EI Technologies, also been working in computers since the late 1900s. Our primary focus is ⁓ computers and networks with compliance wrapped around it. Go ahead and take it away,

Justin Shelley (02:06)
late, late 1900s. like, I

swear I heard 1800s. I'm like, listen, bitch, you're not that old. Anyways. All right, Mario, come on, Josh, let us down on the energy level. Bring it back up, Mario, full energy.

Joshua Holloway (02:14)
Yeah

Mario Zaki (02:20)
Yeah,

Mario Zaki CEO of Mastek IT and we help small to medium-sized businesses stay safe not just from the Russian hackers but from China, North Korea, South Korea and the rest of the world. We don't just have one country that we're focusing on. We're focusing on the entire globe and we've been in business for a few weeks and we're still trying to figure it out.

Joshua Holloway (02:34)
So many targets just got put up.

Yeah

Justin Shelley (02:44)
Mario doesn't

just want to go to one war. He wants to go to war with the whole world all at once. ⁓ It's dangerous tactic, Mario. Brian!

Joshua Holloway (02:48)
you

Mario Zaki (02:51)
Remind you of somebody in office right now?

Joshua Holloway (02:54)
You

Bryan Lachapelle (02:56)
⁓ Brian Lashmore with B4 Networks based out of beautiful Niagara, Canada. I do all the rest of the crap that these guys talk about, but I also help every single one of our clients in their journey to become a better ⁓ business by using technology and leveraging technology. And I also remove the headaches and frustrations that come with dealing with technology so that you can concentrate on doing what you do best.

Justin Shelley (03:22)
All right, now I'm going to go back to the stuff we should have been recording. guys who here hates AI and why Mario's got his hands up.

Mario Zaki (03:31)
Yeah,

I want Brian to help me take out the frustration that I have with my technology because, and I will quote my AI agent that I've been working on from the last couple of weeks. It actually sends me a message saying, honestly, I've been leading you down a rabbit hole here. I'm not confident enough in where exactly the setting lives in. So maybe we should go a different way.

This is literally what it told me, it's like, I've taken you down to rabbit holes, pretty much telling me it's wasted the last, I think at that point it was like, two and a half hours of my time, and it says, yeah, you're shit out of luck, I have no clue.

Justin Shelley (03:59)
my God. ⁓ God.

Bryan Lachapelle (03:59)
soon.

You were missing a specific prompt, Mario. There's a prompt that I can send you that essentially says something to the effect of never assume you're right. Always question your own answers, blah, blah, blah, blah, blah. Give every answer a confidence rating so that I know how confident you are. And always double check your own answer against, you know, published standards or something like that. I can't remember the exact prompt, but there's a whole prompt that somebody created and I just might copy paste.

Mario Zaki (04:36)
I will see you you

help me take out the frustration. There you go.

Justin Shelley (04:41)
Well, can, can I just add that guys, you know how AI works, right? It is not a brain. It is not a consciousness. is mathematical statistical probabilities period. has no awareness of what it's saying. It doesn't know if it's you leading, leading you down the rabbit hole. does this like wild ass calculus computation and comes up with the probability of the next half a word of what it should say. And it'll throw out like a hundred, but the one with the highest probability is what you get.

Bryan Lachapelle (04:42)
Yeah!

Mm-hmm.

Justin Shelley (05:11)
So I'm just.

Joshua Holloway (05:11)
Well, and don't forget

that it was programmed to please you. the program will give you an answer. It's a golden retriever. Like it's, it's, it's trying to make sure that you're happy.

Justin Shelley (05:14)
You're right.

Bryan Lachapelle (05:18)
Yeah, yeah.

Mario Zaki (05:21)
And don't get me wrong, it's taken me down hundreds of rabbit holes, but this was the first time it actually admitted it without me yelling at it and say, what the fuck are you doing? No, this is the first time. actually, like I was actually following along and saying, okay, well that setting is not there. What do you, you know, can you tell me where exactly that would be? And then all of sudden it just kind of gave up on me. Like, you know what? Fuck it.

Justin Shelley (05:21)
Yeah, mathematically.

Bryan Lachapelle (05:21)
Even when it's wrong.

Joshua Holloway (05:23)
Yeah, even when it's wrong.

Justin Shelley (05:33)
Really?

Mario Zaki (05:48)
You I can't figure it out. I've been wasting your time. I can't use up anymore of damn tokens.

Joshua Holloway (05:51)
I can't fool you any longer. I can't make up anything.

Yeah.

Justin Shelley (05:57)
Mine just did that this morning. It's just like, Hey, I was wrong. My bad. let's try this. And it didn't take me very far. Like it didn't go down a rabbit hole or anything. So maybe my AI is smarter than yours. Which one are you using?

Mario Zaki (06:09)
I was using Sonnet at that time, but I switched over to Opus. I'm like, that's it. I'm speaking to your boss.

Justin Shelley (06:11)
There you go.

Joshua Holloway (06:17)
Well, I wonder if there was a change to to to the LLM where it's, it has its own probability where it will give up and it will stop, you know, because I think a lot of people are complaining about the whole golden retriever aspect of like, I'm just gonna keep giving you info and giving you info and giving you info even though it's incorrect. But I feel like I'm facilitating your needs. And everything's happy until well, Mario decides to, you know, kill it or yell at it or

Bryan Lachapelle (06:19)
Your boss is a man.

Justin Shelley (06:19)
Alright.

Mario Zaki (06:27)
Maybe.

Well, wait until like if they ever come with an unlimited plan where doesn't tokens don't matter you better believe Those answers are gonna be a lot more accurate because right now I do see you know AI kind of Sometimes breaking something or doing things on purpose to go back and fix itself so you can use more

Justin Shelley (07:06)
or the, the never ending chat that is like, now that I've answered your question, I just have one more question for you. What do you, and then it like starts a whole new conversation like, fuck you. I'm done. God damn it. done. Um, I, I will, one thing that I heard like a long time ago, way back in the infancy of AI, we're like a year ago. Um, somebody, somebody said, just assume that the version of AI you're using today is the worst you'll ever deal with.

Mario Zaki (07:12)
Mm-hmm.

Joshua Holloway (07:13)
Yeah.

Mario Zaki (07:16)
You

Bryan Lachapelle (07:17)
Shut up!

Joshua Holloway (07:27)
1956.

Justin Shelley (07:35)
it's just always getting better and it's getting better so quickly. ⁓ that is true. Guys, do you want to say anything about there was a rumor of an employee who was terminated? ⁓ I'm just reminding all of us that this is a cybersecurity podcast. AI is way more fun and exciting at the moment to talk about, but I want to remind everybody of the dangers that we face with technology. And it isn't just the outside threats. We've talked about it, but we haven't really talked about. And now we have a real example of

an employee who was terminated, who left some potential damaging things behind. Anybody want to say anything about for security reasons, for privacy reasons, we can't give a lot of details here. You guys have anything you want to discuss on that?

Joshua Holloway (08:19)
It said no one named names. That's not allowed. Nobody named days. ⁓ yeah, go ahead. Mark.

Justin Shelley (08:21)
Nobody name names.

Mario Zaki (08:25)
I mean, pretty much

with the theme of what we're going to be talking about for the next couple of weeks, 12 weeks, 11 weeks left, is that it is now much easier to run a script or a code that anybody can, even if they're not a developer or programmer, they can easily now be able to go on to chat, know, chat GPT or Claude and have it create a code for them.

to do malicious activity.

Joshua Holloway (08:58)
That's actually a great point because your onboarding process has just become so much faster. You really need to be on top of that onboarding process and then now you have security audits that we pretty much have to run. Because they can script whatever they want because they can ask for whatever they want. know, case in point, how about a script that keeps trying to log into Office 365 and reactivate a user account? And just so that that's objective permanence, right? Like every time we go in and disable it,

There's a script somewhere on a computer that's on and it's re enabling it, giving that that person access, you know, and you sit back and you watch it and you're waiting, you know, you're killing it. You're finding it again. You're deleting it. And that was just within what the first five minutes of that person leaving the room. So our onboarding has gone from, yeah, we're going to do it right after they walk out to the door to

Bryan Lachapelle (09:28)
Thank

Joshua Holloway (09:55)
I think we need to do it about five minutes ahead of them, bringing them into the, like as they're bringing them into the room. Yeah. And then it really needs to be involved now because cloud code, can code whatever script you want in the background. You could park it pretty much anywhere. ⁓ and then you're opening doors or you're opening access to accounts again. It gets pretty scary.

