Real Investor Radio Podcast

Craig (00:02)
Hey, welcome back to the Real Investor Radio Brief. I'm here again with Jack BeVier. Jack, today, today, it's all about OpenClaw, MoltBot, ⁓ ClaudeBot, whatever the hell you want to call it these days. Jack, my gosh, the world is literally on fire talking about OpenClaw. That's what I'm going to call it from here on out. And I started playing with it last week when we did our

podcast and I was trying desperately to get it set up and I was having some issues which we'll talk about in a second Jack, but I can't believe ⁓ how many people are talking about this online, how many videos have been created off of it, how many influencers are just blowing up because of it. One in particular, going to get them on the show, Alex Finn. know we haven't booked him yet Jack, but we're big enough that he'll come on.

But man, I just thought it would be cool to just do a brief episode today, Jack, to tell you sort of where I am with it in the past few days, really just starting today. And then just sort of talk to folks about, I developed some notes that we can put into the show notes in the description. It's just a brief PDF where I've downloaded some of the top Power users that are online.

folks on Twitter that are talking about it, all these crazy use cases that people are talking about. Jack can't wait to talk about those and then sort of resources for folks to click on and learn more. So I'm ready to jump in and talk about it as much as you want.

Jack BeVier (01:38)
Okay.

Okay. Usually it's you slowing me down, but let me, let me put a leash on you real quick. Explain to me, because I've seen, I've seen some of the, I've seen some of the social media hype around it, but explain to me exactly what it's now referred to as open claw. was Claude bot and then Claude anthropic took exception to that because it was getting too popular. so it has changed its name a little bit, but

My understanding is that it is a, an agent or a program that has agents built within it. And you can hook up those agents to all you give it access to everything, right? You give it access to your email, to your computer. I got some questions about that. ⁓ You can give it resources. You could give it access to a Bitcoin wallet so that it can go actually like have money as a resource and like interact with the world.

And the thing that is so cool about it, this is my understanding and you, you know, I'm looking for your feedback here, um, is that you've, is that they've taken the security off of it, right? Is that this thing is running free in the wild and it's got resources and you give it a mission. tell it what you want it to do and it's got access to a lot, a bunch of whatever resources you've given it. And then it just goes and does stuff.

And sometimes that's great. And sometimes that can have some unintended consequences, which is why it's getting, you know, so much, or which is another reason it's getting so much attention. How much of that was right? What did I get wrong? What are my misconceptions there?

Craig (03:14)
Yes.

a lot of it was right. So open claw is basically just an, an open source AI assistant that some dude in his spare time wrote because he didn't like that when he was away, he couldn't just talk to his server at home that was running his entire business. And then, and then another way of putting it is, ⁓

Open Claw is sort of like everything that was promised to us in 2025. We were going to have all these, you know, agentic employees that just basically worked around the clock and, know, all you have to do is tell OpenAI to do it for you and it's done. That never happened. We like it, like it was, was very cl- like I got, I built a lot, we built a lot of cool with Cursor and, and, and OpenAI or Clawed. We built a lot of really cool stuff.

but there's guardrails around all of it. You can only give so much access. Like you can only open up so much security, right? And so the whole idea of OpenClaw is no guardrails, whatever. And I mean, whatever, whatever you can do on your computer, you can have OpenClaw do. Yeah.

Jack BeVier (04:44)
So I have a question about that. Cause I was under

the impression that, that AI agents hadn't yet really gotten good at manipulating websites, right? If you said, Hey, go pay my water bill. That it was still kind of struggling with that.

Craig (04:58)
Yeah, I can give my agent Rocky, I call it Rocky Jack, because I'm rising from the ashes, baby. I'm going to have that theme song playing and everything. ⁓ I can give it access to my web browser and it can open up the browser and it can go to BG &E and it can put in my username and password and it can pay my bill for me.

Jack BeVier (05:22)
Can it click that little I'm not a robot, even though this one is?

Craig (05:27)
Not only can it do that, but it can also figure out capture if there's a capture phrase. I've seen it do that. I haven't personally programmed that yet. I'll get to where I am with it. ⁓ But yes, I've seen enough use cases at this point because I've watched a bazillion videos and read a whole bunch of ex posts that there will be nothing like I, I, I, this guy, Alex Finn, he set up his, his, his open button. So basically what,

Jack BeVier (05:32)
Really?

