I Built the Ultimate AI Board of Advisors (free claude skill)

0:00 So right here you can see that I've got Jeff Bezos, Paul Graham, Alex Herozi, Gary Vee, Dan Martell, Gino Wickman, and Warren Buffett sitting here in what I call the roast room. And what's

0:09 happening is all of these advisers are writing their opening positions and then they're going to have a debate about the actual question that I submitted to them. So right here you can see that

0:16 this debate is about the Austin Inerson conference in June 2027 and they have four different options to choose from. I gave them A, B, C, or D. And what's actually going on here is this is linked

0:25 to a cloud code session as you can see right here where I've shot off this prompt where I basically did a slash roast room which is the skill that I'm going to show you guys and give away

0:32 today. And I submitted the question with all this information about the options, business information, resources, what would success look like? And now we're able to actually sit here and watch them

0:41 discuss this live. And so obviously this visual interface is kind of just a fun visual interface that sits on top of what's actually going on. And what's really going on is we're still running

0:49 this skill inside of Claude Code where we have these different personas and they're going to be discussing this. So I've done things in the past where I actually run a slro, but I've made this

0:57 a little bit more formal and I've made this, you know, more formalized personas and I call this the roast room. Now what I think is cool is all of these personas were pulled as inspiration from the

1:05 actual people themselves. Jeff Bezos focuses on one question which is is this the right bet and how reversible is it? Paul Graham focuses on one question which is is this something people

1:14 actually want and is it growing? So, all of these essentially agents or council members are thinking about one very specific thing. Warren Buffett is thinking about what does it do to cash

1:23 risk and long-term value. And obviously, I'm going to break down how I built this and what actually lives inside of each of these specialized agents brains because there's a lot of data that lives

1:32 inside of each of these guys' brains. So, right here you can see they actually start discussing and there's something going on with Jeff Bezos because I have this sort of like minimized so it looks

1:40 kind of goofy but we added in these little animations so we can actually see them talking. They're going to give their opinion on what options. So you can see Jeff Bezos came in with a B and

1:48 he scored them. So he gave B seven out of 10 and we can see why. We can also see that Paul just went with B and he gave it a 6 out of 10. Hermosi is also running B and gave it an 8 out of 10. So

1:58 this is basically how the council works. They're going to go through in one round and then debate and then they're going to go through in another round and then debate and they'll come back to us at

2:05 the end with a final vote and verdict. And I can go ahead and speed this up because all of these decisions have already been made. This is just like I said the visual interface. But in this

2:13 case, you can see that all seven of our adviserss here voted B and they scored it an average of 7.1 out of 10. So now what they're going to do is they're going to read through each other's

2:21 positions and they're going to fire back. In this example, they're pretty in line already. But it's really interesting when they don't actually agree and they have to try to persuade

2:29 each other on why their decision is a better decision. And because I have cloud code open on the left, we can see what's actually being facilitated. This is sub agents. It's not an agent team.

2:37 So that means that each sub aent is basically talking to a chair and the chair is facilitating the debate. So right here it says all seven seats in round one backed a gated B which means

2:47 the chair treats as a warning. So round two is asking Buffett and Martell to argue for option C as hard as they can. And then we're going to see how they think about that and then we'll see what

2:57 the other founder council members think about as well. And now you can see how they're sort of like starting to make these connections between each other. They're arguing with people or they're

3:06 agreeing with people. So Gary Vee right now is talking to Buffett. Dan Martell is saying something about Gary Vee and they all start to actually come together now and hopefully they're trying to get

3:16 to an overall roast room consensus. And as this is happening, you can sort of go through and you can see the key points. You can see what round one looked like. You can see what round two looked like.