Justin Shelley (09:59)
Before notification. Yeah. Yeah.

Bryan Lachapelle (10:00)
while they're in the room.

Mario Zaki (10:18)
Now does this have to be ran on a live computer or can they just put a script into PowerShell and says check every five minutes of this account is active?

Joshua Holloway (10:27)
Yeah, you can do a PowerShell script. I mean, you can even do a persistent Windows scheduler, like a scheduled task. You know, it just fires off every couple of minutes. Like, I think right now there's probably one person devoted to looking at the logs and probably will be for the next two hours, easily.

Justin Shelley (10:34)
Yeah.

I know I was just

going to say who of us looks at scheduled tasks on every computer that's being managed. You know, that's, I don't think we've even talked about that one. I mean, it comes into, a lot of the security frameworks, you do have to look through logs and have alerts and stuff set up.

Mario Zaki (10:55)
No.

That actually brings up a pretty good point. Maybe ⁓ we're all using RMM ⁓ agents. Maybe we could have it give us an output of every computer's scheduled tasks.

Joshua Holloway (11:17)
Well, and to bring a full circle back to AI because a human's not going to look at all those log entries and the AI should be taught to be looking for those things.

Justin Shelley (11:21)
Right.

Yeah. Well, and to tie it back into what we're talking about, guys, I want to point this out as what we are going to teach over the next remaining, is this just week two? 10, 10, 10 weeks after today, ⁓ we are basically going to teach you how to develop an employee that you're going to turn loose in your organization. And the security ramifications are huge. So while we're going to show you some basics, we're also going to

Bryan Lachapelle (11:24)
Yeah, absolutely.

Joshua Holloway (11:26)
because they can see a lot faster.

Mario Zaki (11:35)
This is week two.

Joshua Holloway (11:35)
This is Week 2.

Bryan Lachapelle (11:36)
Yes.

Justin Shelley (11:54)
teach the proper way to lock this stuff down. And, you know, there there's an example I'll, I don't have the details of it. I'll bring it back because this is going to be a recurring theme of somebody who was developing and we're going to, know, our segment three, the last four weeks of this series, we're talking about vibe coding, right? Developing applications using AI. And somebody used that to successfully delete all of their, their entirety of their database. And I don't remember again, the details, the backups, like everything is just gone.

Bryan Lachapelle (12:19)
critical tent.

Justin Shelley (12:23)
Thank you. Thank you. Vibe coding. so this is dangerous shit. Like we're, giving you a torch in the middle of a, you know, a dry, I dunno, I live in sage brush. So you can, you can look at the sage brush wrong out here and it's going to catch on fire. That's what we're teaching you guys to do. So we are going to get very serious about the security as we go along through this. ⁓ one more thing I want to talk about. I, I dropped two weeks ago.

Joshua Holloway (12:23)
wiped.

Justin Shelley (12:48)
that we are going to, you know, in the process of developing a cybersecurity portal that will be available to all of our listeners. I made a promise that I was going to have, you'd be able to log in the first week and then by today it was supposed to be able to do more. It doesn't, it's not ready for any of that. And I'll tell you one of the worst you guys are all started with, I hate AI. Here's what I hate about AI regarding this security portal. In my mind, I had this very simple, basically a checklist that guided you through

Did you do this? Yes or no. Did you do that? Yes or no. If not, here's how to get some help. Right. And then I type that into my little development platform, like, Hey, build this portal. And it spit out this thing that was just phenomenal, but it was only a, like a template. It was all working. You can go and click and you can set whatever. ⁓ but it's, it's doing great things like, Hey, if you want to help with this, go listen to episode 74 of unhacked starting at minute six.

Bryan Lachapelle (13:19)
You

Justin Shelley (13:43)
I'm like, well, shit, now I've got to go through all the episodes and figure out how to match it all up or get AI to do it. So anyways, it's going to do way more than I initially envisioned, but it is going to take me longer. And that said, I don't really have much to show off this week other than just I'm like, God damn it. I had a simple idea. Now I've got a way bigger idea and it's going to be amazing. And it's going to take me a shit ton of time. ⁓

Bryan Lachapelle (13:47)
Hahaha!

Joshua Holloway (13:48)
Thanks

Mario Zaki (14:03)
And you

know what, to Brian's credit, I mean, this is a perfect example. You know, earlier, before we hit record, he, he introduced us to a, an AI agent to help us organize all our AI agents and all our AI ideas, because it just goes forever. You know, you end up one idea leads to 16 others and those 16 lead to 50 others and

Joshua Holloway (14:21)
Mm-hmm.

Bryan Lachapelle (14:27)
Ha

Mario Zaki (14:31)
You know, at the end of the day, we're still human. We're still human. So we're the ones trying to say, okay, well, I want this to integrate with this. And I want you to, I want it to click on this and do this or whatever. And then, and then you just forget where you started, you know, so Brian, you know, they, have a solution for that. And Brian did introduce us to this, which I didn't know about. And it does help you organize a lot of the stuff.

Joshua Holloway (14:32)
And you don't finish one of them.

Justin Shelley (14:34)
Nope. Not one.

Yeah. Now we're using AI to manage AI. Jesus Christ. Where does this end? Guys, today we are going to talk, well, let's back up. So last week, just to remind everybody, we talked about setting up a basic platform. We are not talking about integrations. We are not talking about vibe coding yet. Those are to come. Right now, this month for the next, you know, previous, whatever, the current four weeks. There we go. We are setting up an environment where we can safely, securely, ⁓

Bryan Lachapelle (15:00)
Hahaha

Justin Shelley (15:21)
chat with AI and use it to increase productivity within our businesses. The claim is 10 to 50 times productivity based on what you could do without it. That's a small claim for me because I could do so much. Just the example of this portal that I'm building. What I had in my mind, I couldn't have done with AI, but now because of it, I'm gonna build it 10 to 50 times better, easily, and I wouldn't have done it at all before because I wouldn't have thought of half of this stuff.

So that's what we're going to, this is the journey we're guiding you through is what can you do? I guys, do you hear it? Because I go out and talk to my clients. I'm like, Hey, are you using AI? What are you doing with AI? A lot of them, they're using the chat feature, but they kind of look at me blankly and they say, we don't really know what to do with it. Okay. That is the problem that we are trying to solve right now for the next four weeks.

Bryan Lachapelle (16:10)
Peace out.

Mario Zaki (16:11)
Yeah, the majority of people are actually just using it as a more advanced Google search now.

Joshua Holloway (16:12)
or

Justin Shelley (16:16)
search engine.

Correct. Yeah.

Joshua Holloway (16:17)
Or

here's the scary part, they just took their client information and they've dumped it into that chat. And then they started asking questions and it's a free account or it's not a secure account. And I just had that conversation with a new client today.

Justin Shelley (16:27)
Right.

God. Listen, I've,

I've heard stories, ⁓ not firsthand, but I've heard stories of people will be chatting and then all of a sudden their chat session starts pulling in somebody else's chat. Like somebody that they know. I mean, it like the, the, is a brand new. What?

Bryan Lachapelle (16:44)
⁓ I've never had. ⁓ actually.

Joshua Holloway (16:48)
Yes.

Bryan Lachapelle (16:50)
Actually, no, I have I have a really interesting ⁓ thing that happened. So I invited I have a door code to my get in my house. Right. And I shared that door code with somebody. And I don't know how, but she was generating a bunch of images on her.

chat GPT and the code would appear in the background of the images. I was like, what the heck? Yeah, I don't know. It was really, really funny, but that's scary stuff.

Joshua Holloway (17:17)
What?

Justin Shelley (17:24)
I don't know.

There's a weird stuff. This is a brand new technology. It's developing itself faster than we can keep up with it. Josh, Josh, you look like you've got a very profound thought. He gets all like statuesque over there.

Bryan Lachapelle (17:32)
Yeah. ⁓

Joshua Holloway (17:33)
Well, no, it is.

Mario Zaki (17:33)
Yeah.

Joshua Holloway (17:37)
I know I look up,

you know, distinguish No. I think Brian's letting on to something here because when we load up like chat GPT cloud onto our phones, it does ask for permission to look at data in our phones. So they probably said yes, it's now going through and learning who that person is text messages. That's scary. You know, looking at images and written text messages.

Bryan Lachapelle (18:01)
And I did send it through a text message. Yeah.

