Automatic captions from the recording, lightly corrected. Stretches with personal details, unsourced numbers, dated news or advice the class has since changed are left out. That's 25 stretches in all. If the recording and the steps above disagree, follow the steps.
Open the full transcript, 12 chapters
0:53 Why build an AI operating system
going to work today is I'm going to talk to you about why an AI operating system is going to be probably one of the most useful things you can build.
Okay? And so we're going to build the operating system. And the way an operating system works is it's just like Windows. It's just like your Mac. It's just like kind of that where it's like a broad thing and all of the little things that operate your life and business fall within that broad thing. Right? So like most people when they have a Windows computer, they download Windows apps and that's what they use, right? They use Windows to do the thing that uh to basically be the thing that opens the app, reads the app, interprets the app, does all the things with the app.
So whether that's, you know, QuickBooks or whether that's calculator, it's all happens in the operating system, right? And so the content creation piece is going to be just like that where it happens within your operating system. But to get there, we have to build kind of the dome of the operating system first, right? And so what we want to do
2:33 Too many inputs, too many decisions
the reason that I think an AI operating system is so valuable is that in modern society, we have so many inputs that are trying to weasel their way into our lives for attention, right? So, we have email, we have um website things that people send us, we have social media that people send us, we have Slack messages, like there's all kinds of things that come into our ecosystem that we have to figure out what to do with. And every time something comes into the ecosystem, that's another decision that you have to make, right? And the more it's been make, right? And so the point of the kind of AI or agent operating system is to remove some of those decisions so that 50 small decisions get consolidated into one big decision. Right? So, if I
have to um if I have to send someone a website, like a landing page, um I can choose I can sit down and I can say, "Okay, I want the colors to be this. I want it to look like this. I want it to look like this. I want it to do this." And like give it all of those individual specs that I want that landing page to have. Or with an AI operating system, I can say, "Hey, I need a landing page for this client that serves this outcome."
And then the AI goes and gets the context. It gets the details and then that is what pushes it. Right? So the decision that I made went from the colors, the style, the the text, the all of those to this is what I need to do. I'm delegating an AI to go do it, right? Like we're in a lot more of like the delegation phase of kind of work at this point. And the the thing that is really helpful with the AI system is if you it's kind of like we talked about like if you have a good decision-making structure that allows you to get to AI to help you with your action, right?
It's the same thing here. If you have one capture point, so one place where everything is going in, everything, all of the stuff from your emails, all of the stuff, it looks like it's random, completely disconnected, and not um functional.
Hello.
The train is coming. [laughter]
Welcome.
Good morning. Sorry.
Oh, no worries. So, we're just talking about um AI or agent operating systems and how they can kind of make things super easy. So, the thing that you want out of your system is you want the decision that you have to make to be the same first decision every time, right? So, if I'm like, oh, I need to find something on my computer.
5:40 One inbox and a filing log
Before AI, I would be like, I need to find something on my computer. Okay, so I need to go to my downloads and then okay, it's not in my downloads. What did I do with it organizationally? Okay, it might be over here. It might be and you go and you search your entire computer and end up finding it 20 minutes later or realizing you deleted it or saved it to the cloud or it's just not where you thought it would be, right? And so the goal of having one capture point is everything that comes into your ecosystem goes into that folder. AI is the thing that manages that folder and it creates a log of everything it's doing. So every decision that the AI makes, it's logging where that thing is going. So if you have a client
disclosure come in and the AI is like, okay, I'm evaluating this. I'm going to move this copy to this person's CRM. I'm going to, you know, file it here. Da da. And then it logs that behavior. It logs what it did. So whether that thing filed it in the right place or not, you'll be able to tell. based on the log because so like let's say Jim Smith is the customer and this one accidentally filed it with Jane Smith because Jim has a wife named Jane whatever, right? And it filed it that way. In the log, you will be able to be like, "Hey, Jim Smith's thing isn't where I thought it would be.