3:24 And you can more specifically see what Bezos said to Gary Vee. You can read through it here. You see, you have things like, I disagree with this. I can see it here. Blah blah blah. Here's my

3:32 stance. And you can look through all of the stuff. This is what Hermoszi was saying to Bezos. But now we're moving on to the final round, round three, which is the actual vote. And it looks like

3:41 it's going to come back and it's going to be to go with option B. But we will see. And there you go. You can see that they're starting to come in with their final votes. And what's going to happen

3:48 after is we are going to be delivered a final verdict that we can sort of read through and see like the key decisions and the key points that were made. So you don't have to like watch this whole

3:56 thing. Usually what I'll do is I'll shoot off a couple prompts. I'll come back and then I'll just read through the verdicts. And it really helps me make decisions. And I don't always just

4:03 blindly take whatever the council gives me. But what's helpful is to read through each of the decisions and and read through the perspectives because sometimes that helps me just think about

4:11 it in a different way even if I don't actually go with their opinion. So this was basically the final verdict. It was to go with conditions run B as a $3,000 annual plus Austin tier capped at 120

4:22 seats and sign nothing until 50 members have paid 150,000 by January 15th. And what we can do is we can actually download this as a PDF. So, if I download this and I open this up, you

4:30 can see that we get this PDF with our scores and with all of this information on each round, what they weighed it, and how we can sort of like take this, send it to our team or make a decision,

4:40 whatever we need to do with it. So, that was an example of what this looks like and the deliverable. I'm going to break down how exactly this works, but first, I want to show you how you can get it

4:47 for completely free. You're going to go to my free school community. It's linked in the description. You're going to come in here, go to classroom, and click on AI skills. and there will be a link in

4:54 there that takes you to the full repo for this roast room setup. So that is what the interface is like. You can see that if I click on past sessions, I can actually look at all the different PDFs

5:03 that has been created from this roast or so from the roast room. And we could also actually replay the council. So if I went to this first sales hire or keep selling, it will go ahead and basically

5:13 replay this whole thing for me, which is pretty cool. And what's interesting is sometimes the actual prompt that you use to shoot off for the roast, it might not use all of the personas here. If it

5:24 realizes that, oh, we only really need to hear from Buffett and Martell in this specific example, then it will only run Buffett and Martell. The one that we just did utilize all seven. So that was

5:33 great. But this specific council, as you can see about first sales, hire, or keep selling, it only had to talk to Graham, Hermosi, Martell, and Buffett. By the way, guys, I've got this completely free

5:41 SOP for you about getting your first aation client. It's going to go over the exact steps that has been proven for hundreds of our AIS Plus members to get their first paid gigs. It goes over the

5:51 one sentence service pitch that can get you started today. Why your first client should cost you money. The 5-minute video that answers, can this person actually deliver before you've actually

5:58 received any money. What to do when you have zero case studies. There's so many good things in here that are going to help you out. Even if you already do have clients, I would recommend grabbing

6:06 this because like I said, it's yours completely free. So, if you want to grab this, there's a link for it down in the description. Let's get back to the video. So what this means is it stores

6:13 all of your sessions. So if I come back in here and I go to claude code and I open up the files. If I go to my projects and I haven't renamed this yet. I used to call this founder council but

6:23 I can go to my founder council. I can go to my sessions and I can see these different sessions that I had run and then I can actually investigate them and look at the brief. I can look at each

6:30 round. And that's basically what we're seeing in this interface. We're just seeing it in a nice pretty way. But of course you don't need the interface. You could just run these founder council or

6:38 these roast room sessions without the interface. But I thought the interface was a nice touch. But let's talk a little bit more about how this actually works because I think it's super super

6:46 interesting. So what we have is the roast room. So that's basically us and Claude Code is essentially a chair. It's the facilitator of the roast room. We have seven advisers like I showed you

6:56 guys in the actual demo. And what happens is each of these is their own sub agent. It goes through three rounds where they write their opening position. Then they look at each other's

7:04 positions. And then the chair facilitates, hey, you argue with him, you argue with him. And then they go through on round three and they vote and then everything is saved and we get a