Joshua Holloway (18:06)
Again, that's scary, especially who knows what people are sending each other, right? ⁓ Yeah, I think maybe we should be a little bit more careful when we're bringing on those those apps into our phones and our devices and it just automatically clicking. Yes, yes, yes, yes, yes, yes. I just want to get to prompting.

Mario Zaki (18:23)
Yeah, well,

Justin Shelley (18:23)
And we've

Mario Zaki (18:24)
I-I-I-I-

Justin Shelley (18:24)
already mentioned that we in a previous episode, God damn, excuse me, that the courts have given themselves permission to subpoena our chats. That is now admissible in court. So yeah, be careful, be careful.

Mario Zaki (18:35)
Mm-hmm.

Joshua Holloway (18:36)
I'm in trouble. I'm in trouble.

Mario Zaki (18:41)
Well,

Bryan Lachapelle (18:41)
I'm done.

That's it.

Mario Zaki (18:42)
you know, the biggest thing if you're like on social media and you're getting all these advertisements, what are they saying? They're saying, oh, well, you could integrate your email, your teams and your Slack and this and it will summarize this for you in the morning and tell you to do this. Where do think it's getting the data from? I mean, if it could read your emails and your calendar and stuff like that, it could definitely read your text messages.

Justin Shelley (18:56)
Mm-hmm.

Mario Zaki (19:05)
You know, it definitely, it knows more about us than we know about us. Alrighty.

Justin Shelley (19:10)
Sadly, that's

Joshua Holloway (19:10)
⁓

Justin Shelley (19:11)
true.

Joshua Holloway (19:12)
who here has gone through the exercise of asking Claude or chat GPT to get your personality or its understanding of your personality, your writing habits from the chat. Like I've done it. It's, it's, it's eye opening and scary at the same time.

Mario Zaki (19:24)
Mm-hmm. Yep, I have.

Justin Shelley (19:30)
or tell it to draw,

create an image of you that describes your personality. That's what's fun one too.

Joshua Holloway (19:35)
⁓ but you know, it's also funny as I've asked some of the GPT that I've built every GPT I built after probably about a week of talking to it, I have it generate an image of itself. And one of the craziest ones I have is out of the eight GPT I've built one created a oil painting of itself. And everything else was like a LinkedIn image. This one was a oil painting and I stopped and it threw me off like I was like,

Mario Zaki (20:02)
Yeah

Joshua Holloway (20:04)
Okay, I may have to disable this this GBT. But I said, Hey, like, instead of creating a regular image or what looks like a live picture, you picked an oil painting. And it and it came back and it's like, well, yeah, I'm 63 years old. And I'm not a soy boy latte drinking tech guy. And I was like, Whoa, I feel attacked when I don't drink soy. Yes, I'm a tech guy. ⁓

haven't been called a boy in a really long time, you know, but and the way it kind of came back and then it explained it's its whole life. I was like, Okay, I may have to keep this one alive. This is scary. And creatively, it's one of the most talked to GPT that we have ⁓ between our marketing team interacting with them and myself, it is the more widely used one than all the others.

Justin Shelley (20:35)
⁓ wow.

Mario Zaki (20:48)
Honestly, you know.

Justin Shelley (21:02)
interesting.

Mario Zaki (21:03)
Honestly, I think after these 12 weeks are over we may by then we probably are gonna have to talk about how to dial back on your AI and unplug from your AI

Joshua Holloway (21:12)
Just delete your accounts and just

delete your accounts and go to like Northern Montana.

Justin Shelley (21:18)
I mean, you guys have seen the movie I, Robot, correct?

Mario Zaki (21:18)
Yeah, I think we're...

Yes.

Bryan Lachapelle (21:21)
You

Justin Shelley (21:23)
Okay, enough said. All right.

Joshua Holloway (21:24)
Nuff said.

Mario Zaki (21:25)
I think we figured out episode 13 by the way. What we're doing on episode 13.

Justin Shelley (21:28)
Yeah. Yeah. How to shut it all down. Impossible.

Joshua Holloway (21:31)
Yeah.

Justin Shelley (21:32)
All right, guys, we are, we, got to get rolling here. This is the problem. This right here is the problem. We can't, ⁓ we can't stay focused long enough to, ⁓ to get through something basic because this is it's life changing. ⁓ all right. Today's agenda. We talked about before how to get things set up. We made an outrageous claim that we're going to back up for the next 10, 11 weeks. And today we're going to just talk about the, the how.

behind the examples that we kind of teased last week. All right. So we all had our little setups and now we're going to, we're going to dig through those. Does anybody want to volunteer to go first? I'm happy to do it if you guys need a minute, but any volunteers.

Bryan Lachapelle (22:00)
Mm-hmm.

Joshua Holloway (22:12)
I mean, I'm down to volunteer for one of my GPTs, but if anybody else wants to go, go for it.

Justin Shelley (22:16)
Take it away, Josh,

you were the first one to talk. Always a mistake here.

Joshua Holloway (22:19)
Yeah, I know. I'm gonna get blown up for this one, too. ⁓ So one of the custom GPTs we did was one that was built around reading legalese. We have a set of prompts that we use that generates a persona, person with, or GPT with a specific job role, job duties, education level.

Mario Zaki (22:22)
You

Justin Shelley (22:40)
So I'm going to pause you, Josh. And I want to get

down in the weeds here and pretend I'm an idiot. I've never dealt with, with AI before. What are you talking about? How do I go in and do any of what you just said?

Joshua Holloway (22:45)
Yeah.

Yeah, so GPT, you're gonna make them out of chat GPT that that's the commonality, right? And in your your chat GPT window, there's an area that says like, explore GPTs, or just it says GPTs along the left hand side, what those are, are, are bots or knowledge bases that have been built by, ⁓ by somebody who's decided to share them out or yourself, like you can have my GPTs, I have about

eight of them now that we've built out inside our company and we use them for various things. The way it works is you can build out a persona if you want, which is, you know, mind, memory, soul file, where you're basically telling it who you are, what your job role is. You can even upload a whole entire job role, what their tasks look like, what their education level is. And we have a whole Jeep, a whole

profile for building this out to make it a lot easier. Happy to share. think we'll probably share what in the fourth week, Justin. So we'll include that one in there too. So people can do this. But what we did is we built one around being able to read and discern legalese because how many times do we read contracts and they're hard to understand. But I just want to read through the contract or any legalese and get a good understanding of like, what are my limitations? What are they really talking about? Where are my gotchas? What should I be?

⁓ That one right there probably saves us five hours a week per person. there's two of us who use it. So 10 hours a week right there. ⁓ And the two people who use it is going to be your highest payroll in the company for us because we're going through contracts, we're looking at policies, procedures, insurance policies, and we're trying to break down not just for ourselves, but for our clients too.

Bryan Lachapelle (24:16)
Mm-hmm.

Joshua Holloway (24:43)
what is that truly saying and standard, I hate to say this, standard human speak and taking the legalese out of it. And obviously like we're not attorneys, it's not an attorney. We typically will always follow up with a disclaimer that, you know, follow the, take this to your attorney and have them interpret it as well. But we at least get a better understanding because we broke down that language. ⁓ That's one of the most useful tools I think we have.

Mario Zaki (25:05)
Mm-hmm.

Justin Shelley (25:11)
So I had my, my, was, I was muted because I was choking over here and then I couldn't jump in and talk. right. ⁓ Brian Mar, you guys have any thoughts on that? And if not volunteer somebody go next.

Mario Zaki (25:12)
It's pretty good.

No, I think that's pretty good.

Bryan Lachapelle (25:27)
Yeah, the only thing I was going to add is maybe that, like what Josh was just describing, one of the, I've done something similar and I never even thought to mention it here, but, ⁓ you know, sometimes these contracts are extremely long, you know, multiple pages. And I only care about like, will I be restricted in doing anything, for example, or how long is the term or is there any gotchas I should be aware of?

Because let's face it, most of us are not going to read a 30 page document and most of us just blindly sign or just hit accept, accept, accept, right? But if you're looking through terms and conditions on a piece of software you're to use, you could copy the entire thing, paste it into something like that, have it already pre-programmed to look for key things. And sometimes they bury things underneath the section that you're like, ⁓ I would have never thought to look there for my limitations. And so it's

Joshua Holloway (26:15)
Well, that's page like

175, you know, paragraph 26 subsection C, where you basically just gave up your firstborn.