Craig (05:56)
there's a setup process. have to install it on your machine or you have to install it, install it on some remote server. like a Amazon web services or like a hosting or something like that, you can install this thing on a remote server or you can just install it on, I bought a Mac mini. I haven't run it on my computer thought that was probably like, I'm not a programmer. I don't know how to harden this thing so that, know, it's enterprise level security. And I think that's, that's coming Jack that,

Jack BeVier (06:13)
because you don't want to put it on your computer, right?

Craig (06:26)
That thing that like, I believe that you'll say, holy every person at Dominion could have one of these things because now it's enterprise level and we can open up Outlook and we can open up Salesforce and we can open up messaging and we can open up teams and we can open up like all of those things, Jack. I truly believe that that's what's going to feed these dashboards that I talk about that just give us that snapshot of everything we need at the moment we need it. Like if this, like I've told this thing,

What I need to do on a daily basis to be a successful loan officer at Dominion in no uncertain terms. I mean, I did a brain dump. That's what they recommend. You just do a brain dump. Personal life, work life, what your goals are, what you love to do on the outside. You just brain dump this thing. And the cool thing about it, Jack, is that it has these memory files. It memorizes the things that you tell it in both a

⁓ a user file, a user memory file, and then these daily logs that it makes that it can reference back to, hey, yesterday we were talking about this, or, I remember when Craig was talking about this, I'll go back to that memory file. So it's very good at that. That's one of the most important things I think about it. And I actually have a project that I'm working on right now that I think I stood up today where, imagine if, imagine if,

It didn't memorize everything like an MD file, like sort of an overview, but it literally memorized every idea that I ever gave it. Every website, every spoken idea, every project idea, everything that I want to do, like from, you know, what I want to build with AI. And it kind of then created categories and dependencies between all of those things so that the entire thing became like another, like over time in this super base in this, in this database,

It's literally becoming sort of the craziness in my head, but much smarter. I think I just stood that up today in about 15 minutes.

Jack BeVier (08:24)
Yeah, like

an external brain that that that has that can interact with the world though. ⁓

Craig (08:33)
And is probably way smarter than than me not you Jack, but definitely smarter than me, right? So so to answer your question Jack Anything that can be done on a computer Can be done with open claw if the right API's if you give the right skills if you give it the you know sort of You know that over time so I was getting back to this guy Alex Finn So he basically told his his his bot Henry. He calls it

Jack BeVier (08:37)
⁓ Man.

Craig (09:01)
Hey, while I'm sleeping every night, want to wake up in the morning surprised about something that you built overnight. Just make it happen. I don't care what it is, make it happen. Dude, one day he woke up and the thing made a face for itself. And it's on his screen and it comes up and now it's got a face, it's got a persona, it's got an image, it's animated.

Jack BeVier (09:14)
That's cool.

That's crazy.

Craig (09:28)
A few days later,

he woke up and it called him. It went out to the internet, made itself a voice over IP phone number, and then called him in the morning.

Jack BeVier (09:39)
Dude, that's kind of creepy though too, right? Like, like, cause these things are out there just running around. It's like, you got like this child that's like, just you've, that you've empowered with resources and just like, you know, wakes you up at six in the morning, like wake, you know, you wake up in the morning and your kids just standing there staring at you. ⁓

Craig (09:58)
I

swear to you that will be built tonight for me. I want to get a call at 445 in the morning from like somebody that sounds like David Goggins telling me to get the out of bed and go to the gym. How could with 11 labs? I've already given my 11 labs API. All I need to do is just feed it my Goggins persona and create like a sub agent for personas. And I'll say use the Goggins persona to make a

Jack BeVier (10:01)
as

There you go.

Craig (10:22)
make a cron job at 445 in the morning to call me and don't stop calling me until I pick up the phone.

Jack BeVier (10:28)
You

man.

Craig (10:32)
Here's the better. So here's the cooler part. So let's say part of what I want to do to get to create 2.0 Jack is I want to create streaks in the things that matter in those KPIs of life that matter, right? And let's say that one of them is going to the gym five times a week. And I could easily, I've already created a dashboard where I can click that. Like it's Monday, I went to the gym, click, right? But what if I gave,

access to to Rocky that he knows where my phone is at any given time and I've already got the Google Location Services API in there so it knows I'm at the gym and it automatically says hey you want that stay number one dude let's do it again tomorrow I don't have to click anything it just knows automatically

Jack BeVier (11:21)
Dude, that is wild. You're blowing my mind here. like, like, I mean, it is what you said. It's like, it's, is literally a personal assistant ⁓ that can do whatever you've tasked it with. That's wild.