Where is it?" It'll go to the log and it'll say, "Okay, I filed that on this date. I filed it here." And you'll be like, "Oh, okay. You filed that incorrectly. Jim has a wife named Jane. Add that to the context so it doesn't happen again." Right? And so the decisions all happen from the same starting place. They're all like one action, one action, right? And then a
7:16 Each business keeps its own context
lot of people here have different stuff going on, right? Like I have AI, I've got the mobile home parks, I've got personal stuff, I've got all of it, right? And so each one of those things is an entity. So like I view my personal life as an entity within AI, right? And so it's all kind of siloed as if it's businesses. Um, and so each entity has its own context. And when each entity has its own context within your system, you don't have to regenerate it every time, right? It's just there. And so it knows the background stuff of your business that you don't have to tell it this is my uh MLS number, this is my, you know, it has it. Um,
15:53 Start with low-liability work
people is on your business start with the lowest complexity lowest liability things that you can to learn AI, right? Because if it's simple and not going to get you in legal trouble, you can play around with it and it frees you up mentally to be like, "Oh, I can do whatever I want." Whereas you once you start to get into like, "Okay, this is like this is going to impact my business in a meaningful way." Like, I'm rebuilding my onboarding paperwork. I'm re, you know, doing X, Y, and Z. That's where it's like, okay, I have more time invested into the actual process versus like, build me a landing page. you can type that in, you know, go have, you know, whatever you want. Go watch a show on the television, come back, and the
landing page will be done. And it's like, all right, it's not that big of a deal if it's good or bad. You can make the adjustments super simply. Um, super fast. Um whereas if I was like oh all my context is locked into Claude all my workflows everything is here it takes like some amount of effort to migrate.
17:27 Point any AI at your own files
But if you have all of that stuff built on your system you can just point the intelligence at it right and so what a lot of people are
and what does that mean point the intelligence at it? You can say chat I've got the stuff in claude. Can you go with claude? You can say
or do you need to save it somewhere on your computer?
Yeah. So if you save it into like this is my agent OS folder, right? My agent operating system or whatever it is. If you save things there, you can point whatever intelligence you want at it. So if OpenAI has the best model or let's say Claude comes out with something that's really really good for like doing bookkeeping or a very specific purpose, you can point that like you can point ChatGPT and Claude at it and say Claude, you do the bookkeeping on this. ChatGPT you do the marketing right operating system and we're like in Claude talking about stuff and it's really good real estate stuff then do we need to save that conversation in a folder on our computer for that for future reference
yeah so what you
just can you click a lot of things and save them at the same time or does it have to individually be done
you can just say hey Claude I would I'm really enjoying this conversation I think there's a lot of value in this uh take the best and most useful parts of this conversation and add it to the context in my agent OS and it will go do that for you.
Right. And that's the value in the agent OS is now you're not going and doing all those little things yourself. You're telling one thing everything.
Sure.
And that's what's getting it done right. And so your decision-m loop is so much less cuz you're like what do I need to do with this? I need to go to AI and then I need to ask X Y and Z and it will go get me information. And now that very point. And then you go all the information comes back to you and now I have all the information and I can just make the decision instead of I got to go to this system and pull this. I have to go to this system and pull this. I have to make sure this is, you know, work and everything. And so it just it allows it to function in a much more seamless way.
So it has access to your computer where you say I have a folder on my um desktop and you can put it in
like it could tell it.
Okay.
Yeah.
Dragging it over.
So, yep.
Then with our OS, it's essentially folders with context cards
basically. Yeah. Yeah. It would be like Yeah, it would be it look just like your like downloads folder and however you organize things in there or however you organize things in Google Drive. Yeah.
20:01 Update the context you already have
So I guess then my next thing is what I really need to do is and how do I use AI is I I think that like the context cards that I have used I really need to update
because it's was weeks and weeks ago that we did those. I don't remember what's even in them. Yep.
So like how do I have it
based on what it knows
create those or update those. So honestly what I would do is the way I build my system is like very custom to me and I'm would recommend doing the same. And so like what I would do is I would say hey I'm building an agent operating system. The first thing I need to do is get my file organization uh on point and I also need to have one input folder where I put all my stuff and the AI agent manages it. Um, and I would say help me design that. And I'm going to have a prompt that I'll send out to you guys as well. Um, that'll like
I've already missed it because I don't I think I missed all the context cards part of the
AI classes and I just need to
do that.
So yeah. And so what you'll do is like I have templates that I'll send and all you do is you upload those templates into the AI and you say make these personal to me and it'll and so what you'll need to do then after that is you'll have to go read it and you'll be like this isn't relevant. I need to get rid of this. I need to change this. This was updated two years ago. Blah blah
Yeah. So, it's like personal like they have it's like an about me one, your brand. Yep. [clears throat]
It's like and then it's like the guard rails. It's all these things about how it like does you.
