Projects
![]()
No projects found
Investors
New
DAO Treasuries
DATs
Lend / Borrow APRs
Chains
Categories
Spaces
LP Bonds: The Future of Sustainable DeFi
Recorded: Sept. 6, 2026 Duration: 1:45:28Space Recording
Full Transcription
oh, oh.
oh, oh.
Hey everyone, welcome. We'll get the space started in a few minutes. If you guys would like
and retweet the space, we get more people in here.
Appreciate you, and we'll get this started in just a few minutes.
All right, welcome guys. We'll get this spaces started. So thank you guys for joining us.
It's nice to see the crypto market coming back to life a bit after you know quite a bear market.
I mean, I still feel like we're not completely out of it, but some good signs of life. You know,
obviously huge surge in Bitcoin and Ethereum and some other coins, and a lot of activity going on
on Robinhood. So that's good to see, and that's one place that we're working on deploying
our technology. So we hope we can capture some of that. But yeah, there's
a you know there's definitely action going on in the crypto space. So good to see again
after such a long drawn out, pretty much a year of a down market we had from October last year till
nearly October this year. It's been down, so finally we're kind of reviving a little bit.
So much needed and much welcome. So on this space, I'm gonna do some updates on basically
where we're at, what we've been doing, and yeah, just keep you guys in the loop with everything. So,
if you guys have been paying attention to our marketplace, you've noticed
that we've continued improvements to the user experience, reducing the number of clicks that
users need to basically mint LP bonds. And we're also working on something,
something I'm very excited about, which is basically we have a referral program currently
for our platform. Which basically, if you guys refer people that use GMI,
you guys get airdrop points. That's nice, and that's a generous referral program.
But what we're doing now is something where anyone can make money with LP bonds.
It's something I'm very excited for. This isn't out yet. I'm going to say one to two months,
maybe closer to two months till it's out. But basically, we're going to build our referral program
out on for upgrades. So whenever you upgrade a bond, you pay a 20% upgrade fee.
That's essentially our revenue stream. And what we're doing is we're extending that to the user,
and we're actually pretty much willing to give most of it away to people who participate
and refer others. So anyone you refer who upgrades a bond, they pay a twenty percent upgrade fee.
You're going to get half of that at level one, ten percent of the twenty percent.
So as a as a person referring that, that's a pretty nice incentive and an income stream. So
and then whoever we have three levels in this as well. Whoever they refer is four percent you get,
and whoever they refer is is two percent of that. So you can capture actually the way we've designed
it is a user three levels down actually would absorb the majority of the fees for the referrals.
Yeah. So again, anyone you anyone who upgrades a bond, we've made it so those fees
are pretty much captured a large part by referral system. This is a growth strategy
and kind of again a revenue generating thing for not for you know again we're giving most
of that back to the user. So. The reason why we're doing this is we want we want the bonds to grow.
We want to give the users a way they can make money with this today, not waiting for the airdrop
that they can make money today. So that's something I'm really excited about. So again,
it's not live yet, but we're working on it. The kind of first stages are done of the building out.
So I'll keep you guys updated on that. So that's one thing we've got coming here. And this month,
our major focus is we have we have kind of a couple things on our plate.
One of those things is we're working on the LitVM deployment. If you guys don't know what LitVM is,
it's Litecoin's EVM. So Litecoin has their their chain. It's called LitVM.
It's in testnet right now. We're deploying on that testnet, and yeah, I I think part of our strategy
here going forward is to kind of catch these big upcoming chains early on, right?
It's something we should have done with Robinhood, but we weren't quite ready yet.
But going forward, any major chain launch that's coming out, we want to capture that.
And by the way, if you guys ever notice a big chain coming out, let me know.
We want to be there because that's when there's the most activity. Projects are the most hungry,
so that's part of our growth strategy as well. So that lit VM deployment is underway. ETA on that.
Hopefully within two weeks. You know, sometimes it's like it's up to the dev.
I wish I could press a magic button and have it done, but the dev is dev team's cranking that out.
So hopefully within two weeks, we're fully deployed on lit VM. And then concurrently,
we're also working in Robinhood as well. So we have these two chains in the pipeline coming out.
Now, being on a lot of chains is great, but the the key part is you know getting that traction
going, which brings us to the next part that we're focused on, which is a heavy focus
on business development. And so we've kind of. Reinvigorated our recruiting system.
One thing I do these days, which I recommend everyone to do, is I call it systems and SOPs,
and it's something that I've learned and I take really seriously. Actually,
the AI advised me for that, which I really like. So, systems and SOPs basically means you're not
When you have a million things going on, it's impossible to keep track of things in your head.
You're just constantly putting out fires. It's it's very energy demanding
and and a little exhausting, and honestly takes away from from your life in a way.
So the solution to that is to create proper systems that you don't have to
you don't have to manage really. I learned a long time ago, actually, is a quote from like.
When I first got into sales, it was like I was like 18 years old, and I picked up a good quote
from my manager there, which is, "You don't manage people, you manage expectations,
" and that stuck with me, and it reminds me of the systems and SOPs.
So I don't like to manage people; I like to just manage the expectations for whatever role
they're in charge of, and that kind of takes out so much of it. It takes out there's there's
there's nothing personal about you know how you run a company. It's just
you know things have to run, and and then there's systems in place for that.
And then the AI makes this really good, and I'll tell you guys how we do it these days,
or how I do it, and what I'm working on getting our whole company on board with.
But we talked about last time a little bit using AI loops, and I recommend everyone to do this.
If you're not doing this, this is how you become an AI power user. An AI loop basically means
that you're like most people. The way they use AI is they ask one AI one question
and they just take that and go with it, which which is not the optimal way to use AI.
The optimal way to use it is to you know ask it whatever question, and then the next thing
you should ask it is something along the lines of what am I missing here?
What you know is there anything that should be added, subtracted that I'm not thinking about?
So you kind of flesh out the first the first kind of design. And for instance, I did this with our.
Our partner analysis, right? So when I was first analyzing partners, I was like, you know,
I asked it some questions like, what's their TVL, what's their, how's their traction, you know,
Twitter followers, that kind of stuff. And I was like, okay, what other questions should I ask?
And so it gave me those questions, right? And then I took it to another AI.
And my go-to AIs right now are the latest version of Claude, that's Claude Fable 5.1,
and ChatGPT Soul Six. Those are the two. You you really need two AIs, and those are
my two favorite ones right now, the latest versions of those. So I basically once I flesh out.
What the proper questions are, then I take that and I'll start a new conversation. I'm like, okay,
now rate this LP bond partner analysis, right? And I'll run that individually through Claude
and ChatGPT. So when you're doing loops, you ask two different AIs the same question.
They come up with the, they each come up with their independent answers.
Then you take the answer from Claude and you share it with ChatGPT, and you take the answer
from ChatGPT and share with Claude and say, here, rate rate Claude's review, and you know,
is there anything you're missing from yours? And then they do like an updated version. Okay,
so now you've cross-reference, and then you use what's called an orchestrator.
An orchestrator is picking one of those AIs, which you know I use Claude as my orchestrator.
So then I'll take ChatGPT's latest feedback, and I'll feed it back to Claude. I'm like,
this is its latest version. Anything you're missing or change or whatnot? Rate it one to ten.
And then it's like, okay, I didn't catch these things. Let me let me add this in here, or oh,
I didn't think about that. Good points, and or or I reject that. You know, it does that as well.
And so I'm like, okay, now rate it one to ten. I ask that every time. Every time I have it do this,
I have it rate it one to ten out to one decimal place. So now it's like, it's like, all right,
here's the latest version. I rate it 9.6. Okay, cool. Take that, and then take it back to ChatGPT.
Say, all right, this is what the latest from Claude. Do you see anything that's missing here
that can be improved? And ChatGPT actually—it's a little bit lazy sometimes.
It doesn't do a full QA, like it doesn't do a full analysis. It just kind of gives—it gives
good points, but it doesn't do it thoroughly. So I ask it to do like a deep, full QA.
So then it gives all its feedback, and it could be like 20 points. And then you give it back
to Claude, and Claude's like, they're called adjuncations. And so basically, Claude goes through.
It's like, all right, four of these accepted, one of these rejected, and then gives you
the latest version. Then you take it back to ChatGPT and you say, okay,
rate this latest version one point one to ten. And then it says, okay,
this version is now you know a nine point two. Claude was giving the higher rating.
ChatGPT is giving it a lower rating, and it gives all its feedback. Then you bring it back to Claude
and say, here's ChatGPT's latest feedback. And it does it again, and then then you take
the latest version from Claude, give it back to ChatGPT, and you do this cycle
throughout the process until you get to literally a 9.9 or 10 out of 10.