7:12 final verdict PDF. Now, the reason I wanted to use sub agents is because that way we could have one chair facilitating. It's also cheaper than if we ran this as agent teams. I used to

7:20 run the roasts as agent teams, but I switched back to sub agents because I just found that the agent teams and the sub aents didn't really give me different answers. They were giving me

7:28 very similar answers and sub aents were cheaper and they felt more like I was able to keep the actual real voice of whoever I was creating. It just felt more organized. So these are all their

7:39 own individual sub aents and I will show you guys what those sub agents look like in a sec because like I said we scraped through a ton of data to make these. So what that looks like is for example Alex

7:47 Herozi is focused on offer pricing and sales. We have an actual agent file. So it's a sub agent file that gets called in the skill and it says hey you're going to you know spin up this Alex

7:56 Herozi council agent. This file basically contains the way he thinks who he is the question he owns and how he answers how he makes decisions. And it also links to all these other files that

8:06 are basically his brain. So the eight rules he decides by his named frameworks, principles, positions, stories, voice, and sources. And all of this was just scraped from public stuff

8:16 that he's put out there, YouTube videos, exosts, LinkedIn posts, blogs, things like that. You can see in this specific example, we went through about 124 sources, over 1.3 million words, and we

8:27 put that into his brain. So, we did that six more times with all of these other members of the actual council as you can see right here. And the cool thing is if you wanted to maybe swap out like Dan

8:37 Martell and swap it in for someone else based on your industry or based on your use case, then you could basically just run the same thing. You say, "Hey, you see how these seven advisors are set up?

8:44 I want you to build me another one that's just like him, but let's build this one for um you know, Barack Obama. I don't know, whoever you want it to be." And then they go through and they

8:54 actually score every option. They wait it by relevance and then that's sort of what's debated on. So when you actually come in here and you clone this repo, you'll basically just give your agent

9:01 the link to this and then you'll be able to just say, "Hey, install this and it will put it into your project or you could create a new project for it if you'd like. Now, what's cool is let me

9:09 open up thiscloud folder to show you what these agents look like. So in the cloud we have agents and here's where you can see we have our council Bezos with the brain and how you talk, how you

9:18 debate. We have Buffett, we have Gary Vee, we have Hermosi, Martell, and Wickman. We've got all of them right in here as sub aent files. And you can see the sub engine files are not that long,

9:27 but what happens is we're routing it to the brain. So we're saying, hey, you can read your hot MD and then you can look in your frameworks or your positions or all of these things when you need to

9:36 make decisions. And so what that looks like if I open up our skills right here, you can see that we have these things called brains. So in the brains, we have Hermosi, Martell, we have all these

9:44 people, right? So for example, let me just open up the Hermosi brain. We have the hot, which is basically his hot sheet, which is the eight rules he decides by. We've also got some

9:53 frameworks here like value equation and grand slam. We've got money model and client finance acquisition. These are all things that he talks about a lot if you've read his books or if you watch

10:00 his YouTube videos, stuff like that. Decision loop, signature questions, etc. We also have information about his books. We have frameworks, we have positions, we have sources, we have

10:09 stories, stories that he has told, which is really, really interesting and it makes it feel more personal. And then of course, like I said, we've done the same thing with these other brains. So if we

10:17 look at Jeff Bezos real quick, same thing. We have the hot goes over his worldview, eight rules he decides by. We have his positions on certain things. We have his sources. We have his stories as

10:26 well. All of these are just a bunch of different markdown files that were scraped and created by having a bunch of agents go do research on these different people. And now we have this framework

10:36 of, okay, if I wanted to add another person to our council, here's how we do it. So, it's really, really a simple setup, guys. And like I said, the visual interface I showed you in here, purely

10:44 visual. You don't actually need this. The skill will do all of the same stuff. It'll still generate the report for you. It will do all of that. But if you want to watch it, then you can watch it in