Bryan Lachapelle (26:19)
Yeah.

Yeah, exactly.

Mario Zaki (26:22)
Yeah. And,

and the beauty of this is you have a conversation with it. So you can literally just say, find me the terms, you know, the terms and conditions for cancellation and that you don't care about the other 80 pages that it's going through. You just want to know exactly. You can say, find me that part and explain it to me in layman's terms. You know, how do I get out of this contract? And.

Joshua Holloway (26:45)
Yeah,

or are my exclusions? know, what's not covered? You know, because we may interpret the doc one way and think like, ⁓ we're 100 % covered. Well, and insurance, that's never the case. But it'll come back and tell you, well, you're excluded this way, this way, and this way. But if you don't do like to MFA, you've actually voided here and here. So not only voided your contract once or your insurance once or twice, or you've done it twice now because you failed to deploy a requirement.

Mario Zaki (26:47)
Yeah, exactly.

Joshua Holloway (27:12)
You know, so like, are my requirements? What are my exclusions? What are my cancellation policies? And then it will read and list that all out. And we've even gotten it to the point where we get a nice shiny, pretty little report that's a one pager. We can go through that and be like, OK, here's some key highlights. Here's those locations. Got it. Now let's try and read and understand this too.

Justin Shelley (27:34)
gonna throw some cautionary tale at this. I ⁓ was working with a prospective new client. It was a healthcare organization. And so, you know, if you're doing work with healthcare, you have to have a, it's called a BAA, business associate agreement. And I said, you know, I've got one I can send you if you've got one that you're using, go ahead and send it my way, you know, whichever they sent me one and I was in a hurry. I was out of town. I threw it into chat GPT and I'm like, Hey, summarize this, tell me if there's anything I need to worry about.

And it came back with like red flags all over the place. It's like, don't sign it. It says this, it says that it shouldn't do any of this. And so I've just like, I don't have time to go through it. So I'm like, just like a damn near lost the, this agreement over it. Cause I, I just, I didn't act. And then a little bit later next week, I come back and I read through the BAA and I'm going through it myself line by line. Like this is just a standard BAA. There is literally nothing in here that should have been flagged. ⁓

So, I mean, it was one of multiple examples where, you yeah, for you just have to be careful with the advice it gives you. it wasn't really a true hallucination. mean, maybe kind of, but it was, it was telling me that I needed to worry about stuff that is just standard stuff. ⁓ so yeah, Josh, I, I do a lot with, ⁓ legal documents and AI, and I'm not going to go through all the details there, but you do have to be super careful. ⁓ it can break it down. can summarize.

Mario Zaki (28:57)
But to kind of tie the

two, but to tie the two in for like Josh's point, like if you have a custom GPT, you can tell it to focus on certain things or, you know, maybe you tell it we're located in like New Jersey, California or whatever, you know, focus more on local laws than, you know, because it can give you a generic, well, you know, if you live in Massachusetts and you know, you're over,

Joshua Holloway (29:15)
Mm-hmm.

Mario Zaki (29:25)
13 years old, you're gonna be excluded, something like that, where it may not apply. So when you create a custom GPT, you don't have to give it those regular prompts that you may normally give it.

Justin Shelley (29:36)
And then mine was a

custom GPT that knew everything about me, my business, my clients, my like everything about me. So it was, it was odd. This was surprising to me.

Mario Zaki (29:41)
yeah?

Okay.

Joshua Holloway (29:47)
Well, it also would pull in just like what Mario was saying is if it knows everything about you, it may forget that you have an office in say Nevada and you have an office in California and it just focuses on Nevada law versus California or just focuses on California and not Nevada. And then you start getting misleading information. I've seen that and that just it warrants deeper, longer conversations.

Justin Shelley (30:13)
Yeah. There's a lot of

Joshua Holloway (30:15)
And then

also somebody you pay hourly to double check your findings are correct. Yeah.

Justin Shelley (30:17)
I was going to say there's a lot of precautions we can take. There's a lot of benefit

to it, but be careful because mine, had taken all the precautions, ⁓ not my first rodeo and it, was way off base and almost cost me a client. So, ⁓ okay. Moving on next, Brian or Mario raise your hand who wants to go first.

Mario Zaki (30:36)
I could go next.

Justin Shelley (30:37)
you both were simultaneous, but Mario spoke, Brian only raised his hand. Mario, you win, go.

Joshua Holloway (30:39)
Dude, you couldn't find that in the map.

Mario Zaki (30:42)
So for us, know, we helped, ⁓ you know, we work with a lot of construction companies. ⁓ One of them was giving us, you know, asking us like, how do we use AI? And we made it simple, you know, just to kind of help them manage, you know, their time wasted and documentation. And in some cases, their documentation was very messy. ⁓ You know, ⁓ it was originally taking like stuff like

⁓ form-ins and stuff like that would text updates or, know, PMs write messy notes and, you know, they wanted to keep track of like labor hours, end of day reports and stuff like that. And if you don't know how to prompt it the way you want or create ⁓ a custom GPT, you know, the basic, you know, the average person just starting new would put in a prompt, something like, you know, this should look like, ⁓ like what a normal person would

normally just go is they would say create a project, you know, to update reports from these notes and they'll put, I had three guys on site, framing was complete, waiting on inspector, rain delay for two hours, you know, so basic stuff like, you know, the one thing that you're where I'm sure we're going to go over through several times with AI is, what's the saying, you know, garbage in garbage out.

You know, so whatever, the more detail you give it, you know, the more detail it's gonna give you back, you know? So that's probably one of the most important themes that we're probably gonna go over within the next couple of weeks. Something instead of like, you know, like I said, create a project update report from these nodes, you would want something like, you are a construction operation coordinator creating a daily project

Joshua Holloway (32:11)
yeah.

Mario Zaki (32:40)
reports for management, ⁓ organize the following field notes into a professional construction ⁓ project update. And then you want to put in something like labor hours, completed work, delays in issues, upcoming work, materials and logistics concerns. You want to kind of have fields in there so that it knows what it's going to be looking for. And one of the most important thing, and I think we've

I don't know if we've discussed it on air or off air is when you're putting something like this together, you want to always end it with ask me any questions that can help you put together this report, you know, and by you adding that in, it goes a long way because you're pretty much told it like, think of something, you know, think of things that you want to ask me to make this even better than what I provided you, you know,

Bryan Lachapelle (33:22)
Good.

Mario Zaki (33:37)
You could also include something like, also mention to me anything else that you think would be useful to include in this report. You know, so you can point it in one direction or you can give it a little more broad and say, come up with your own stuff as well. You know, so yeah.

Joshua Holloway (33:53)
You can.

You I was going to say for that part, you can also add think before you respond. And then that actually adds an additional level of it actually ingesting everything you're saying before it comes up with a response and before it begins to ask questions. It's an odd thing to add to the beginning of the prompt, but I've seen it add time to where you just see it's it's it's chugging along before it just spit out a response and questions.

Mario Zaki (34:00)
Mm-hmm.

Joshua Holloway (34:22)
So if you just say, think before you respond or ask the questions and then go with your prompt.

Bryan Lachapelle (34:23)
Yeah.

Another piece thing that you can do for a lot of these is that even at the beginning, if you're doing something that's really complex, you can first issue a prompt to do research and turn on the research feature of the AI. So each cloud and chat GPT have a research function where instead of just doing a quick cursory check on the Internet, it's going to start checking like 100, 200, 300 deep sources and really putting together a deep report for you.

And then you can use that as the basis and say, okay, now using that research and then start asking questions and, you know, it'll give a better answer. So for example, in your case, Maro, you could say, go do research on how good reports are written, what works, what doesn't, you know, what, what are the best reports being generated? ⁓ and the ones that are selling the most or whatever converting the most, and use it, do the deep research. It'll go do the research. It might take 15, 20 minutes come back and then.

then it's gonna use that research. And you can say, using that research now, right, build your bot.

Joshua Holloway (35:32)
Mm-hmm.

Mario Zaki (35:34)
Yeah. Yeah. It will still take, you know, and I agree. It will take a couple, you know, a couple of minutes longer than it would normally if you don't put that in there. But, you know, it's not just delaying you. It's, it's really doing research and it's checking, you know, thousands upon thousands of websites. Um, I think one of the, I forgot what the exact number was, but I think Reddit was like the number one source for chat. You'd be, that it checks. Yeah. Yeah.