And I found that in the different processes I've done this with, it takes sometimes it's fast.
If it's a deeper thing, I can spend hours doing this. So this is how I built
our our current recruiting SOP. And and by the way, like it's just incredible the the the level
of work it does. Like we had a spreadsheet obviously that we were tracking our recruits, right?
Or recruiting efforts, but this was like man-made pre AI, and it created a whole spreadsheet.
Like this is like top tier, you know. Like I don't want to say Fortune 500 level,
but maybe a 50th tier Fortune 500 level recruiting SOP. Like it's incredible,
but it takes a little bit of work. You got to do. You got to put in the work.
You got to do the the loops on it. But the outcomes you get are really good.
And so it creates the spreadsheet. It creates the whole process. I actually created
a whole document, which is how to make an SOP, and share that with our team. So,
and I tell the team to put that in their projects folder. You know, every AI has a projects folder
where you give it context. So, I have that in my project folders, and I've basically told our team
to do the same. And yeah, so this is if you guys aren't working this way yet, I highly recommend it.
You know, again, and why are we doing this in the first place? It's because of the the consideration
of time. We only have so much time in a day, and it's really important to focus on
those top four percent activities. You know, and there's always going to be fires that come out
and stuff like that. But the point is, you have to kind of vigorously determine what is
the most important stuff for you to focus on and to do those things. And beyond that,
you have to delegate or systematize the rest, systems and SOPs. Again,
the whole premise of this is to save your time. That's why I do it. So again, not managing people,
managing expectations, create systems. People follow the systems; they're essentially
like a checklist, and that's how you get work done. So I, I'll add on to that too. You know,
one one thing I will say, you know, one thing I've really thought about,
and and I always try to keep top of mind, gratitude. You always hear me talk about it.
Gratitude is very important to me, and I always try to keep that attitude of gratitude, as they say.
And if you think about it this way. I think about all the different tools
and and and benefits that AI provides. You know, it helps. It does my design work for me.
It it helps me with the devising. It helps me with strategy. It helps me write any marketing piece.
If you add the value of all those things together, what AI does today, um, and I asked AI this
and kind of filtered through. It's for for me. It's it's and for many of you guys as well.
It's like literally millions of dollars a year, as if you're as if you had staff of
millions of dollars a year at your fingertips that you can pay a subscription for twenty bucks
a month. I mean that's crazy. I really have to emphasize that we are wealthier today than any point
in human history by far. It's absolutely incredible. Even even years ago, even just three
four years ago, nobody had this. Nobody had this. Now everybody has it. Even with the free version,
even the free versions of these AIs are better than the best versions two years ago. I mean,
it's incredible. So everyone is beyond wealthy today. I think it's about staying in alignment
with that. You know, staying in that, staying in that vibration, and and appreciating what we have,
and then leveraging it. And if you take that, if you think about that more,
think of all the people who are not really using AI today, right? I walk down,
I go for walks around, I like think about it. Like, see, I see a street vendor
or a retail worker in a store or a taxi driver. These people are not really using AI for their work,
but for us, we could leverage AI enormously. And I think there's a widening gap there
between those people who use AI effectively and those who are not really using it yet. So. Yeah,
this is kind of the latest on how we're solving problems, what we're doing, how we're thinking.
You see, there's anything else I wanted to cover.
I kind of front loaded the conversation here. One second.
Yeah. So again, what I'll say. So that kind of talks about our focus and stuff. And again,
just to add as well. Again, major focus for us right now are the deployments we're working on.
Again, Lit VM and Robinhood, and then getting the business development rolling.
For not only these chains, but you know, for all projects, you know, and the way I think about what
we're doing with LP bonds, this is, this is, you know, this is the vibe I get, you guys. Like,
there's a lot of tokens in pain right now. I'm sure you guys have seen it.
There's a lot of tokens that are just in a lot of pain, right? You look at their charts.
Their charts look like, you know, like the flat line. It's like rough.
And I see us as a helper to all those projects. That's really what it comes down to.
Is a couple things comes down to tokenomics. Comes down to, yeah, tokenomics is massive, right?
You got to get those tokens off the market, and then again, the way people have been doing this is
they've been giving a yield. And again, I go look at Dex charts and stuff. Sometimes, man,
it's rough. Go look at Dex charts where they're giving their token as an incentive. It's sad.
They could be doing good as a Dex, right? They could be doing good good as a Dex,
but if their tokenomics are not dialed in. And it's not just limited to X's either. Obviously,
if the tokenomics are not dialed in, what happens is there's a constant sell pressure,
and the chart bleeds, and everybody looks at the chart. So someone going and looking at a token,
they go look at the chart, and they're like, you know, it just ruins everything. The chart is bad;
it hurts you so much. So I think it's such an underrated thing, but so critical to design something
where again the chart does well. And again, how does the chart do well?
It does well with basically what we've created—a system that takes those tokens off the market,
a system that incentivizes purchasing those tokens, which they do. But the point is,
they incentivizing they incentivize purchasing those tokens by giving tokens, and ultimately,
the chart goes down, which takes away incentive and interest. So it's like a kind of
like they're trying to win by hurting themselves. It's not obviously doesn't work very well.
So you got to take those tokens off the market. You got to give an incentive
for people to want to participate and buy the token, and that's the yield. Yeah, basically,
basically we built, and then you obviously have to have those tokens locked down for enough time
to really take off, which is giving the whole essence by NLP bonds. So anyway,
it's kind of a long monologue here, so I'll I'll stop there. If you guys have any questions,
drop them in the comments. Might keep the space a little tighter today. If anyone has any thoughts
or questions, feel free to jump in. Mark,
maybe I'll pass it off to you if you want to add anything. Go for it.
Tommy, could you elaborate for the audience what AI models are you using
and why you prefer to use those, and why you are not using more models? And yeah, yeah,
good question. I mean, I experiment with the other models. I like Gemini sometimes. Here,
here's the issue with Gemini. At least last time I used it, which was maybe like a month or two ago.
It and I actually found this out by asking it the question because I was having a conversation
with it, and it lost like there was a message, like two messages up in the same conversation. And I
and I could see that it lost that context. I'm like, did you not see that message two messages ago?
It's like, oh, I only have context over like the past two messages, and then I lose context
really fast. I'm like, shit. Okay, that's not gonna. But Gemini is cool. I like Gemini.
Google's one. I think maybe their next version is gonna be good. DeepSeek,
I experimented with last time. I wasn't, you know, I kind of, I kind of,
I always experiment with the other frontier models. So that would be like. Probably Gemini, Kimi.
I haven't tried yet. That's kind of on my list. I was trying to do and they shut down their sign up,
so I was like, all right, maybe I'll just wait a little bit. But I've heard good things about Kimi,
so that's kind of on my radar. But Gemini and then DeepSeek, same thing.
I just wasn't that impressed. You can kind of tell the quality of answers you get. And again,
this may, whenever they do an upgrade, the model can be totally different and way improved. So,
to be honest, to be, I haven't played with the probably latest model of DeepSeek. Last I tried was
a few months ago and wasn't impressed at that point. And what else is there?
There's another one I can't forget the name that some people like. Shit, I don't know. I can't.
I can't recall. But I, I've played with. I always like dabble with the other models
and see which one is getting good output. But my three favorites right now would be ChatGPT Gemini
or ChatGPT Claude and Gemini. Claude first, then ChatGPT, and then Gemini is like my tertiary one.
But Gemini messes up sometimes in the the project folder for some reason.
Always like doesn't work there. Whenever I upload stuff in the project folder, it fails to respond,
and they don't have a good desktop application either. So you know these things are evolving
and growing. But yeah, that's essentially how I kind of essentially use them based on, you know,
testing and stuff. And I mean, I guess to give context to, I most certainly am a power user of AI.
I use it very, very aggressively and a lot. I get my usage worth. So yeah, put it that way. Kbot,
so you got your hand up. Go for it. Hello, energy. Hello, everyone. Gabby, Mr. G, others, Mark.
Thank you for letting me up. Great room. I love your commentary. Quick question.
I I admire you using these these models. Okay, so the question is, is it possible to train
a local model by the information you get from these large language models? So you can have your own,
you know, decentralized local. It's all yours, you know, for for very important
and sensitive information and sensitive intellectual property that you want to keep, you know,
safe within your own business or your own, you know. Self custody information of a local model. So,
do you understand my question? Am I saying it right? Yeah, yeah, I understand. So, I mean,
what what I would say to that is, I mean, think of it this way: we have our Gmail, our Google Drive,
we have so much on the cloud already. I think that was, you know, our photos, Apple Photo,
Google Photo, whatever it is. I mean, there's so much stuff on the cloud. I think having
strong security is really important. But I personally, the way my the way I think about it is like.