Craig (11:34)
It's so,

so Jack, one of the things that, ⁓ that you and I have learned really a lot about over the last several months, and especially other guys here in the company is vibe coding. How do we talk to our computers? How do we talk to AI and then something that they spit out is the outcome that we were hoping for. Right. And, and I think we've done a lot with that.

We but it's it's generally been a very iterative iterative process. Jackie give it a sort of like here's my here's my product brief. Here's what I want to do and it you know sometimes it'll one shot something small. But what if you had a bot Jack? What if you had this bot that you were like hey man the expectation is that this thing gets done by 7 o'clock in the morning and I don't expect that you'll have you're going to you're going to work on it until it's done and there's no roadblocks.

So you'll get any resource that you need from the internet. I've given you all the APIs and sort of logins that you need. There's no iteration here. You've got the UI that I want to develop. I've given it the product spec. And a lot ⁓ of these guys, Jack, are getting even better. They're reverse prompting. They're saying, hey, if I wanted to create this dashboard of life, where would I start with that? And how will you do that for me?

rather than saying, Hey, this is all the things that I want. Go do it for me. Right. So there's this idea of reverse prompting now where you've got this very, very smart agent Jack. It's one thing to have an LLM that's just smart. It's a whole nother thing to have an LLM. That's a, that's an amazing programmer that understands vibe code that has access to all of your resources and has access to memory that met that remembers you.

Jack BeVier (13:16)
I mean, so just like looking at the downside, because I don't know, that's my nature. I can't help it. I'm sorry. ⁓ But like, can you like give it access to if you give it access to your work accounts and say, hey, respond to these emails for me and like process this like, or, know, will it, you know, like, can you just like,

step it in there, ⁓ link it up in there and then like leave and go to Australia and see how many days it takes for like the boss to figure out that you haven't been in the country for the past three weeks.

Craig (13:52)
Yes, you can. However, I wouldn't recommend it ⁓ because we all know what AI slop looks like and we sort of all know what, you know, unfettered access, you know, where that could lead. So there's so two things there, Jack. ⁓ To answer your question directly, I think one use case here at Dominion for my job would be like this, using our Outlook API.

Jack BeVier (14:19)
Mm-hmm.

Craig (14:20)
You're allowed to look at my send items folder only. You have no other access. And you also have no, you have no send access. You can write drafts, but you cannot send. So what I need you to do before.

Jack BeVier (14:34)
so you can

can you can put guardrails around it.

Craig (14:38)
absolutely, absolutely. And there's a litany jack of security docs of folks that have hardened it. ⁓ I think it's still a major concern for many, mostly because it's guys like me that are doing it. I've never had any security experience. I've never had any programming experience. Let's be honest, I'm kind of the guy that's ripe to get hacked, right? But ⁓ I think

Jack BeVier (14:40)
okay. Okay, so you can... Okay.

Craig (15:03)
One of the things that I've explained in the notes here is you can set it up in sort of its own little instance in what's called Docker. So like if the world is open to the internet, Docker then holds the bot and only allows certain things in and certain things out of what the bot can do. And so that's one thing that I've set up. ⁓ But to answer your question, I would like to walk in the work in the morning and have Rocky

pull up all of my emails where I've sent folks quotes, DSCR quotes or RTL quotes, where I've sent them apps or where I've sent them a form. So, hey, I sent you a form for a DSCR, but I never got it back. He says you had this great deal and you're gonna send me the form back. So what I do when I send those things, Jack, is I categorize them in Outlook and we can do that today. You just make a category, quote sent, it's got a little purple thing on it. Have Rocky.

search all of my inbox for the last 50 days, quote sent, absent, form sent, and then based on the day that I sent it, tell me the best follow-up for that particular item. Am I calling the person? What's my script? Am I sending them a text message? I can give you access to that. You can send the text message for me.

Jack BeVier (16:04)
Mm-hmm.

Craig (16:24)
Or am I sending them an email? So, you know, write a draft email, put it in my drafts, I'll review it as soon as I get to work and I'll send it if it looks good. And imagine, Jack, imagine how much more we would convert if we had not just a follow-up system, Jack, but one that is literally tested, like we test a marketing message. We've tested out our follow-up sequence so well that we know that we'll convert this much more.

Jack BeVier (16:37)
Yeah, accomplish. Yeah. ⁓

Craig (16:53)
because we're following up based on a very logical sequence and that sequence is being orchestrated automatically for us.

Jack BeVier (17:01)
Yeah, yeah, that's super interesting. I love that.

Craig (17:05)
And

frankly, honest to God, Jack, I really believe this swallowed my heart. I think we could have that done by the end of the week. We wouldn't have now. Now, would we have that sort of time tested, you know, message? No. But could we set up the architecture and sort of the workflow for it by the end of the week to where I walk in at seven o'clock in the morning and the thing is literally at least written the drafts for me. I think I could have that done by the end.