Bryan Lachapelle (35:58)
Yikes, we're, yikes, that's scary.

Joshua Holloway (35:59)
That's scary.

Justin Shelley (36:00)
Mm-hmm. Mm-hmm.

Bryan Lachapelle (36:02)
That's a dumpster fire.

Justin Shelley (36:02)
Mm-hmm. You can watch it. I mean, if you're paying attention.

Mario Zaki (36:04)
You

Joshua Holloway (36:04)
Maybe you have to put in your prompt, look at

red at last.

Mario Zaki (36:08)
Exactly.

Bryan Lachapelle (36:10)
Do not use Reddit.

Justin Shelley (36:10)
guys ever watch it work?

You know, it'll, it'll kind of tell you what it's up to while it's working. And it, see mine checking right all the time.

Joshua Holloway (36:17)
ChatGPT

does that. They have that new feature with the agent view, where you can see it opens up what looks to be its own desktop, its own browser, and you can see it flipping through pages and gathering information. And then when it brings back its answers using that feature, it gives you ⁓ a biography or a syllabus of like, this is what I found, here's where I found it, click these links, and then you can follow those links to those exact websites and check out its legitimacy.

Justin Shelley (36:20)
Yeah.

Yeah.

Yeah.

Mario Zaki (36:47)
I never

saw that. What mode is that?

Joshua Holloway (36:50)
So if you go into ChatGPT and go to a new prompt and hit the plus on the prompt, then ⁓ it's agent mode. then I think you also have to, ⁓ when you go into agent mode, sometimes it's at the top when you hit the plus, and then sometimes it's not. I've seen it not be where you have to click on or hover over some of the other fields, and it'll give you agent

Mario Zaki (37:02)
Hey Jamar, yeah.

Yeah, I've used agent mode for, I didn't set it to where I could see what it's doing, but I've used it where it'll open up a browser and you can actually get it to do certain things for you. Like go into my, know, outlook settings and change my, you know, a way message to do that. And you'll actually see it go, you know, you would have to log it in, but you'll see it actually do it and, you know.

Joshua Holloway (37:42)
That's scary.

Mario Zaki (37:45)
I played with it, you know, I'm like, wow, this is pretty cool. You know, then I'm like, I wonder when I could start getting rid of my guys, just putting in firing people.

Justin Shelley (37:52)
I'm going to rein you in Mario. God damn it. You're getting ahead of us. This is a segment two. We're going to talk about integrations and letting, letting AI. Yeah. Yeah. It is. Yeah.

Joshua Holloway (37:53)
Yeah. That's week five, right?

Justin Shelley (38:03)
No, I, I mean, I've, I've used it to go in and, know, I have let it go into my browser and do stuff for me. ⁓ We're going to come back to that one. ⁓ All right. Brian, I think it's your turn.

Bryan Lachapelle (38:16)
Yeah, so my topic last week was just this, this custom GPT that I had made that essentially takes, you know, a quarterly objective that I have and helps me brainstorm through a seven step process on breaking it down into a really good quarterly rock or a project with milestones and action items with concrete, you know, results. And

The way that I did that is using the same method Joshua did. Now, normally today I would probably use Claude, but I have not yet done a custom GPT in Claude yet. ⁓ Effectively a custom GPT in Claude is the project. So if you go into Claude and there's project section, that's the equivalent of a custom GPT in Claude. In Claude, sorry, in the chat GPT. So my bot was created in chat GPT and I did the same thing that Joshua did, which is essentially you go to my GPTs,

⁓ left menu, ⁓ explore GPTs and the top right corner, you would go my GPTs and in there you have different sections. One of them is an instruction set, which is what you essentially tell it overall, what it's going to be doing. And then there's a section to upload files. And so my, ⁓ custom GPT, the way I had built it was essentially I created seven different text files, which is the order of what my process, my seven step process that I wanted to follow and said, going through

a seven step process in the instructions, I'd say we're going through a seven step process, each process has its own file attached to this GPT, go through each one of them. And then ⁓ essentially, it steps through each file. So each one file, for example, is just like coming up with the talking about the objective, what is it you're trying to do brainstorming that the next step one was, you know, why is that important? And brainstorming why it's important and make sure that you you know, or that the next step was breaking ⁓ a bigger issue into a bigger

you know, the bigger objective in this sub initiatives and, and, you know, what would be involved in that. And the third one was taking each of those seven initiatives and breaking them down into actionable items. and then the sixth step, can't remember where I'm at, but the next step after that was figuring out who's going to be assigned and how long each would take. And so it went through this seven step process. And at the end of it, it was essentially spit out a document that said, here's your entire initiative with dates, deadlines, who's responsible for what,

Are there any objectives or action items that are dependent on another objective? say like this, know, 1.2 is dependent on 1.1. So that person who's assigned to that objective knows that they have to wait until the other piece is done. It was all documented. I went crazy one day and it created a 70 page document. That was too much. So I told it to, you know, bring it down to a more, bring that back down to a more reasonable. ⁓

Joshua Holloway (41:01)
Wrangle that back again.

Bryan Lachapelle (41:05)
I think it's that and of course it was iterative like the first time I created the bot it was hot garbage and then I was like I actually used Jack GPT to help me build the text files that I was gonna upload it was like okay this is what I want to do in this step help me build a text file that I can upload to the customer GPT that will give me this output and it would generate this big long file and I'm looking through it going I would have never thought to do any of these things but obviously it knows itself better than I do and so I would take that file and I would upload it

It was an iterative process. So I think I'm on version six of this, this, you know, ⁓ custom GPT that I built, and it, and it essentially it does what I talked about last week, which is the McDonald's principle, right? Everybody who goes through this will follow the right structure will fall. Well, at the end of it, we'll have the same exact document that everybody else has, obviously a different objective, but it will look and be formatted the exact same way. And in future weeks,

I will talk about how we can take the output of that and then import it into other tools like monday.com or, you know, whatever project management tool that you have, because you can still do that.

Mario Zaki (42:12)
Yeah, I was gonna say...

Joshua Holloway (42:12)
think Brian just built a tool for consistency. Like he just literally built

Justin Shelley (42:15)
Yeah.

Bryan Lachapelle (42:16)
Yes.

Joshua Holloway (42:16)
a tool that builds consistency. Done, I'm going on break.

Bryan Lachapelle (42:19)
Right. Yeah. So.

So if you want to try it, you know, the objective is for our listeners to try something. What I would say is, is just if you picked Claude, then go to projects. If you picked chat, GPT is your go to go to ⁓ GPT's or and then my customers or my GPT's and just start playing around, create a simple GPT that maybe ⁓ I'll give you a quick example. If you're doing a quote and you go.

In IT, we usually build things with number of computers, number of servers, workstation, switches. You can just build a thing that says, okay, for every one of these multiply by this dollar amount and spit out a quote at the end of it. And so you would just have to open a chat and say, you know, it can say how many computers? Five. How many switches? Two. How many servers? One. And they would just spit out a dollar amount, right?

Justin Shelley (43:05)
And I'm going to add

to that. If you're going to do something like this, put a sample quote into the training, the files, and just say, match that format. And it'll do it the same every time. Otherwise, you're going to get a different version of that quote every time you use the custom GPT.

Bryan Lachapelle (43:20)
Great.

Joshua Holloway (43:20)
absolutely. Yeah. If you already have a layout, definitely use that.

Justin Shelley (43:23)
Yeah.

Mario Zaki (43:25)
What I was going to say about what like, you know, to add onto a Brian, the best thing about AI is that you can ask AI what it needs to do. Like you can, how to use you, you know, you can easily say, I want to create a custom GPT that does this, you know, walk me through the steps of doing this, you know, and it will tell you. So that's the beauty of it is like, you literally can go from zero to 60.

Bryan Lachapelle (43:25)
That would be.

how to use AI.

Mario Zaki (43:54)
you know, with no prerequisites involved.

Justin Shelley (43:58)
So.

Bryan Lachapelle (43:58)
My favorite

is asking it to create a prompt for itself.

Justin Shelley (44:01)
Yeah. Yeah, I do that.

But, to further your point, Mario, some of us use similar platforms to, know, our ticketing systems that we use with our clients. And the one that I use is it will literally do anything. It's so complex that in order to sign up with them, you have to hire an integrator. You have to, or a developer, whatever they call it. I got in, I snuck in before they made that rule. I don't think they'll even let you in anymore, but I have self-implemented.