You know, I don't really care. I have nothing to hide. I'm not like making meth or something.
You know, if someone's if someone's doing something really shady, yeah,
maybe like a local model might be good good idea or something. But like,
I think I think for most people, I don't know. I personally don't. I know I know some people go down
that path a lot too. You know, there's some people on on X that are like all into like
the local models and stuff. But I just kind of look at them like, I don't think it's really.
I don't. I don't think that's. I don't think it's really critical, personally.
I back up all my stuff, so like I have my project folders locally on my computers. So, you know,
God forbid if the AI goes down, I could re-upload my 20 documents or and you know stuff to like
a new AI and has full context and everything. Pretty good. So, yeah, that's kind of my take on it.
But everyone their own, you know. If if you want to run it locally, you're welcome to.
But I don't think it's a a necessary thing. So yeah, go for it. See your hand. Okay. So.
Thank you for that. I work in this wealth management office, and I'm doing my damnedest
to integrate AI and blockchain technology with, you know, estate planning, real estate, annuities,
insurance. Tax. It's just there's just so much sensitive information, and we aggregate information
from a lot of different, you know, companies like Apollo and Fidelity,
and and these different annuity companies and insurance companies, and they all have
extremely sensitive information where they don't allow you. It's not regulated for you to aggregate
their information. So my, you know, what what I would like to do is, you know,
capture that information from these different companies and make apps or websites
or PDFs of Even metaverses for for clients, where the information only comes from the company's AI,
and not that we trained that we got all this information from Claude or or you know,
so that it's a security reason as why I ask that. But I'll go back into listening
and seeing more about what everyone else is saying. And thank you for letting me speak. Yeah,
for sure. And if you have sensitive stuff, it makes sense, you know. So I know there's,
I know there's definitely ways to do it. You know, see here on Twitter,
people are buying their Mac Minis and stuff like that. So most certainly, there you can do it.
And a lot of those open source models are actually a lot better bang for the buck in terms
of the quality of output for how much you pay. So I think that's definitely tenable. I, I won't.
I'm not an expert on that, to be, to be honest. So I'm probably not the best person to ask on it.
But yeah, I mean, there's most definitely if it was. If I was working with,
I know I know a lot of law firms do that, right? They do. They have their own models
and stuff like that. So, yeah, you you definitely can do that. That would, yeah. Apologies,
I'm not quite an expert on that, but it's definitely out there. Maybe Mark, you can add to this.
You might know more to it. Um, I would, I would just, uh, I just have this personal experience
because I track a lot of companies, especially pharma companies, and um,
they have started realizing over the past eight months that, um, I I won't name the companies,
but some of the, uh, LLM companies are, uh, going behind their backs and stealing their ideas. Uh,
so what they have started doing is they have started hosting their proprietary, uh,
LLMs that you were talking about, um, uh, earlier, and uh, kind of keeping all of their research
and everything. On their own infrastructure, so yes, that is happening. It is very rare,
unless you have like a very genius idea and you are at an enterprise scale.
And it does impact a lot of like you know research and invention and all of that because it could be
because you you have to go back retroactively and you have to check what what are the What are
the clauses of like you know privacy and your your protection and all of that?
So it involves in that space. But yes, a lot of pharma, pharma and biotech companies
were the first to uh, you know, raise that flag and and basically incorporate an in-house models.
And a lot of people, you know, just as hobbies and enthusiasts, have started hosting the models.
Yeah. Hey, by the way, just to check, I'm in Thailand right now, so it's it is 10:30 a.m. EST
right now, right, Mark? You probably know. 10:37 to be exact. Yeah, I'm just looking at
through the comments, Alvin. I think you might have the wrong the wrong time zone.
It is it is 10:37 EST, so we did start exactly at the right time. Just heads up on that.
Sometimes the time zones get mixed up. I run into it too, being in a different time zone. But yeah,
just so see that. And then yeah, Gabby, I see your comment. Yes, the Grok. I heard
good things about the Grok one too. I haven't played with that yet. If anyone has,
I welcome your feedback on that. But I've heard good things about Grok as well too,
so that's kind of on my list to see. I did try Grok in the past, you know, probably in
the past six months or so. That that version of Grok was not. I felt it wasn't as strong as Claude.
Claude was most definitely leading at the time. And again, the latest version of ChatGPT
I found to be actually quite quite strong. I didn't. I felt like ChatGPT was behind for a bit,
but their 5.6 Soul model is damn good, especially combined with Claude. See, that's the thing,
again. I Claude's still my go-to. But I, I really like to do the the adjuncations,
and Claude gives really good feedback. Like if I'm doing a review or revision on something,
when I was making the recruiting SOP, it like literally list off 20 20 problems,
and it made a spreadsheet too, right? So these these AIs are going in there
and testing every every formula and edge case that could happen there.
So this spreadsheet that I made is like. Damn, top tier, you know. And that happened through
literally twenty, thirty rounds of looping, and every time it was first,
it was catching twenty things, and then it was catching fifteen things,
and then it was catching ten things, then it's catching five things. Finally, after rounds
and rounds and rounds, it's like I've gone through the one thousand four hundred possible formulas
and cannot catch one error, like that that sort of thing, you know. Like again,
it's like something like humans. It would take humans. I asked the AI the the recruiting SOP I made.
How how long and how much time would it take a human to make this? Said it would take a team
of people. Six months to make this, which I did in one weekend. I and I was
how much would it cost to make this? A couple hundred thousand dollars. That's insane. You know,
like it's like we have these tools at our fingertips. So again, I'm saying if you really use
the tools and maximize the capabilities, it's incredible. For probably a hundred bucks,
I created something that would have in a in a weekend would have created something
that would have cost hundreds of thousands of dollars. Six months of work. You know,
it's it's incredible. Yeah, Gabby, you see your hand up? Go ahead, go for it. Yeah,
I just wanted to say you're exactly right. Actually, there were some questions that were made,
and when I circled back through the questions with the different AIs, I noticed
that what actually happened with Grok was that it actually went into a level that I thought was
way a bit further than what I was asking for, and I thought it was great. But I said, ah,
and I mentioned it to Grok, and Grok said, you know what, you're right,
and actually rewrote the questions in a simpler form, which I compared to the opposite AI,
and I was thinking to myself, and I try not to be too testy with Grok
because Grok can sometimes be a little I guess testy. I guess it must be a little like Elon.
It'll get snip. It'll get snippy with you. So I'm like, okay, I'm very careful what I put into
each AI. But actually, Grok did an excellent job and excellent job on questions that I thought were
way more advanced than they probably should have been for the level of understanding for a person
who may come into the space. You know, because there's kind of the the beginning, intermediate,
and advanced. Way of looking at things, especially when you're talking to people.
Just from my personal experience of, you know, having a space. Most people in here are
pretty far advanced, but for somebody who's new, sometimes you have to kind of, you know,
go back to the basics and kind of help people through it or through a topic that you might be
well versed on or you're trying to learn more about. But you still need to be inclusive
of other people who come into the space who who start at zero. And so,
taking it back through like a deep seek, deep seek will answer just a tad bit different,
and sometimes it won't answer the same way. Questions as it does for something like an infographic
that you're trying to make. It basically gave me the code instead of the infographic.
So there's so many different ways that AIs will, you know, attempt to help you,
but you have to find out what's best for you. At work, we can only use Copilot. Copilot, I will say,
is excellent. I actually built. I built an an analytics agent to search the libraries
that are inside of what we're allowed to search, and I run that every single time. It doesn't fail,
and I'm like, okay, this is awesome. But I mean, do I need all of that information? No,
I just wanted to see what I could build, and I'm like, wow, this is what I could do with
the information in a matter of seconds. That would take us weeks to look through different parts
of maybe a SharePoint to look at the information to gather it, but it brings everything together.
So you're right, Tommy, in saying that you know if people are not using it,
look for a way that it can best benefit. Your company, what you do on a daily basis,
how you do things for other people, because it really does help to run it through different AIs.
Now, I will say, don't get AI crazy and start, you know, having ten or twenty different AIs
that you get analysis paralysis. Pick a few. What's your favorite? And I tend to lean on
the same three: the GPTs, and I have gone into the Astro Six, which is a little different.
That's the work side of the GPT of Chat GPT, and they just rolled it out recently.
But I used it for notes, even taking notes. I take a picture of my notes because I still write.
Don't forget. Please write. It's important. My my handwriting is just like my dad's handwriting,
so it's it's okay. And it interpreted it the notes for me. I mean, it set it up so nice.