10:52 here. And the last thing really that I wanted to talk about here is how to actually create a good question for your console. There is obviously the idea of building a good question, but then

11:00 there's also like the context. So, of course, when I'm asking it questions in here, it knows everything because it's already in my AIOS, which means if it has other questions, not only can it ask

11:09 me because if it doesn't feel like it knows enough to send it off to the council, Cloud Code will ask you, hey, in order to send it off, let me ask you this question, this question, and this

11:17 question, but if you have your AIOS set up correctly and you have all your connections and you've got your own personal context and stories and meeting transcripts and whatever it is already

11:25 loaded into Cloud Code or Cloud Code has the ability to touch those, then it will actually go pull that data and it will help the council make an even better decision because all that context

11:34 already exists. So, what do you actually want to give it when you're asking a question? Give it the decision. This is what I want you to tell me. Give it options. Giving it two or three options

11:42 at least is better. Otherwise, it's sort of hard to like categorize what potential paths you could go down. And it may even be more overwhelming because then you might feel like, oh, I actually

11:51 have seven paths because each persona gave me a different option. But in reality, maybe all seven are saying some combination of like option B. So, give them options, give them real numbers.

12:00 So, if you feel comfortable telling them like revenue and team size and and things like that that contextualize this that are really important to actually making a decision, give that

12:09 information. Give it limits like deadlines, hours, and budgets. Give it what success looks like. This one's so important. You have to give it context on why do I need this decision? What

12:19 does good look like? And how do I get there? How should I be thinking about that? And then, of course, which way you're currently leaning and why? Because, as we know, sometimes AI likes

12:27 to agree with you. But this roast room is basically set up to make sure that it's not just agreeing with you and maybe it's going to convince you the opposite or maybe it's going to validate

12:34 you, but it's not just going to blindly do that because they all have their own principles and they all have their own way that they think about things. And of course, it should be very, very simple.

12:42 Once you get into this interface, you can click on what to put in your question and you can look through this. You can also say, "Hey Claude Code, look at the roast room and help me turn this

12:50 into a good roast room prompt." And then you just talk for a little bit and it will turn that into a prompt and you can shoot it back in. And then you basically just run the skill and the visual

12:58 interface will pop up and start running through all three of those rounds. So anyways guys, I know that this is a very simple skill honestly like a very simple walkthrough of the skill and how it

13:07 works and what it does, but it is something that I do use a lot and it's something that I think that you should all get in the habit of. It's a really important idea of being able to sort of

13:16 outsource your thinking but not the understanding and be able to leverage other people's subject matter expertise because at the end of the day we're all human and we all have subject matter

13:25 expertise in a few domains but we won't be able to have subject matter expertise in every single domain possible. But what you can do is you can use agents to go store some of this knowledge. So

13:35 let's say you're building an app in cloud code and you don't really know that much about security. What you could do is you could build a security agent and you would say, "Hey, go research

13:42 everything there is to know about security. Go read these articles. Go read these books. Go watch YouTube videos and store all of your knowledge in a skill or in, you know, your agent

13:50 brain." And then if I have any security questions, I can just ask you. And it's just about building all these little specialized agents and outsourcing the thinking, but making sure when they come

13:58 back with, hey, based on my findings, based on my research, here's what I'm recommending. You still need to take that recommendation, process it, understand it, and don't just blindly

14:07 accept what what it comes back with. But building specialized people like this really helps you feel more confident in the answers you get back, especially when you have them debate each other.

14:17 You could have a, you know, a security agent and you could have like a CTO agent and you could have them discuss things and bring that verdict back to you. So, it's all about figuring out how

14:27 can you leverage other agents and other perspectives to help you make a decision. But like I said guys, you can find that for completely free inside of my free school community. But that is

14:35 going to do it for today. If you guys enjoyed the video or you learned something new, please give it a like. Helps me out a ton. And as always, I appreciate you guys making it to the end

14:41 of the video. And I will see you all in the next one. Thanks everyone.