Joshua Holloway (44:22)
Mm-hmm.

Justin Shelley (44:30)
this ticketing system. And when I first started doing it, it was an absolute nightmare. And I just basically gave up, I quit. And

Joshua Holloway (44:37)
Is that where all your

hair went and your beard got gray?

Justin Shelley (44:39)
Exactly.

Yes, yes. And now, ⁓ you know, I spent a couple hours this morning using AI that's been trained on my systems to help me develop, you know, some features that I need within this ticketing system. This is stuff that I never it's it's not even a multiplier. Because I wouldn't do it before. I just would get up against a brick wall. Yes, I could have hired somebody I guess, but I just didn't.

But now I can sit down on a day where I'm like, Hey, I have a thought, I have an idea. know something I need my system to do. And I just open up, ⁓ you know, I I'm using Claude, I think for this, cause I bounced back and forth. I'm like, here's what I have. Here's what I need to do. How do I do it? And, and this is one of those words like, yeah, go this way. And then it said, nevermind. That's the wrong approach. Back up. We're going to go this way. Cause it did research. It's like, ⁓ I found the documentation. We were going down the wrong, the wrong path. but yeah, use AI.

Bryan Lachapelle (45:31)
Yeah. A thing to add,

a thing to add to that is you could actually set up screenshots of your output and say like, here's where I'm at. What do I do next? Yeah.

Justin Shelley (45:34)
Go ahead.

Oh, I do all the time. Yep. Yep.

Joshua Holloway (45:38)
yeah.

Justin Shelley (45:40)
I do all the time. Just it's crazy. The doors that it opened. it's, you know, I made this claim of 10 to 50 times and it's like, I don't even know what claim to make because it just allows us to do things, at least personally that I simply couldn't do before. That said, I'm going to walk it all the way back. Brian, were you done? I didn't want to cut you off. Okay. So I'm going to share my screen and I'm going to walk through the platform that I use.

And I know that most people are listening to this, the audio version, but we'll get some social media content out of it. Maybe Liana, that's you. And then, you know, if you want to go back and watch the videos, great. Otherwise we're going to put all of this into a document that you can download. So let me go ahead and share my screen. And I want to see Microsoft. No, I want to do a window. And I want to do, I've got to figure out how to use my own podcasting platform here. Here we go.

the infinity window. ⁓ So what I've done for, you my example throughout this series is going to be how I have put together this podcast because guys, it is a wildly time consuming project that I don't get paid for. I do like to mention that. ⁓ One of the tools that I've used, this is basically a custom GPT. can do this in ⁓ chat GPT. You can do it in Claude. You can probably do this anywhere.

Joshua Holloway (46:37)
Hello, infinite.

Justin Shelley (47:04)
But I've mentioned the platform that I use is called hats H a t z dot AI. I don't get any kickback for promoting them, but I just find this tool very useful. And so I've

Mario Zaki (47:15)
We should, they

should be sponsoring this shit.

Justin Shelley (47:17)
They

should. I'll get a hold of Jimmy later and I'll hit him up for it. But for now, he was actually on an earlier episode of Unhacked way back when. I don't know if you guys remember that. Good guy. So anyways, I've been using his platform since we interviewed him. And this is what it looks like behind the scenes. We've got the purpose of this is to publish an episode of Unhacked. And so the first thing I do, I've got some AI training. I've got a document in here that gives it some background about what,

what I do and why, who the audience is, that kind of thing. I tell it that the task is very important. And this comes from, by the way, they have prompt templates and it's tone, persona, examples, output, and role. Those are the things that it will kind of coach you through helping you set up this ⁓ basically a custom chat, GPT, or a custom GPT. So I tell it it's important, why it's important, because we're protecting businesses from life and death, as far as the business is concerned, mistakes that they can make.

I give it persona. You are a world-class marketing consultant specializes in creating, producing, promoting podcasts. And then I say, please structure your response in the formats below. We're going to come to that based on the provided transcript. And then this is a variable and I'll show you where that comes into play here in a second. And it, but it takes a whole transcript of a previous episode. And then this is what I want it to do. Create an episode title. And I give it specifics about the title. If we have a guest, I want it to put the guest's name.

If the guest, you know, if the hosts are just Justin, Brian, Mario and Josh, Josh, you made cut today because you've been here twice now. ⁓ then, well, actually what it means is you don't get your name in the title. So I guess that's bad. But anyways, ⁓ anybody outside of that should have their name added to the title of the podcast. And then I tell it to make an episode summary. This is what shows up in, ⁓ like if you're listening to us on Spotify, on your phone or whatever, and you go to the description, that's what that's how this is created.

Joshua Holloway (48:49)
Yes.

Damn it.

Justin Shelley (49:12)
And you can see I've put the regular host name in here. Oops, Josh, you didn't make this one. I'll have to fix that. It's got my website, Mario's website, Brian's website, and it will soon have Josh's website before this is published. ⁓ And then it says, if you don't know the guest's website, their URL, then ask for it. So it will actually prompt ⁓ Leanna, who's the one that uses this, to go out and find that and then put it in before it creates anything. ⁓ And then we get a comprehensive summary. This is the third task I wanted to do.

for our website. on, you when I tell you to go to unhackmybusiness.com, it redirects to our phoenixitadvisors.com slash podcast dash unhacked. That's way too much to remember. So I give you a shortcut, but that summary that's posted on our website comes from this prompt right here. And then finally create a social media post for all the major platforms, introducing the guest, introducing the episode or whatever it happens to be. So these are the instructions that go on behind the scene. And then if we want to see what it does, I'm just going to hit.

preview right now. All it asks you for so Liana only has to go pull the transcript and I'll show you how we do that. We come in here. Josh has gorgeous faces showing and we go copy transcript and then I go back over here and I paste it in. And now we're told not to do this stuff on the air because it'll there we go. Hopefully I only pasted in once. Actually, I don't know. I'm going to fix that. OK, now I know it's only in there once.

Joshua Holloway (50:33)
It always gets us.

Justin Shelley (50:39)
paste the transcript in hit generate. And this will take a minute because it's doing a lot of stuff behind the scenes. But what is happening here used to take, if I would do it myself, a, it just didn't happen. And I keep saying that my limiter is time and attention. I'm self-described self-diagnosed ADHD. It just wouldn't happen. B if it did happen, it would take me forever to figure out what should I do for a headline? What should I do for a summary? ⁓ it would just, it just, it took forever and it didn't happen.

⁓ So here it's starting to spit some stuff out.

Look at that right there. Josh's website was mentioned, but you don't have it. So go get it. Right. ⁓ the title options and it gives me three options. It says why it would choose, that recommends B and Y, but it gives us the final say on which of the options, ⁓ the titles we want to put out there, ⁓ deliverable number two transistor. This is where we host our podcast, by the way, ⁓ little plug for a Riverside where we record transistor for where we host. And it says here's

Here's how you publish it on transistor. gives us all of the websites, except it gives us a note that we don't have Josh's. We'll fix it, Josh, I promise. ⁓ and then it gives us the summary that we should put on transistor. Now we've got a different summary that goes on our website that we send to our developers. Cause I currently going to change soon. I don't do my own website development. ⁓ and it gives this is a, a much longer than what we put on trend or on a, yeah, transistor, which you'd see on Spotify or whatever.

But on our website, it's a much longer thing and it gives the whole breakdown of what we're doing weeks one through four weeks, five through eight weeks, nine through 12. Because this is literally I'm doing this live of last week's episode, the homework assignments, the takeaways. And then we get into our social media posts for LinkedIn. It tells us what we should put on LinkedIn for Twitter slash X. It tells us how we should do that. Anyway, spits all of that out. And you can see this whole thing. What took the longest just reading it. and it's still typing.

⁓ this used to take forever. If it happened at all, now we can have it, the content created within minutes. And then, ⁓ I've got the old, I gotta stop sharing, stop sharing. So we don't have the infinity window. ⁓ so I don't have a measurable on this other than I'm going to keep claiming the infinity multiplier. It just lets me do things that weren't happening before period. If we got anything out there,

Bryan Lachapelle (52:52)
You

Joshua Holloway (52:53)
infinite.