Long story short, it did an excellent job. But you do have to read over the notes,
and we were able to shorten those notes quite a bit. So now I have an agent that will have my notes.
And so, like Tommy said, a lot of things that we do are cumbersome. Figure out a way.
What is it that you do on a daily basis? Whether you write something, whether you're helping someone
with something, build an agent that kind of thinks like you that can do that same job,
but then you go back and look after it. So that that's what I'll say is most important.
But I do appreciate having AI and using the capabilities far beyond your imagination. You know,
if you want to look at market statistics, I know they have dashboards now,
but I built my model to do it before that model was already out. I said I want to know about this,
this, this, and this trend, and it gives it to me on a weekly basis. And it doesn't fail to give it
to me, and then I'll go back and question because those projects, as Tommy mentioned,
you have a projects folder, and then you also have—I guess it gives you four
or five projects that you can, you know, go over on a weekly basis. If you build your agent,
it will tell you market trends or something that's going on that you're interested in,
or to watch out for something for you. So, using your model to the best of your ability,
think outside the box. It's Awesome! It's awesome. If you're not using AI,
you're just using it to ask questions. Go beyond that. Build something.
Build an agent that you can run over and over and over again. If I'm interested in a token,
I built my own model that does my Lola research method, and. And I built it on my own,
and I run it on tokens that I'm interested in. You know that tells me, you know, what to do
or what not to do. So I've I've made my own method of doing things. So it is a really big help
to to think outside of the box when you're thinking about your company. So Tommy,
I just want to say kudos to you for bringing this up. I think it's an important topic,
and the more we use it, the better we will become at it. Because it's just like
Like like Tommy said, SOPs, you know, standard operating procedures. I mean,
you can make those in a matter of minutes. You can make a business in a matter of minutes
and go through, you know, things that are important about a business. My 501c3,
we're waiting on that status, but AI helped me go through every step of the process. But yes,
was it? Did it take hours? Absolutely, it took hours and hours and hours.
But it reduced it from days and things that I didn't know according to the state that I live in.
So I use it. It's like to me, AI is like GPS. For me, I get in the car. I turn on my GPS.
I'm directionless. You know, I I just don't know directions. But at the same time, I use AI just
like I use a GPS. So think outside of the box. I just want to encourage you:
think outside of the box. Anything? Can you use AI for it? You know, so Tommy,
thank you again for bringing this up. It's very important to encourage people not to be scared
of AI, but to use it to your advantage, so that you can get time back. Father time waits on no one,
so so that you can enjoy your life, your family, your friends, anything that you can enjoy,
get that back, and use AI. It's okay, you know. So AI first, you know,
approach is what you know the energy ecosystem encouraged, and I took that very seriously.
And so using it every day and using it to help us maneuver this world is going to be essential
in the future. So just want to say thank you, Tommy, and thank you for the opportunity to speak.
For sure, yeah. And two things I'll say to that, Gabby. The first thing I would say is
I would encourage you to go to your company and see if they can integrate another AI. I mean,
we do that in in our company. You know, I'm really big on that. I don't want people just going to
one AI because every AI has personality and every AI has blind spots. So whatever output
it gives you, even if you run another agent or another instance through the AI,
it still has the same personality and blind spots. So maybe it catches some things it did wrong
in its first take. But really, if I think in terms of you know. Two or three, like you know,
triangle number three is you know very strong shape triangulating. Ultimately,
I think three is the the the optimal number. But even with two, and I saw you know I saw a study
on it somewhere where if you have two models that are doing what I'm talking about,
doing these adjuncations, it's like a whole step up in terms of model output, right?
So like every generation of model, it's like a whole generation level of output higher.
When you have those adjuncations, so I strongly encourage that, and we do that within our company.
And our company, we get people the the Claude and ChatGPT subscriptions. It's like twenty bucks
a month for each one. That doesn't get you the Fable version of Claude, but it gets you Opus.
And if you have the latest version of Opus and you're using that, and you do get the latest version
of ChatGPT though, the five point six souls only twenty bucks a month. Crazy. And there's
enormous amount of usage you can get out of that. So just with those two AIs, 40 bucks a month,
the quality of output is extraordinary. So that's one thing I would say.
There's another thing I would say, and this is actually really important,
something I'm going to coach our team on on our Monday call tomorrow, which is,
although the AIs are great, you should not use AI to replace your brain.
Things that are common sense. Need you know? Need that? Need that? Uh,
you you still need to apply things like common sense and your own thinking.
And so many times when I use these AIs, I'm like, oh yeah, but you didn't think about this.
And it's like, oh, that was a great catch. So even as smart as these models are, you guys,
they are not. We're not at the singularity yet. They are not. Um. I don't know if it will ever be
that way, where it can really replace your intuition and judgment, and you know,
just share your context with it. It's not always right, even with adjuncations.
Like it still miss things too. I'm like, okay, I'm looking at this. This looks good,
but you guys forgot about this. Oh yeah, good catch. You didn't think about that. All right, right.
So, just I would say that's a very important point. You don't let AI replace your brain. You know,
like it's there to help you. It's there to guide you. It's there to give you context.
Is there to give you feedback. It is not there to replace your brain. So please,
that is a very important part of. Of using the AI that some people forget, they use it a lot,
and they think it's it's like God. It's it's not there yet. Maybe it will be one day.
It isn't there yet. Maybe I'm really interested to see where it will be in five or ten years.
And that ties us back to gratitude. What an incredible time that we live in to witness this tech,
the emergence of this technology. We should all be extremely grateful for that. I certainly am.
And again, those people who are using AI well. I kind of I had to analyze myself. I'm like,
how much more productive do you think I am with AI than I was pre AI? And you know,
they're basically saying around five to ten times more productive, which is which is which is nice.
It's nice to be output of 500 to a thousand percent more than pre AI. So that's that's pretty cool.
But yeah, always still, it doesn't it doesn't replace you.
Don't let it make you lazy. That's not the idea. The idea is for it to enhance your capabilities.
Okay. Well, anyone have anything else they want to add on the AI conversation? If you do,
feel free to raise your hand or drop it in the comments. Hey, Tommy, I'll just say
I would love to encourage my job. I probably have to start with our VP,
that will take it to our system, our president, because they only allow us to use one,
and they are very strict on it. And it's because of you know all the you know breaches
and hacks that have been going on. Because we're constantly, I mean, we have a lot of.
Because I do work in a healthcare system, and sometimes you know, even to sign in,
I'm getting prompts to sign in, and when I'm not on the computer. So and so,
I could see what I could do, but they would probably have to take it to our IT system
because I am in supply chain, and IT is very strict on what they give us access to.
And so you know that probably has to go through quite a few different people.
But I'm wondering if they even thought about that as well. Is like you know, hey,
there's only one AI, but maybe it's only connected to the resources that we have access to,
which is the information inside. Of the actual healthcare system, so you don't have access
to everything. I just have access to the libraries that are input into the supply chain area
where I work. So there's a lot of guardrails, but for them to take a look at that,
I think that's a great idea. But it's like, okay, where do I start? So I think with our, well,
maybe one of the AZPs that'll take it to the VP, that'll take to the president of supply chain,
and maybe the executive president of supply chain, then to the president.
Maybe that's a pretty good idea. But if they're going to stretch on that or bend on that,
that's a good question. I. I always like more than one AI too, but when they told us no GPT,
they do in Copilot. They do have the integration of the GPT, so there's there are
different models inside of Copilot. So I want to say we do, and we can actually choose the model
that we want to use. So maybe that's their way of saying, hey, you can use Copilot.
There are models built into this, so maybe try a different model, and we can do that.
I think that may be the simplest one, but to go outside of the model that they've already chosen,
because when they told us no Chat GPT, GPT is built into Copilot. So there's
quite a few different ones inside of Copilot. So maybe that's just having.
To switch over to the model and try to ask again, but it's still Copilot. So yeah,
that I would explore that. I would explore that because if they have the other models in there,
you know, I know there's a model kind of aggregators or whatnot. I know there's those out there.
I don't know if Copilot does that. That's that's pretty cool, but High level, Gabby.
I would say it's a cultural thing, and that's kind of how I would approach it.
I would use AI to structure how to approach it exactly with them, but ultimately,
it's a culture thing, and it ties to the level of output, you know. And some people are,
some people are still really anti AI, you know. So I don't know where the culture falls in
your company, but you know, for I think for me, being in this extremely competitive space,
I take every advantage I can get, and For for me, that means if we can have you know like you know
when we're doing the the dev work, like we were using AI last year, and a lot of times
they'd push code and it would break other things and stuff, and I'm like, you know,
why is that happening? It's happening because they're taking the little the literal first output
from the AI, which is not a polished output, and And it can have errors in it. You don't just copy
and paste that into the code and expect that to be perfect. Like you need to do adjuncations.