Yeah. Exactly. Exactly. And and then like as you're building out your system, you'll see the AI will like you'll like go into a folder and you'll be like, "Oh, what's this?" And then you open it and it's like the AI has added a memory to that folder. Yeah.
Yeah. Exactly. And it's like and that's the value in the operating system. Um and so like having just one point where that's you you go there and you know what you're doing
is like super duper useful because again that decision fatigue is super real. And again what it allows you to do is it allows you to point any intelligence you want at it. So at a certain point um I
So, but yeah, the goal is to have a system that you can point any intelligence at and it's going to get you the outcome that you want because the better your system, the lower the intelligence you can use. There are elite engineering talent people right now that can get better results out of the first generation of ChatGPT than I can get out of this one, right? because they know so much more than I do about how all of this functions.
26:10 A sample folder structure
So, and this is kind of goes back to the decision fatigue, right? You have to explain who you are again. You have to remember which version of the thing that you created actually matters. You have to decide where all the new things go and then you have to restart everything in the next tool, right? And that's why having desktop agents and your own system is highly useful. And so, this is kind of a sample of a system. Um, you can see it's got inbox, so that's where the new things go. It's got a me folder, who I am, how I work, like my personal context, and then it's got
you're going to have this for us, right?
Yeah. Yeah. Yeah. Yep. Yep. And then it's got work, content. So content could be work, content could be personal, but and so this is just this is a sample of how to chop it up, right? So this is not the end all beall. Um, you can divide this in whatever way makes sense to you. And that's what the cool thing about AI is, right? Like I historically would have to go and install my system in other people's business and then they'd be like, "Oh, I'm having problems with your system." It's like, "Well, of course you are. It's my system." Like I will come and like have to x, y, and z and get them up to speed. But if I can install the fundamentals of my system in a way that everybody is personal to them
and they understand like I like to start my file naming with the date first. I like to start my file naming with the entity first. Like there's all these little decisions that go into building something that is so personal to you. And the nice thing about doing that is one it gives you the ability to see everything but two it also makes it a little bit more difficult for your competitors. like if they see like your computer like I don't know you know what I mean like the the naming is it's not it it's very different it gives you a little bit of a competitive edge not a big one I would say um but a very small one and just being organized so that everything is at the touch of your fingertips where you say AI do this for
me and AI knows where it's going it knows what it's doing and it's actually getting the stuff instead of being like oh I need to research everything that this person wants like that's where you're no longer starting from 0% and that's like the biggest kind of head start, right? And so you just have to the way what I would tell you guys to do is have AI interview you and say I need I'm building an AI operating system. Help me figure out what is going to work best for me and then you can give it ideas, you can um you know operate it how you want.
Um, as an example for me, I like to base my operating system off of how a computer works because I that's just the most simple for me. Meaning, a computer has RAM, right, which is are things that it's doing right now. It's got storage, which is all the context for things that it may or may not be doing right now. And then it's got an actionable interface, right? And so it's, you know, what to do to take the actions. And so that's kind of where I'm at with everything is I like to set it up very similar to a computer because it makes sense to me that I have things that I need to get right now like today and then there's storage and then just kind of every all the processes happen in the back end. Um so I base I think
it's called the von Neumann for the computer model that I use but um so
29:30 Map the homes you already have
this [clears throat] is again where we are aggregating everything right so you can see here's the AI map of all the stuff that you're putting into your your entry point for the system right and we have computer Google drive iCloud I use all three of those storage systems and they are all a mess right now that and this is the operating system that I'm kind of installing in. So my computer's good drive working on it and iCloud is an absolute mess. So all of these things are going to be functionally pushed into the operating
But and that's exact. So, it's digital needle in a hay stack. That's kind of what it is, right? And that's what we're working on is we're working to find those values in a bunch of swamp sludge, right? And that's kind of like Yeah. So parsing is is super useful, but it all goes into like that AI maintain map. Every time something goes into your capture, that map gets updated. So you always know when things go into your capture where they went to at the end of it. And that's like my biggest problem is things go into my downloads folder and then they either stay in my downloads folder and it becomes a massive mess or they go places and I'm like, I know I have that. I don't know where I have it, but I know I have it.