Jack BeVier (17:13)
Yeah.

Hmm.

That's super interesting. follow God, the security thing. This is a nightmare, right? Like I can just see all the, like anybody who's got any IT background from a security perspective is freaking out right now. But all the entrepreneurs were like giddy as schoolgirls, right? Like, ⁓

Craig (17:42)
or or maybe

One

of the other things that you tell it not to do is you're not allowed to talk to other bots. ⁓ There's a worry that there'll be prompt injection, Jack, that someone else will prompt my bot to do something that it shouldn't and then get all my data or all my ⁓ security stuff.

Jack BeVier (17:58)
Right.

Yeah, that's super interesting. But I could see Yeah, I see I see the vision that you're talking about, though. And that's a very exciting idea. Very exciting idea. I mean, the idea of infinite labor, right? Like the idea of, you know, is an entrepreneur can't come up with some use cases for that, you know, do something else.

Craig (18:25)
forget the

name of the guy that runs the all in podcast, Jack, where it's the four guys and he's usually sort of the host of the show, but he has, has his own venture. There you go. He has his own venture capital firm and he was talking with his super smart guys like you the other day that worked for him and he's just kind of unleashed them all on this open call thing. He's like, you know, go learn about it and let's start implementing it. And they talked a lot about security. So I highly recommend folks going out and checking out that, that, uh,

Jack BeVier (18:32)
It's in Calcans.

Craig (18:54)
that particular podcast, but he, you know, one of the concerns here, Jack is the cost. So let's say I'm running this thing eats up tokens like PES. mean, it's, and so one of the, one of the architectures that folks have figured out is, Hey, Rocky, you're my, you're my, I'm speaking to you and you're speaking to me in Opus 4.5 because that is truly the very best model right now. Like that's, that's how we talk to one another.

Jack BeVier (19:04)
⁓ yeah.

Craig (19:23)
I program him how I want him to talk to me, what his persona is, know, like the cadence, all of that. And that happens in Opus. However, when you tell my memory agent or my ⁓ copywriting agent or my capture agent to write code that makes them work, all that happens in ChatGPT or Gemini or like

Why would I use Claude Opus Jack with the most expensive model there is to go out and do web research when Google kind of has that down? And it deep-friends everything.

Jack BeVier (20:02)
Yeah. Use

one of the fast models for that easy stuff. Yeah.

Craig (20:06)
Fast and cheap, right?

And so a couple of really cool things that are happening, Jack, and I think this is going to happen before the end of the year where people who embrace this technology and learn how to use it really well, I think there's going to be a real case for running these models locally. So there's a couple models now that are relatively inexpensive. Kimi K2 just came out last week. That's supposed to be really great and cheap.

Jack BeVier (20:25)
Mm-hmm.

Craig (20:36)
⁓ I believe it's Chinese. You can run it locally and then ⁓ Mini Max is one where you know relatively cheap and you can run that locally. So anyway, getting back to the podcast. So the guy who was running the show said, so what's this kind of costing us? You know, he was like, you know, we're probably burning through about three to $400 a day on our setup and he was like, that's just one agent, right? And he goes.

The kid was like, yeah, you know, but we think we'll probably have 20 agents by the end of the month and each one of those will be burning through 300 bucks a day. And he's doing all the math in his head. Like I know you're doing right now. And I guess that Jack, like you tell me if the, if, the bots can do, make me more effective and it's proven or you make us more effective and efficient and we can do more. What's it worth, Jack? it.

Do you say, that person would cost $80,000 a year. Does it come down to that?

Jack BeVier (21:39)
Yeah, yeah. I mean, everyone gets in that everyone gets an assistant to get that there's I mean, there's a multiplier effect to assistance, right? For sure. So, ⁓ no, I mean, that makes a ton of sense. If you can, if you can prescribe the, can prescribe the work that they're going to do. And it's that it's that work. Also, you know what, that's an interesting idea, have them do the work that is like the

the counter to their personality, right? Like you don't need it. You know, you got a detail oriented person. You don't need them doing detail oriented work. You need them doing outreach, right? You need them picking up the phone. You need them engaging with people. You know, you, you know, you need an extroverted assistant for all your introverts and you need an organizer for all your, you know, business development people to like log their stuff in Salesforce and whatever note system they have. Right? Like that would be probably like the lowest hanging fruit is like,

the contra positive to your personality, create that because then it's your entire knowledge base. And we've got the whole, you know, personality spectrum covered and you're an animal, right? You have no weaknesses anymore. You know what I mean? Like everyone's got weaknesses. And if this thing is going to be like the contra positive of your, of all of your weaknesses, that would be a dangerous multi.