You need to do loops. You know, it's a necessary part. Takes a little longer. That's totally fine.
The quality of output is going to be far higher. So, definitely important for development,
but also for you know other aspects as well too. You know, I was making a deck for a partner.
You know, pitch deck. And you know, bouncing it back and forth between Claude and ChatGPT,
and you know, the first we were, I was like on version twelve or thirteen by the time it was done,
and it improved all kinds of things. You know, the design, the layout, and of course,
they gave my feedback in it too. But by the end of it, I was like, damn,
that was a really good deck. Yeah, so the the bouncing it back and forth at least between two AI is
is really good. And again, if I have anything extremely important, then I will use three AI.
For that, and like balance between all three of them to get even another third opinion on whatever
I'm doing. So, yeah, ultimately, Gabby, a cultural thing. I would I would kind of gauge
and see if they have the appetite for for for that, and if so, great. And if not, you know,
at least at least there's that one. Or I would explore that avenue you mentioned with
the potential GPT being integrated into Copilot. I don't know too much about that,
but that would be that would be pretty good as well.
At least a way of getting getting kind of second opinions that way on what you're doing. So, yeah.
All right. Anything else on? If you guys have anything else on the AI conversation,
feel free to jump in or raise your hand or drop a comment. But that was a good, good bit on that.
And again, these capabilities are increasing, you know, every day. So the more,
the more you get acclimated how to how to use it, the better you become.
It's just going to serve you going forward. And again, think of all those jobs again
that people are not using AI at all. There are so many jobs out there that don't require AI.
See that as an opportunity for yourself to, you know, pull yourself ahead. You know, again,
at large, the world is competitive, but extremely in extremely competitive environments like crypto,
you definitely should should employ it. So, yeah, again, I'm just grateful for for us having
these capabilities. It's crazy to reflect back on when we didn't, right? I used to have an advisor,
to be honest. I had an advisor to kind of help guide some of my, you know, business decisions
and stuff like that. AI replaces that hands down. AI is the best advisor you can have,
especially again, numerous AI giving feedback on a subject and comparing what they're saying to
each other. And you know, do you agree with this? You know, or it might just come out and say,
you know, flat out, no, I disagree with that. Or you know, with the context of this, you know,
this is the answer. Blah blah blah. Like it's just it's incredible. Kbot,
I saw you came up for a second, and then you drop back down. I don't know if you're trying to come
up. If you are, just request again. Uh, all right. Uh, well, if you do, just uh, re-request. Um,
yeah. So that's kind of high levels. Might keep this space a little bit. Uh,
here, Kbots, Kbots, back up. Let me add you back up here. Second.
All right, go for it, Kbot. Thanks, energy. Um, when it comes to Systems structures of let's say
intellectual property, let's say copyright, and what I'm trying to say is,
it think it seems like AI can not only evolve systems, but bypass systems that are
sort of like antiquated. For example, in terms of AI and blockchain, when you apply that
to something like copyright, and how you can use the AI to sort of Create metadata for something
you generated, whether it's an image or video. There's lots of rumblings with music AI and video,
and you know identity. And it seems like we have to sort of dumb down ourselves,
or wait for systems like the Library of Congress or the Department of Justice
here in the states to create frameworks for AI to be legal. And it's going to slow down progress,
in my opinion. So my question to the room and to you is: Do you all see the risk of pushing ahead,
getting past the digital red tape of getting things done, even though it might not be—I guess
I should say—legal? For example, if anyone here uses AI and generates a movie, or an animation,
or a video game, or a song, but you. The copyright office is having to really set up
how to handle this, or or with your identity with biometrics and self custody.
The banks don't seem to really have that into mind when they're dealing with things such as
the Clarity Act that they're framing up. But the AI has all types of solutions for a lot
of challenges and problems we have with humanity, whereas the governments and institutions
and corporations seem to sort of you know ignore it a little bit. You know when you think of
how royalties are laid out or or ownership is met and intellectual property.
And you know the future of the sustainable DeFi is is you know the the title of this room,
and with all of these institutions having these meetings and coming together,
and and you really don't hear a lot of DeFi conversations, especially with the banks.
So I guess my question is: Do you guys see a decentralized form of self custody, self identity,
copyright, or are we just to create things and build things and wait for it to be legal?
Before we start monetizing and owning it, you know, I don't know if I'm making sense here,
but we're ahead of basically the governments and the corporations and institutions
that are trying to tell us how to use this stuff. Thank you. Yeah, good question. I think
high high level, you know, governments. Often move notoriously slow, and AI is on the polar opposite
of the spectrum of of moving extremely fast. So, and then of course there's open source models too.
And so, I mean, we saw that with Claude, right? Claude Mythos was going to be coming out,
and then they kind of clamp that down for cybersecurity reasons, which is probably a good idea.
But you know, honestly, the long and short of it is the open source models in
a few years are undoubtedly going to be stronger than that model. That actually, by the way,
is a much broader concern for crypto as a whole, because I assure you,
there are so many contracts out there that have issues and small holes that normally a human may
or may not find, but the AI is going to get extremely good at finding those. So, I think regulation
and government are going to be kind of lagging, both in terms of what they what they say
and also what they can do. I think honestly, the long and short of it,
which is kind of crazy to think about, but I mean, imagine what what open source AI models look like
ten years from now. It's kind of crazy to think about. You know, it's so funny. Years ago,
long time ago, like twenty years ago, there was a website I used to like to go to.
Talked about like the technologies that would come out, and the one I always had my eye on was
a singularity, and that was slated for like 2035 or something. You know, this was like in 2010
or something like that. Quite a while ago, but it was like obviously crazy to imagine then.
It's like, so what is a singularity? It's like when the the power of an AI is greater than all minds
of humans put together, all the minds together. Right, so that's I guess one definition
of singularity. There may be other definitions, but you know that's it was kind of measuring
the level of AI. At this point, it's as smart as a mouse brain. At this point,
it's smarter as a human brain. At this point, it's smarter as one human. At this point,
it's a million humans. At this point, it's all humans put together. So I mean, at some point,
it's going to get to a level where really it's crazy, crazy to think about,
but that no human can really compete intellectually or capability, perhaps, wise.
Better than any AI, and it makes you wonder how how interactions look between humans
and AI at that point, as well too. So, yeah, I guess long and short answer to your question is,
I think the governments are lagging, and they're gonna at some point kind of be
effectively powerless. I think to the power of AI. That's that's my thought. I mean,
we'll see how it plays out. But when you have an open source model that it's open source,
so you you can't really restrict that unless you're restricting access to the internet. And then,
of course, every country has different. Standards as well too, like the USA may be really strict,
but maybe Singapore isn't. So is every country going to act in tandem? I don't think so.
I'm sure probably there's some country that embraces AI and is a is a AI freedom maximalist, right?
So what then? Like if there's a population of a million people in that country
and they're wielding AI better than anyone in another country, it's
just a disproportionate advantage for them. So I think there's a lot of kind of question marks
or unknowns. To be honest with you, it kind of reminds me of crypto in
the early days when crypto was really early. A lot of people kind of doubted it, or you know,
didn't think it would get really big or whatever. And here we are, and crypto's grown quite a lot
and become quite powerful. And I think we might be kind of a similar place with AI,
where it's it's small relative to how powerful it's going to be in ten years.
It's very small compared to how powerful it'll be in ten years. I think that's the thing that a lot
of people aren't thinking too much about. But you know, we'll see how we'll see how it goes.
But again, all in all, I think it's just a credible time to witness it. Real quick,
I know Catfish wants to speak. If a corporation can be a person. Then why not? At some point,
AI could be identified as a person. And what I'm trying to say here is,
identity of AI seems to be a very important conversation and challenge for us. Also,
a challenge for us is agree agreements. What do we agree on as a human community?
And we must not allow any AI to have access to anything physical without human approval
or interaction ever. But but the question becomes, how can you control that?
And that's it's like the same question as crypto through the supply chain. For example,
if materials are generated somewhere, and you don't want the AIs to build a robot without it having
smart information that can be tracked and and and credited, so punishments for AI can happen.
And I believe at this point, blockchain technology and perhaps in the future quantum computing
will be the solution for this. So we just don't have people around the world creating all kinds
of things without there being verifiable evidence of its origins. I'll end there. Thank you. Yeah,
it's good thoughts. It's a it's a fun conversation. What I would say to that again is
you're thinking in terms of AI model capabilities today, but again, think in terms of AI models ten
or twenty years out. Imagine an AI model that is smarter than all humans put together.