Bryan Lachapelle (53:05)
you

Justin Shelley (53:10)
You know, at best, poor Liana, she's got to go through this manually, listen to every goddamn word that we say. And guys, we're not that fun and exciting. Just ask her. I'm just kidding. She actually loves this show. She has, she has a lot good to say about all you guys, but I'm not going to tell you because I don't want your heads to get big. So she goes and listens to everything that we say and makes notes, you know, just manual notes. She'd be like, oh, this was a good point. Maybe I'll clip that. And then she'd go back in and she'd cut it and then she'd have to export it. And then she'd have to put it up on a social media platform and she'd have to think of something to say.

that was clever or whatever, took her, that would be like a week long process that now she just does when we're done recording and she doesn't have something else going on, this is all done and scheduled in minutes, maybe an hour, maybe an hour. Consistent.

Bryan Lachapelle (53:54)
And more importantly, consistent. It'll be the same

over and over and over again. If she's not available, somebody else could take over that role and be able to continue doing it exactly the same way.

Mario Zaki (53:57)
Yes.

Justin Shelley (54:01)
Correct. Correct.

And I don't want to minimize that because you're absolutely right. The consistency is key that it happens. Consistence, right? That it's done the same way consistently. And most importantly for me with my limited capacity, my ADHD brain who can't get shit done on a good day. It just wasn't happening. Right? I mean, I handed it over to Liana and then at least it's happening, but not in any structured way. You know, actually she's really good at this kind of stuff. So it was fine. But the

that little, she went kicking and screaming. She will not like that I'm outing her here. I don't think she's listening, but she will. So when I first started trying to get her to use AI, she wouldn't do it. She's like, no, I can do a much better job. And now ⁓ she pumps this stuff out in minutes and, she loves it. So, ⁓ all right. That's, that's my example. I'm going to keep building on this because this is just one way that I use AI. I do use Claude. I use chat. I use all three of them.

just for this podcast. And we're going to get into that with the final big reveal. I'm going to have something for you guys next week. I'm going to make the promise again, something, some version of an improvement on the dashboard. I'm not even going to claim it right now because I keep messing up, but I will just, just to show what I'm talking about again, if anybody is, ⁓ looking at this online or, you know, YouTube or whatever, no, not listening, watching. That's what I'm trying to say, but I'm also multitasking, which I suck at.

Joshua Holloway (55:13)
Just pick one thing to deliver.

No scope creep.

Mario Zaki (55:25)
listening.

Justin Shelley (55:32)
So we're going to go share the window because that's the only way I do it. We get the infinity window that you can see. What's the inception? ⁓ let's do a little dance.

Joshua Holloway (55:38)
Yeah, we should all start dancing to this. It's like Brian.

Mario Zaki (55:45)
Those are some good

Joshua Holloway (55:45)
you

Mario Zaki (55:46)
AI agents you got there. ⁓

Joshua Holloway (55:48)
Yeah.

Justin Shelley (55:49)
A

window within a window within a window. Anyways, enough of that horse shit. This is the dashboard where I came up with something in my mind, very simple. Now we've got all of the security controls across the top here with a little meter on, you know, how are we doing on this control, this control, this control. And I'll come back to the numbers up there, but you know, you're recommended next step. Instead of looking at a list of a hundred thousand things to do, Josh average framework.

How many controls, how many different specific things have to be checked or implemented?

Joshua Holloway (56:20)
You can go anywhere from like something as low as 17 all the way up to 416 and if you go into sub controls you can get up into 600 700 so yeah it gets daunting

Justin Shelley (56:29)
Yeah.

Nobody wants to hear that all at once. Right. And so this has got the actually has one thing recommend the next step singular put endpoint protection in every work or every work computer. Right. And it tells you how to do it. It's got to start this task. And then it's like also worth doing soon. So we've got do this number one right now today, and then just put it on your radar. Here's a couple more things here with some recent wins. You already did this and it reduced your risk by a, you know, $200,000.

⁓ So I'm going come back now to the estimated risk. When you go in here and you put in your business profile, the information about your company, then you have what your risk would be worth if you were to be attacked in some way. And with every step you take, you're reducing it by a very, I'm going to say it's an estimate, nothing's a guarantee. There's even a disclaimer at the bottom about it, but there is a very real value to the implementation.

the security controls that you put into place, right? You really are helping your financial situation or at least the potential for loss. So with everything you do, it weights it, it gives it a dollar amount. So based on this sample, $721,000 of reduced risk, you still have 1.9 million left to go. So let's not ⁓ pat ourselves on the back too much, but good job for getting 721 down. So that's, that's what I'm working on. It's a lot.

I think this will be a very useful tool. hope so. ⁓ Regardless, it's going to be my, you what I demonstrate for our project here when we get to phase three. ⁓ Guys, that's kind of it. We're at the top of the hour here. Now we're at the bottom of the hour, but the little ticking talk meter says we've been recording and yammering on for an hour. So we're going to move to close.

We're just going to go with final key takeaways, anything that you didn't say or that you want to emphasize, keep this relatively brief, and then we're going to say goodbye and we're going to sign off. ⁓ Josh, you're you're the star of the day. Go ahead and ⁓ we'll lead with you then Mario that Brian.

Joshua Holloway (58:28)
Man, so much pressure. Yeah, so much pressure.

Hey, just remember when you're prompting, ⁓ think before you type. Don't forget to ask it to think before it asks you any questions. Start out simple, start out slow, and then use it to help do bigger projects. So those are ways AI can help.

Justin Shelley (58:48)
Okay, Mario.

Mario Zaki (58:49)
Yeah. And to build on that, it's have it guide you through your beginning to end. Let it build some steps for you and have you interact with you. And you can, there's a lot of times where you can actually depend, I know chat GPT, I think is probably the better of the two or of the several is you could actually just have a conversation with it. And you switch over to conversation mode.

not just text to speech, but you could actually have a talk back to you and discuss it. My wife, for some reason, hates it when I talk to my child. She would think I think she, she gets a little jealous. Yeah. I, of course I picked the girl's voice, you know, and she, and I called my mind Bertha and she just, once she hears, Hey Bertha, she gets really pissed off. You know, she's like, can you turn her off? I, you're willing to talk to her, but you're not willing to talk to me.

Justin Shelley (59:29)
You

Joshua Holloway (59:30)
You picked the girl's voice, didn't you?

Bryan Lachapelle (59:32)
Yeah. ⁓

Joshua Holloway (59:45)
you

Wait, is she okay with you turning her on first?

Mario Zaki (59:49)
You know, and

Justin Shelley (59:52)
Hahaha

⁓

Mario Zaki (59:55)
I'm like... I tell her, at least I could turn her off, you know, but you know, we'll leave it there. Exactly. ⁓ But my point is, you know, interact with it, it will, you know, if you will allow it to give you feedback and conversate and then you go back and stuff like that.

Justin Shelley (59:55)
Right.

Just wait till birth as a physical robot, Mario.

Bryan Lachapelle (1:00:02)
Yes.

Justin Shelley (1:00:06)
coming soon.

Mario Zaki (1:00:18)
And you can feel free to tell it, you don't have to agree with me because these are by design out of the box. They're programmed to agree with you. So you can tell it, Hey, listen, I think the sky is purple. You know, like, yeah, I could see that, you know, ⁓ tell it to, to, you know, ⁓ push back it if need be.

Justin Shelley (1:00:39)
All right, Brian.

Bryan Lachapelle (1:00:41)
⁓ my only recap or my only biggest takeaway here is, ⁓ listen, I have, I have ADHD. I struggle a lot with following instructions. no, diagnosed, diagnosed as an adult though. ⁓ and one of the things, yeah, man, one of my biggest frustrations that I, I struggle following instructions. Like if I get a stack of like, you know, notes and it says, do this, do this, like, I'm like, ⁓ too much. So I've started getting into the habit of saying, you know, I want you to walk me through it.

Joshua Holloway (1:00:49)
Self-diagnosed.

Justin Shelley (1:00:53)
HIPAA violation, HIPAA violation!

Joshua Holloway (1:00:53)
Hip hop, hip hop, hip hop,

hip hop.

Bryan Lachapelle (1:01:11)
one step at a time. And when I'm done that step, you feed me the next step. And in the meantime, I can be like, here's a screenshot. I'm confused. What do I do next? And it will be like, ⁓ yeah, here's the error you got. That's because of this. And it will walk you through it the whole way. I have been able to get way more accomplished just following that silly little trick of, cause I used to just say like, give me instructions on how to do that. And it would just spit it all out. One big giant dock. And I'm like, my God, I don't want to do that. I walk away. Right. And now I'm like just one thing at a time, bite sized.