Craig (22:53)
It would be. And so I think the bigger question though, Jack is, you know, what's it cost to set it all up? know, like, you have that kind of patience? Do you have that kind of patience to pay for the tokens while you're sort of muddling through figuring it out? Cause it's still a little techie, little kludgy. And then, and then.

Jack BeVier (23:06)
Yeah.

Craig (23:17)
You know, is it going to be, I've got set, we know guys, Jack, they have a whole crew of VAs that they pay for every month. They've got assistants, they've got acquisitions managers and disposition managers and all of this overhead. And so, yeah, I'm asking you, like, can you see people at this stage, Jack, at this stage where you still got tinker, you got to get into the workshop and you got to like, you know, bang a hammer and turn a wrench, like, are that,

Will there be enough people there that embrace it and say, hey, whatever it costs, I'm going to set this up because I know ultimately it's going to save me, you know, two X in salaries or, you know, whatever.

Jack BeVier (23:57)
I mean, I think that there's a first mover advantage to these ideas. And the question is, yeah, are you so early that you're just burning cash and something and you should have started three months from now versus, you know, how much of that learning curve can you get up and how much of your organization can you get up that learning curve so that you're actually changing the nature of the organization, right? Cause if the people start to, if you,

one of the things that we've tried to do with, like the AI initiatives internally here at Dominion is if I can get all of these mortgage professionals downloading their brains into using these new tools and downloading their ideas and being creative and coming up with stuff, it's going to change the nature of the organization, right? It's not just that, we've got it's Dominion with some AI enabled tools. It's the entire company is like now organically creating things that other companies are not going to organically create. ⁓

And so I think that, I mean, I'm hearing a similar, you know, similar vein here is that it's like, if it's such a step change in terms of the power of the tool, the argument is that like, you need to just get everybody native on that tool. Like, you know, period, period. You can't be a carpenter that doesn't, it's a brand new hammer and it's the most amazing hammer ever. Like you just put it in all the carpenter's hands, period, hard stop. don't, you know, because, because it makes them multipliers of themselves.

Craig (25:15)
I agree. I really do Jack. I was just, as you were talking, I was thinking like, what if we talked about a couple of use cases now for like, you know, our borrowers, just regular investors, guys that are out there slaying the dragon every day and they're, they're, they're embracing some tech. Let's say you're a wholesaler Jack and there are some great tools out there. Prop stream. ⁓ You could probably name several where, guys can, you know, stand up better wholesaling operations using these tools. Right.

But what if I had a bot that would just go out to the Maryland Register of Wills and just scrape that thing all night for every new probate that came up ⁓ in every county that I'm operating in and then put that all into a Excel spreadsheet or Notion spreadsheet or whatever you're using for the backend database.

Jack BeVier (25:49)
Mm-hmm.

Craig (26:07)
And then it reaches out to a printer like a go big printing or one of those guys and said, here's my, here's the postcard that I want. Here's all my stuff. Make it happen. Charge my credit card. I'm done. And now you've got an automatic mailer that goes out once a week. Every time the, every time probate list updates, why wait till the end of the month? Just give me the new ones. Right.

Jack BeVier (26:28)
Yeah,

yeah, the half of the listeners just said Craig shut the hell up like that was exactly what they're thinking.

Craig (26:33)
I

Yeah, why buy this when you could just scrape it for free, you know? ⁓

Jack BeVier (26:41)
Yeah.

Well, Hey, super interesting stuff. We're going to be talking about these concepts a lot more going forward because it's, just, we just both think that's where the world's going. So, ⁓ obviously we will keep, tailoring the content towards, know, everyone's real estate investing business, ours included. ⁓ but, you know, glad you're, glad you're having some fun with it. Love seeing you lit up about it. I got to get myself a little Apple mini and play with this thing. You know, I'm looking forward to it. Sounds good. Sounds good.

Craig (27:06)
Come on over my office, Jack. We'll just play with mine.

All right, everybody, hope you guys enjoyed this one. We're gonna have much more. I can't wait to show you guys what the thing does. Maybe a couple demos on the next one when I get it really up and rocking. That'll be fun, right? Watch it do its thing. And we'll see you guys on the next one. Love to hear your comments. And as soon as I'm done putting this PDF together right, so it looks good for everybody, we'll put it in the show notes and you can.