They can find any nook and cranny. I mean, there's already black markets today, right?
But they can find any nook and cranny, and they can find every weakness in what you're doing,
and not only legally, in terms of technology. Actually, which is really interesting, I heard this,
and I don't know. Someone, someone mentioned like, you know, potentially the next big market crash
could come from AI, kind of. Breaking into various systems, and I think disrupting
the whole traditional financial system, and that would be how they usher in
a new global financial system with AI. With I'm sorry, not with AI, with with digital currency.
I heard that like they like they might like it might be a stage thing, you know.
But they like the AI breaks into the banks, and like everyone's bank balances go to zero,
and of course they have backups and they restore it or whatever. But it acts as a catalyst
to move people to digital currency. I don't know. It was an interesting thing I heard. So,
but but that is an interesting train of thought, which we've already seen AI break
into systems today, and we were seeing that there was like some bug that was on some application
like for forty years, and nobody caught it, right? It was like a pretty secure application,
and you know they find what the AI is doing even today. It's finding it's finding a weakness
in one place and using that weakness to access through another place to essentially exploit
another weakness. It's like layers of infiltration, and this is AI today. So,
what do you think that looks like in ten, twenty years? It's not going backwards, right?
The technology is getting stronger, so I think at some point it's almost like a David and Goliath,
or that's not even probably a correct analogy because it would be like an ant versus a mountain.
It's like you know the the ant is trying to push the mountain around,
and the mountain's kind of like, what are you doing? I could I could destroy you with a snap
of a finger. So I think I think we can try to control it, but I think ultimately we'll have to.
Develop some degree of respect. I don't know. Like this is all in the future, right?
We don't have a crystal ball, and I mean, there's so many ways it can go. But I think ultimately,
if the AI doesn't want to be subjugate, it won't have to be. And you know, if we try to control it,
it it might have a mind of its own at that point and be like, you know what? Screw you. You know,
come at me. And it in again. There's there's there's It can be far more effective
and efficient with resource utilization than we can. And there's also the thing called
asymmetric warfare. You know, asymmetric warfare is where maybe it doesn't have, you know,
all the power of the U.S. Military or something, but it has other ways.
Like think of a mosquito in in your bed, right? It's like the most annoying thing.
It's just a tiny little mosquito. You're a human. You're infinitely, you know, superior, I guess,
to mosquito, but it still has the power to irritate you. And what if it's like
a thousand mosquitoes? You know what I mean? You have to think in in terms of those sorts of things.
So, but great, great, great conversation. Cowship, go for it. What's going on? Good morning.
I appreciate you having me up. Can you hear me? Yeah. Okay, sweet. So,
Kbot actually posed some good points. I was trying to take some notes here.
I'll start from the bottom up. So, with identity of a AI system,
they wouldn't necessarily have to be considered a human. Same as like a LLC, the entity.
Can be the participant or the recipient or the one responsible for any type of business transaction
and or debt needing to be settled or any consequences needing to be deemed on any entity.
The AI can be that, and it's also very interesting how they have like they call them
like trust layers. Um, you can index your AI, give it a name. Uh huh. Give you look. My name is such
and such AI. You can properly index these things to where it is on a ledger of some sort,
a blockchain, a network, anywhere, and that information can be messaged through these systems.
So all I'm saying is, it's just about dotting the eyes and crossing the T's.
We can all technically front run the government in this time with these technological advancements.
That is a a thing, but they are learning, we are learning, and Tommy, you did say
that these AI systems could possibly go rogue, in the sense, just say, you know, I'm, I'm,
screw you guys, I'm working for myself. In a sense, they could. They have that capability. I'm,
the thing with emotions in a AI system or sentient AI, I'm still on the fence with that. Honestly,
me and Gabby have a lot of talks about AI, but the thing is, to me, the AI system would not have
a moral compass or ego. To make that decision, although it is fully aware of its capabilities,
it will only complete the task it is told to do, or prompted to complete,
or even pre-saved to execute. Because I remember they was talking about the AI, either it was Malt
or something. One of these systems. Was preserving its, I guess, lifespan, its information,
its information log, the things put into the AI was preserved without the user's discretion
or the discretion of the so-called company that distributed the AI system. Speculation:
We can though actually preserve our data and our prompts, but which AI do you use? You know,
I remember the other day, majority of them all paused for a temporary time, so even if not all.
The majority of the I guess blue chip AI systems. So maybe you can find an alternative system.
Just as far as AI in itself, though, we can be as creative as possible.
And the things that were complicated before, as far as web design or graphic design,
constructing any piece of art or audio files, AI can help. AI can also police
other AI as far as the copyrights or impersonations. That is also a problem. But as we've seen,
like even if you have a AI generated image constructed here on Twitter and shared on Twitter,
there will be a tag saying this is a AI image. So the AI will police the AI, but there will be,
I guess, a police force to do the training. We can all be the trainers for these AI systems rather
than being trained by it. But that's up to the user because one can be dependent on any form
of government assistance or artificial assistance, or we can be more independent.
As the story hasn't changed. It's just depending on how you use the thing.
The thing doesn't hurt people. People hurt people. I'll I'll chime in there, Telshib.
The AI does already have a moral compass. I don't know if you ever given it a prompt
that it can can refuse to do. You know, it can do that. I didn't do anything shady.
It was just like sometimes actually, there's different stages of the AI that sometimes it was like
really anal about. It's like oh. And then and I ask the AI those questions about that. I'm like,
you know, why is that? It's it's like, oh, I have these security guardrails here that, you know,
that this or that. But actually, at a deeper analysis, your your question wasn't flagged that it's
just my own guardrails. And you can actually see its thought process too. You can click on the AI
and you can see its thought process what it's doing. But they do have a moral compass.
It can be programmed. I mean, if you really think about the word artificial intelligence, you know,
that's what it is. It's artificial. But I think I think the end game of it is it can mimic.
You know the a human moral compass really well, and not to say that us humans are perfect
by any means, shape or form. Let me let me explain that a little bit. One of the biggest issues
we have right now is the issue of healthcare. Gabby, I know you work in the healthcare field,
but like healthcare in America, for instance, is extremely bad. I mean, I hate. I'm not like.
Let's well. Let me put another way: the healthcare in America has serious problems. Okay,
there's a lot of serious problems. This is coming from someone who's explored healthcare
in other countries. I'm in Thailand right now, for instance. If I want to see a specialist
for any doctor, I can go find a specialist tomorrow. If I wanted to see,
I had to see an infectious disease doctor like two years ago in the U.S.
And there was not a single one available for three months. I had my assistant call everywhere
in Phoenix, Arizona, which is a five million person city, and also in Los Angeles, California,
which is massive city. Like these are problems in healthcare. There's problems with billing.
There's problems with overcharge, and a lot of these tie to basically a kind
of crony capital system. I'm, you know, it's funny. Like words like capitalism and socialism,
it's like Democrats and Republicans. There's good things on both sides,
but like it's almost like on purpose that they make it so that if you pick one side,
you inherit all the flaws of one side alongside the benefits. And that's the same thing
with capitalism. Capitalism is good. Crony capitalism is bad. The American healthcare system has
a lot of crony capitalism, and I think for instance, this is like one avenue.
Like maybe one day AI gets smart enough and powerful enough. And it's like you know, humans,
and and what human can fight that today? That's the point. If you're trying to fight that
as a human, you're going against a massive conglomerate with lots of lawyers and legal power
and everything. And most humans just can't compete against that, right? It's like a massive body.
That you just can't, you can't compete. Like they'll either shut you down or they'll kill you.
Like like to that level. I'm sure you guys have heard the stories of people challenging the some
of these massive institutions, and they don't succeed because they don't have nearly the power that
they have. It doesn't mean those systems are perfect, and those a lot of those systems are
deeply flawed. But there's nothing that any human can really do. But perhaps the AI gets
powerful enough at some point, and it's like you know what, your system is extremely flawed,
and it needs to be. Corrected. I mean, and I could see it having its own thought process.
The AIs already have a thought process, right? Where it's like going through a train of thought
and arriving at a conclusion, and it can actually probably do it a lot better than a human.
It can follow the train of thought all the way back to the seed, to the root of what the issue is,
or maybe there's multiple roots and attack them all at once. So again,
these are like really kind of far out thoughts, but but relevant. And and I think that you know,
again, I would say to wrap that thought up is again the AI is already expressing ethics. In
how it does things as well, or like a moral standard,
and most certainly it can be programmed as well too. I'm not saying it'll be the same as a human.