Right? And so um so this is um kind of one of those things that is super it's more useful when you actually have the um system built out. But you basically say help me turn this idea into useful content and then without the system you're going to need to know which project, what source, what sounds like you like all the things that you just start from generic over again. Now, I will say one of the things
35:27 Get the setup right, then delegate
Right. And and that's the thing is like every company wants to be an AI company, but very few companies have the data in a set up in a way that allows them to actually benefit from AI in a way that allows enterprises to benefit from AI, right? And so if you get that setup piece right, it makes everything else so much easier. And like you guys are effectively each independent businesses, right? And so like just understanding how to move quickly because the AI is kind of eliminating a bunch of organizational friction is like one of the most useful things I think possible.
Um, and that's exactly um, the reason that I kind of revamped the class is I woke up and I was like, I'm pretty sure I forgot that at one point I was learning all this stuff, too. And now I'm teaching it as if they're farther ahead in the game than they are. And so just go back and just rebuild the the railroad. Yeah. Yeah. And so if we can build like a railroad track and then all you guys have to do is change the cars that are running on it, that's where you have like really useful business infrastructure, right? And so this is again kind of going back to like the individual infrastructure of your life, right? Like I have my personal, I have my business and I have uh the a project, right? So again, the way I
divide things may not be the way you divide things. Like for me, I have my top level outcomes, which are basically my top level outcomes are protect your finite resources, which is time, focus, energy, and tangible assets, right? You have to protect the things that you can't get more of, right? And so all of my projects go into those goals. Like I want to have as much of those resources as I possibly can. So all of the projects I build go into those goals. All of the tasks I build go into projects and then all of the resources help that manage
projects. What did you say?
So it's pro the way it goes for me is outcomes to projects to tasks. And so each one gets a little more refined. And then if you combine resources to that and you look at it, you say, "Oh, I have everything that I need here to do this project." Right? And so at that point, you point the AI at the project. You don't do the project. You have all of the materials needed to get an outcome that you're excited about. Then send the AI to go get that outcome, right? So get your stuff in line, then go get outcomes. And if we can do that, it makes the entire AI process so much simpler. So much simpler.
47:08 A question on posting everywhere
What I think would be ideal is you have a post of something that you want to say and and it goes and sends it on your Instagram, it sends it on your Tik Tok, it sends it on your Facebook where you want to send it. Can it do that in all the platforms?
That's exactly what I do with this class. So what I the only thing I do with this class is I build a video editing thing. Um, and then I upload this video to YouTube. I take that YouTube link and give it to my video editor. It chops all the clips. So, it looks for things I say that might be attentiongrabbing or like a good hook or whatever. And it it does all that, which is great because that means that I don't have to be searching for those things in my teaching. I can just teach how I want to teach and the AI is going to find like I don't have to change my personal style to be more like salesy or markety.
Um, and so the AI does that and then it goes through, it creates the thing, it does the captions and then it will post for me. It posts on uh my AI acceleration Instagram, my buildings Instagram, my personal LinkedIn, my AI acceleration LinkedIn, my Tik Tok. Um, it's going to start posting to my real estate Tik Tok. It's going it posts to uh two Facebook profiles that I have link uh said LinkedIn uh X. It posts on X. Um, and if I ever decide to get a Blue Sky account, I would post there, too. Um, but it it does all of it. And
49:37 A folder for each listing
So, I have only 10 minutes. So, what I want to do is what do I need to prepare for for next week for because I have two listings in California that I want to do what you and then I have one coming up that I'm going to be listing in a week. So, um what do I need to create to make this easy for us next week? So, what I would do is I would uh put a folder for each property or each deal
and then I would just put all of the resources relating to that deal in there. Whether it's a PDF, whether it's a note from the seller that you want to include, pictures, videos, um old listings, uh tax stuff, like whatever it is that context that you want in there, and it will go through. And one of the like, there's going to be all of these little like micro opportunities for people who are set up for the AI to do what it's supposed to do. Whereas, a lot of people are going to miss out because they have no context. They're just hoping every time they go to the AI, they type in the words and they're hoping that it gets a result for them.
Whereas, we're going to have intelligence that is data feeding the intelligence and then it goes and gets the result. And so, it's a different
Right. Yeah. And that's the thing is if you can have the intelligence literally put in front of you, hey, there is an actual reason to talk to these people. And it's not, hey, just call in to check in. Hey, just call in to take two minutes of your time, right? like, "Hey, I'm calling because I saw this thing happen that may benefit you positively or like