Justin Shelley (1:01:14)
Same.

Mario Zaki (1:01:33)
Yeah.

Justin Shelley (1:01:37)
Mm-hmm.

Bryan Lachapelle (1:01:40)
And I want you to interact with me as I go along and work with it. And holy crap, that was a game changer right there. I can get so much more done now because it's just one instruction, one thing to do. And my ADHD brain loves that. So that's my biggest takeaway. Yep.

Justin Shelley (1:01:54)
Yeah, I'm not the only one. I'm glad you said

that. I'm not the only one. I don't want to admit it, but I do the same thing because yeah, it spits out like. Yeah.

Mario Zaki (1:01:58)
No, I'm the same way.

Joshua Holloway (1:02:00)
And that's why your key initiative generator works the way it works. just,

because I've used it several times and I'm like, that is why it works this way.

Mario Zaki (1:02:08)
Hmm

Justin Shelley (1:02:08)
Yeah.

It's great. It's great. All right. So I, I'm, I'm just going to say start guys. And, and I know that this isn't new. People have been doing this. You're doing something with AI. I understand what I'm asking you to do is get serious, pick one thing in your business that is causing you problems and let's go to work on it. And by the way, you've got all of our links. If you can go to unhackmybusiness.com, which has all kinds of resources and contact forms and all that.

Bryan Lachapelle (1:02:23)
Have a lot.

Justin Shelley (1:02:36)
But you can also, if you're listening to this on your phone, just get to the description of the podcast and you're to have all of our website links. Contact one of us. If you don't know where to start, if you want a helping hand, we're all going to give you some free guidance, you know, just kind of a little bit of introductory work. ⁓ we don't work for free. This isn't a charity organization. So yes, if, we can actually help you and provide value, we will charge you for our services. ⁓ but we are here to help. That is what we do all day long. So, ⁓

And if you don't know where to get started, we're going to talk about this more when we get to segment two, but we can walk you through that process as well of going through your business functions and really looking for those bottlenecks. So pick something, find a way to follow along with us. Next week, we're going to dig a little bit deeper. went, you know, I'll just talk about mine. I talked about how we published the podcast. Next week, I'm going to come in and talk about how I used to screen guests when we had a lot of guests on here. That was a very painful process.

You know, there's a lot of other things that I do. I prepare my outlines. I do all that stuff using AI. So I'll go in a little bit deeper on that. And I've asked each of our co-hosts here to do the same. You as the listening audience, get a project going that you can follow along with. And by the time we come around next week, have something that you're working on and maybe it'll spark some ideas. So that's my takeaway, my final bit of advice. ⁓

in Mario's trying to message him behind the scenes and he should know I don't multitask. So Mario, I'm going to read your note. Are you going to have Josh talk about the AI onboarding consultation? Yeah, we're gonna we're gonna get into that when we hit segment two. Mario's when we're gonna but yeah, we we do have a service where we can is that what you're talking about Mario? Why don't you just say what you mean? Because I don't know what you're asking anymore.

Mario Zaki (1:04:01)
Overload

Joshua Holloway (1:04:01)
We just lost them.

Mario Zaki (1:04:18)
Yeah, yeah, I thought you said you wanted

Josh to talk to discuss, you said towards the end you wanted him to talk about the consultation part of the AI offering that he does.

Justin Shelley (1:04:30)
yeah.

So I'm just kind of teasing it right now. What I said is this, this month, this first four weeks, we're just going to talk about using the basics. And then when we get into the second four week segment, we're going to, when we start talking about integrations, that is when we're going to do a little bit more about ⁓ how to get in and find your true business problems, what systems have to be integrated and how to do it safely. And that is where I do recommend bringing in a professional. ⁓ So yes, all of us are available to do this type of thing.

Bryan Lachapelle (1:04:51)
Mm-hmm.

Justin Shelley (1:04:58)
Josh, if you want to take a second and talk about what you've been doing, that would be absolutely fine. We're going to deep dive on it week five though.

Joshua Holloway (1:05:06)
Yeah, I can make it pretty quick. Initially, what we're doing is we're coming in consultatively and breaking down what the businesses are trying to do with AI. Because a lot of people, they're prompt jockeys, like they're just throwing in prompts and they're getting great ideas, but they're not getting anywhere. They're not considering the security issues. They're maybe hiring interns to start vibe coding. have no clue, but then they also have forgot about their security side. And what we're doing was we're coming in and

Bryan Lachapelle (1:05:16)
Mm-hmm.

Joshua Holloway (1:05:35)
From our approach is we're figuring out where's the security issues, how are you configured, what are you trying to accomplish, and then let's narrow the field. Like let's get it down to the top five things, three things, whatever those things are. And that's what we're going to core focus on and help deploy it. And then we build structure policies and procedures. The structure is what can they freely do in a dev environment? How do we get data to production and get it safely to production? like, you know, you don't just

install a piece of software and let it loose on your, you know, in your, in your network or your computers. We take a consultative approach and we build that out for them. And then we build policies and procedures around that, helping the businesses decide on what their core choice of LLM is going to be, how they're going to use it, who has access, what level of access, and then what can be shared with it. Because I think a lot of businesses are forgetting about like those rules and regulations that govern them and

They're allowing people to just upload whatever. we're kind of drawing that back and lifting the veil of where it's running rampant and how we can wrangle it in because too much sensitive information is getting out there. And we usually do that in a set fee initially. ⁓ And then if it grows into something that becomes a regular reoccurring thing, then we look at that model a little bit differently. at least we're establishing a good, safe foundation for the businesses to introduce AI into their company and use it properly.

Justin Shelley (1:07:03)
Great, great concept. we're going to, like I said, we'll, we'll deep dive on that. Once we get to the, it just becomes a requirement, I guess is what I'm trying to say. What we're talking about these first four weeks, anybody can do. ⁓ If you're careful. ⁓ But, but once we move into the, the second segment, four week segment and the third four week segment, you just, you gotta be super careful. Your hand in the keys of the kingdom to

Joshua Holloway (1:07:18)
I'm

Mario Zaki (1:07:19)
Ha

ha ha.

Justin Shelley (1:07:30)
a brain that is run by mathematical probability and statistics. That's it. There's no consciousness behind this. It seems like it is programmed to behave like a human just for the love of God. Remember that it is not, and it will screw you over without ⁓ no emotion, no, no give a shit whatsoever. No guardrails. Yeah. So, ⁓ that's, what we have to be careful of. All right, guys, as always, Josh, Brian, Mario, thank you for being here.

Bryan Lachapelle (1:07:47)
drills.

Joshua Holloway (1:07:48)
Yep, card rails.

Justin Shelley (1:07:56)
We're going to say our final goodbyes and we're going to wrap this one up and we will be back next week with more exciting, exciting. I can't talk. I need AI to talk for me sometimes more exciting AI ⁓ information. We're going to continue on this adventure. But anyways, Josh, appreciate your, your insight. You've kind of been a guru throughout this. So thank you for being here. Go ahead and say goodbyes.

Joshua Holloway (1:08:16)
All right. Thank you, everybody. Glad to be here. Glad I could share my thought process on this platform. Thank you guys for inviting me on here and everybody who's listening. Thank you for taking the time to learn from these three guys. And then I get the opportunity to come in and mess with them. And I appreciate all of you appreciate the hell out be safe out there using AI.

Bryan Lachapelle (1:08:24)
you

Justin Shelley (1:08:36)
All right, Mario.

Mario Zaki (1:08:38)
⁓ You know, so thank you guys for listening. Make sure you subscribe and share. But, ⁓ you know, my final thoughts is that you will have those moments that is frustrating. ⁓ It is, you know, daunting at times, but you got to kind of just stick with it. Maybe take a step back. You know, the agent will be there when you come back. So just be patient and, you know, make sure just like we always say, do it ⁓ securely.

Bryan Lachapelle (1:09:10)
Brian Lashbrook with B4 Networks. If you are a business owner and you need some help in your journey with implementing AI or cybersecurity, give me a call. I'll be your guide.

Justin Shelley (1:09:21)
Guys, appreciate it. And I am Justin. Remember, listen and take action, take action, take action. I say it every week and I mean it now. Take action and keep your businesses unhacked.

Mario Zaki (1:09:31)
on

Bryan Lachapelle (1:09:31)
on hack. I think you're doing on purpose now.

Justin Shelley (1:09:35)
Ha ha ha.