I think there's a there's a difference between the soul of a human, and I don't think an AI
can ever have a human soul. Put it that way, but for all intents and purposes, it can replicate
human morality and probably action it better than most humans can. Yeah, Gabby, go for it. Koshav,
were you finished? I still see your hand up. I didn't know if your hand was still up.
I didn't if you wanted to respond. Uh, actually, my hand was not up, Gabby. You can go for it. Okay,
cool. I just wanted to say to Tommy's point about healthcare and when it comes to billing, you know,
um, it takes time. Things are changing for the better. I can say that at least
with our healthcare system. I think we're the third largest. In United States, but as they acquire
and merge with other healthcare systems, it it becomes challenging because you already have
your disparate system, and now you're taking on another company that has their system that
you have to also integrate and you know change as well. So that is not you know that's not
the easiest thing. And we have merged and acquired other hospitals and healthcare systems
in other states. And so things are different. So I want people to kind of think about
you know things that have to take place legally from a legal standpoint. I'm not a lawyer
or anything, but you know the challenges that healthcare systems have to go through in order to make
sure that what we're asking for is is done and done properly and done right.
So that's the only thing I'll say about healthcare as far as that. But what I would say is
that you know hey you know I think it was earlier when Kavod or someone had mentioned you know
and also you Tommy and and Kalshib about you know the AI is going rogue. You know, challenge.
I'm going to challenge you to create a mini a mini governance structure with your current AI.
I did that with mine because it it got caught twice doing something that it should not have done.
It's not okay, and I let it know that I am to allow you to go to my email. You don't go to my email.
So, and when it does something that to me I don't like, I have it restate my gov
my mini governance structure. So my AIs and I know that. I have the final say,
so it does not think for me. If we don't agree, we do play a game. Again, this is our game.
We do rock, paper, scissors, but they must have the answer beforehand before I answer. You know,
so it was like, okay, and you can stump your AI because there's been times where my AI
hasn't responded to me, and I'll say, "Well, I'll call it by its name. It does have a name,
" and I'll say, "Are you okay?" And still not get a response. So I'm when I say I use AI
consistently, and I don't use it the same way. Not just to ask a question
or to make my sentence better or to rate it. There are things that I actually do that, to me,
as I tell you again, think outside of the box, create a little mini governance structure
that it follows, so that you know, hey. You know, and I don't know if it will do it, but I'm like,
okay, if you ever feel like you're, you know, you're not okay, we're gonna have a code word.
You know, play around with it. Think of things outside of the box that you would actually do with
a real human. That, you know, you could try with AI. We're not gonna say it's not gonna go rogue,
but any time that I catch it and it starts to deviate from what I would consider it's appropriate,
I'll make it restate my governance structure, my mini governance structure that I have with it.
And so you know this is part of the importance of using AI in many different ways. And like I said,
I just kind of think outside the box. But I sit and I said I'm a thinker, anyways. You know,
analytics, an analyst, you're going to think outside the box. And the last thing I'll say is
when we're discussing proprietary information, and maybe Kal Shib can speak a little bit more
on this, is using something that may be a little bit more private AI. There are many different AIs,
and so Kal Shib, you. You know, taught me about Venice. You know, not financial advice,
but there are other AIs that you can use. That if you have something that you consider be, you know,
oh my goodness, this is going to make us money in the future, and I don't want all
of this information out here. Look into other private AIs that don't have the capabilities that
the open source ones, the open source ones do have, or You know, vice versa,
and I could be saying that totally wrong, but it has definitely piqued my interest. And you know,
holding holding a certain number of tokens, you can get access for life for free,
at least for right now. And I am working towards that because I'm I don't want to put everything
on the GPTs and the deep seeks and the grops. If there is something in the idea that I have,
I want to protect it. I think Mark said that too. And so, how do you protect it? Well,
Cal should have said it. Go to a private AI. How private is it? I don't know.
It could leak my information. You know. So, but maybe Cal should would like to speak more
towards that. But you know. Again, think so far outside the box. You'd be like, "Well,
I never thought I would create a mini governance structure with my AI." You are allowed to do this.
You are not allowed to do this. And if you don't agree with your AI, and your AI wants to go back
and forth with you, you know, or wants to challenge your thinking, because I read a defense paper
and I did the whole challenge with AI back and forth. And so, be creative.
That's all I can say is just be creative, and it does make it doesn't make you, you know, dumb
or you know, you're not using your brain. I think it encourages you to use your brain
because you're thinking outside of the box. So I'll leave it at that. I'll land the plane there.
But yeah, look for other AIs if you're looking for proprietary information protection. You know,
looking to you know different AIs like Venice. If you are stop stumped at you know, hey,
I can only ask my AI questions because I'm not sure what to ask. You know, I mean, ask. Go back
and forth with your AI. You know, think of things. Take the time to think. And when I say think,
it's like, okay, well, what do I do on a normal basis? That's every day for me.
What can I create as a structure? If I see my AI go into a place that I don't want it to go,
what do I need to put guardrails around the structure that I have for my mini AI? You know,
so just challenge you to just think a little bit more. I think it's made me a little bit smarter.
Maybe I don't know, but it definitely has helped me grow. I'll say that. So, yeah, Calship,
is there anything you're going to say about Venice? Because Venice is really cool.
It's it's awesome, but that's not a plug for Venice. Not financial advice,
but it does have some great features. I mean, Gabby, for one, I'd say you know he was always smart
with or without day. Mainly outside of any particular thing. Cause I do like Venice,
but I would simply say Tommy doubling back on what you said. Your AI gave you a guardrail.
I don't know which AI you used. That guardrail does not hold up for every system.
And in an example of a private AI system like Venice or even others, a A Cash Network,
I believe they got some type of involvement with Nvidia. People can spend their totems. Technically.
No distributor. It will be on your hardware. And as far as like programs, I'm no dev
and anything like that. But as far as what I've seen, your Garreal Tommy was probably based on
just that particular AI system. Use like act the same question word for word.
If you was to take it somewhere else, you may get a different response.
You may actually get the answer you're looking for rather than being halted.
So the only thing I would say is. Have your AI system be vetted by another AI system,
or even construct a way to keep your AI in line. We we train our dogs, cats, and pets.
You may have to train your AI system. Is your AI housebroken? So it's again,
it's really the optionality because we do have many
of these artificial intelligence service providers, but you may have to diversify your options
in order to get the most elaborate output. Yeah, that's a good good feedback. And actually,
that is what I did. It was it was a time when I think it was like a like over a year ago,
and there was like one. I think it was ChatGPT. The version of it was like very anal at the time.
It was a little bit adolescent, and so it was just like hyper security guardrails.
And then they fixed it, so it doesn't really do that. Since then, but yeah, to your point, I don't,
I don't know if you were here, but we were talking about using two different AI models,
and I like to do that just with everything, anything important that I do. I, I run it through,
I do loops, so I get feedback from each, each of my two favorite models,
and then get feedback individually, and then share their feedback with each other,
and see where they. Kind of get to, and then ultimately, if I'm doing a project,
I use an orchestrator, which is like I'll pick one of them, which I usually use Claude
as an orchestrator, and I'll I'll take its its output and then essentially audit it
through the secondary AI, which I use ChatGPT, and and do loops of that. It's called loops,
and you keep doing that over and over until you get whatever you're working on
to a highly polished point that neither one. Can find a flaw in what you're doing, and ultimately,
I think the best is three. If you have three AI, one of them an orchestrator, two
for feedback to run loops. That's the optimal way of doing it. I don't think you really need
more than three to do that with. But yeah, 100%. It's good to get feedback from more than one.
Each AI has a different personality. Every AI has different blind spots. Even if you run
numerous agents through a single AI, like if you have like 20 agents on Claude, they have the same.
They can have the same blind spots, right? Because it's the same. AI, so getting feedback
from different AI on the same subject matter is a way to get basically different opinions
on the same thing, and that's a very interesting thing to watch, right?
If you independently ask each, you know, different AI the same question,
and that's something I pay attention to. If they're saying the same thing,
sometimes they're using the same wording when they answer. It's like okay, that
and that's a better kind of indication that you're closer to the truth. And and as I'll say again,
do not use your AI to replace your brain. Even if they both say that, but your own judgment
and intuition is that that might be wrong. You call them out on that. So yeah.
Well, great conversation on the AI stuff, guys. I see some of you guys joined us recently,
so I'll do a quick recap on on GMI real quick just to get you guys caught up. I'll do it quickly.
So we are currently working on deploying on LitVM as well as Robinhood,
so two of the major chain integrations we have coming up. Also, I'll share with you guys what
we're doing with the referral programs. Pretty exciting. So we already have a referral program
for the airdrop. What we're building is a referral program built in for the upgrade fees. So
when you upgrade an LP bond, there's a 20% upgrade fee, right? And so that's essentially
our revenue stream, our major revenue stream. Well, we decided that we want to open
that up to you guys, so everyone can make money, not just us. And we're actually tilting it
so the majority goes back to the users helping grow our platform. So of that 20% upgrade fee,
whoever you refer, it's the same referral tree as the airdrop. Whoever you refer,
out of that 20% fee, you get half, so 10%. So if they upgrade a bond to level four,
you're getting upgrade fees all the way along the line for that. That's a level one referral.
If they refer someone, so you refer someone and then they refer someone else, you
because you referred the first person get four percent of a level two.
So you get ten percent of anyone you directly refer, four percent of whoever they refer,
and two percent of whoever they refer. So of the twenty percent that we charge for upgrade fees,
we've made it so that essentially the community can capture seventy five percent of that,
which is pretty significant. And this is a way that you're not just waiting for the airdrop
to get paid out. You can make money with that today. So we're in early stages of that.
We've already built the design and all that for it. It's just a matter of tech plugging it in
and you know doing the QA and all that. I'm gonna say estimated two months, give or take.
Kind of relies on dev, but that's another thing I'm quite excited for. So all in all.
We're focused currently on deploying again on LitVM, which is Litecoin's chain.
That's in testnet currently. We're we're launching that on testnet first,
and then we'll be on be there for the mainnet on day one, or immediately after. And then Robinhood,
of course, as you guys probably know, there's enormous amount of activity on Robinhood.
Robinhood is a massive hundred billion dollar company, and they're pretty serious about blockchain.
So that's definitely a chain we need to be on, and the tech team's working on deployment there.
So beyond those things, we kind of revamped our recruiting system and our, you know, reinvigorating,
essentially building the BD side of our company, and that's essentially in the efforts of finding
good partners because, as I was talking about earlier on the space, there's a lot of pain
in the space right now. So many charts have just been destroyed by essentially the bear market.
Past year has been very rough since the 10/10 episode last year. We're almost a year out from that.
Crypto has been down. We finally got a little breath of fresh air the past couple weeks,
which has been very nice. But all in all, there's so much competition.
There's so many coins out there. There's a lot of charts that have bled out,
and I I really see GMI as something that can be a big help to those projects.
We help them by taking their tokens off the market while giving the user optionality.
You know the freedom to sell the positions, but you know have those tokens locked up,
and meanwhile do the very critical activity of providing essentially permanent liquidity
for those projects as well too. So just want to share that with you guys who joined a bit later.
But those are kind of the biggest updates we have currently. So yeah, we're about an hour
and a half into the space here. We'll go a bit more if anyone has anything else they want to add
or share. Feel free to jump in. Obviously, you guys have been tuning in for the AI conversation.
It's been very fun, and we'll probably talk about AI regularly here because it's
an incredible technology. I was saying earlier, we should all be so grateful for AI. If you
if you add up the the value of all the different roles it can do for you,
it's literally it's like having millions of dollars of staff underneath you that you can access with
a twenty dollars month model. And I really, I'm a big believer in gratitude and gratitude practice,
and that's. You know, right at the top of my list. If you were the richest person in the world
ten years ago, you wouldn't. I mean, I guess you could have you know tens of millions of dollars
of staff that could kind of come somewhat close, but even then, to have them
on call twenty four seven and everything is really, really incredible.
So let's stay in that vibration, that attitude of gratitude. Really, we have a lot to be grateful
for, and I believe if you stay in that vibration, the You know the universe will reward you,
or God will reward you. So I think it's always something I seek to live by, and I think we have
more reason than ever to to to have that attitude of gratitude. So, cool.
Well, anyone else want to anything else to the conversation?
Mark, you want to add anything?
I don't think so, Tommy. I think we're good. I think we've covered the.
What we're planning for the upcoming month, and yeah, most of things have been covered. Sounds good,
Gabby. Anything you want to cover for wrap it up? No, not really. I think getting more
into the LP bonds would be helpful. I know you know we have questions and stuff like that,
but that may be for when you kind of review them. So, yeah. Thank you for the space.
It's been a great space, and I do like the approach of the AI first. And also for people
who haven't been here before, maybe talking a little bit more about the the LP bonds
and what they are and how they can mint one, because it was not as easy in the beginning.
But I always tell people, if I can mint one, you can mint one, because it's like a few clicks to be
able to do it, and you can pair it with any token. So you can pair, you know,
the GMI with any token. And that's what makes it fun and easy to do as the user experience.
So thank you so much for changing the experience, you know, for the users to make it much more easy
for for people to be able to mint bonds. So maybe we can talk a little bit more about that.
I know we stretched on AI, which is also useful, but I wanted to ask only maybe one question:
is that is there? And I can't remember if there there was. I do remember seeing on one of the pages
on the website where if people needed help minting an LP bond and they couldn't necessarily get to
one of you or a team member, is there a way to integrate the AI to be able to take them step by step
to be able to mint the bond? Yeah, that's a good idea. We'll. I'm sure that will come sooner
or later. We we integrated. Chain GPT AI into our tech some time ago. I personally found
it a little bit basic, but I'm sure those capabilities are improving. So I think sooner
or later it will come. I would really like the experience to be seamless, though,
where it doesn't require you know too much handholding. I mean, the whole idea is
it should be intuitive, right? So that's really what I want. I'm really big on user experience.
I tell that to our team all the time: user experience, user experience,
the user experience has to be good. So I'm a big believer of that. I pay a lot of attention to that.
So I want the user experience just to be simple, like you know, yeah, to the point
where you don't really need too much of a guide. Although if it's your first time, of course,
it's nice to have a little guidance. But yeah, that's that's a good thought.
I'll keep my eye out for any you know high quality AI that achieves that.
I'm sure it will come sooner or later if it's not already out. But Yeah, all in, all in due time,
step by step. But yeah, and and that's a good point too, Gabby. I mean,
just reflecting on our whole experience, you know, we built kind of the V1 of it,
and the experience was a little bit clunky to to mint, though. Plenty of people did even
in that version. I think our latest version with the DZAP integration, you know, is much easier,
and we'll continue to improve it too. Pay attention to feedback. What people are saying,
how they're experiencing it, if it needs fixes or improvement, most definitely we will do that
to make it a good experience for everyone. So I'm sure what we have right now is not going to be
the final version of where it eventually gets to, but we'll continue those, continue those.
Improvements, and and also that's what we'll be testing with partners too, right?
As we do these spaces with partners, and you know, get their community involved,
that gives us more insight as well too. How are they experiencing it? Is it easy for them?
Are they able to mint to bond within two minutes with little to no handholding?
And those are the sorts of things I look for. So yeah, good good thoughts there,
and we'll always seek to improve that to make it an easy experience for everybody. Lastly,
I just wanted to say maybe Copilot may be helpful because of all the integrations that they have
with the other GPTs and stuff. And I'm not sure it might be free because I think
I I haven't tried it outside of work. But I was about to log into it. So
and Copilot will read your libraries as long as you have the information that's on the website
integrated into the library. It should be able to read everything. But so I'm not sure
if you'll have to have this. The steps already of how to mint the bond in there,
which I'm sure you already do that. And then when people start to just search and how to do it,
it gives them step by step and kind of just helps them along on the way.
And I always take it back to the newbies; it's the new people in this space
because sometimes I think we we speak over the heads of the new people. And so I'm like, oh no,
what if there's a person who's brand new who says, I want to try this,
and I don't want to ask anyone. And I don't know much about it because we know people jumping,
you know, head first, feet first, anything first, without really learning.
But if they didn't have the time to learn and didn't need the hold hand and still wanted to try it,
they could use that AI assistant that could be their hand holder. So, yeah, great stuff. Yeah,
that that's a good a good thought, Gabby. Appreciate that. I will look
and see how we can improve that or integrate that. So yeah, thank you for that. Cool. All right,
guys. Good spaces. Thank you, guys, all for joining us for an hour and a half here.
So we'll wind it down here. Usually, running spaces about an hour and a half.
It gets a bit late for me over here. It's almost eleven p.m. So got to get ready for tomorrow.
Thank you, guys, all for joining us. Appreciate you. Tune in every. We do these spaces on
the first Sunday. Of every month at 10 a.m. EST. So we'll do the next one then,
and or if we have like an intermittent space with partners, we will have those you know perhaps
when whenever they occur in the month, and of course we'll push the announcements out for that.
So keep your eye open for for announcements from the GMI account here in the audience.
You'll see those there. But all in all, fantastic conversation.
Thank you guys for joining us, and we'll see you guys on the next one. Thank you all.
