The Wolf Of All Streets - FTX Lessons Learned | What Brett Harrison Is Building After The Collapse Of FTX
Episode Date: April 25, 2023I sat down with Brett Harrison, the former Head of FTX US, to talk about his latest venture, Architect. In this episode of the Wolf Of All Streets, Brett shares insights and lessons learned from his t...ime at FTX and how they influenced his new project. He delves into the inspiration behind Architect, a platform aimed at revolutionizing the world of financial infrastructure, and how it solves existing problems in the industry. The conversation also touches on the growing role of AI and its potential impact on the financial world. Tune in to discover the innovative ideas and perspectives of one of the most prominent figures in the cryptocurrency and fintech space. https://twitter.com/BrettHarrison88 ►►THE DAILY CLOSE BRAND NEW NEWSLETTER! INSTITUTIONAL GRADE INDICATORS AND DATA DELIVERED DIRECTLY TO YOUR INBOX, EVERY DAY AT THE DAILY CLOSE. TRADE LIKE THE BIG BOYS. 👉 https://www.thedailyclose.io/  ►►BITGET GET UP TO A $8,000 BONUS IN USDT AND GET MASSIVE DISCOUNTS ON TRADING FEES! 👉 https://thewolfofallstreets.info/bitget   ►►NORD VPN GET EXCLUSIVE NORDVPN DEAL - 40% DISCOUNT! IT’S RISK-FREE WITH NORD’S 30-DAY MONEY-BACK GUARANTEE. PROTECT YOUR PRIVACY! 👉 https://nordvpn.com/WolfOfAllStreets  ►►COINROUTES TRADE SPOT & DERIVATIVES ACROSS CEFI AND DEFI USING YOUR OWN ACCOUNTS WITH THIS ADVANCED ALGORITHMIC PLATFORM. SAVE TONS OF MONEY ON TRADING FEES LIKE THE PROS! 👉 http://bit.ly/3ZXeYKd ►► JOIN THE FREE WOLF DEN NEWSLETTER, DELIVERED EVERY WEEK DAY! 👉https://thewolfden.substack.com/  Follow Scott Melker: Twitter: https://twitter.com/scottmelker  Web: https://www.thewolfofallstreets.io  Spotify: https://spoti.fi/30N5FDe  Apple podcast: https://apple.co/3FASB2c  #Bitcoin #Crypto #trading Timestamps: 0:00 Intro 0:42 Architect 5:16 Access to CME 7:50 Who is Architect for? 10:00 Innovations in crypto exchanges 14:55 What’s left for Americans? 17:30 Regulatory clarity 19:00 Collapse of FTX 21:00 Compliance 26:10 What do we lack in DeFi? 29:00 Data provision 30:50 AI 34:30 Analytics x AI 38:20 Final thoughts The views and opinions expressed here are solely my own and should in no way be interpreted as financial advice. This video was created for entertainment. Every investment and trading move involves risk. You should conduct your own research when making a decision. I am not a financial advisor. Nothing contained in this video constitutes or shall be construed as an offering of financial instruments or as investment advice or recommendations of an investment strategy or whether or not to "Buy," "Sell," or "Hold" an investment.
Transcript
Discussion (0)
Every American investor, retail, institutional, or otherwise learned quite a bit from the contagion
of 2022. Brett Harrison, who was the former CEO of FTX US, arguably took the brunt of that and
learned a lot more lessons than the rest of us. The good news is that he's building something
on the lessons that he learned, which is Architect, a platform that's going to allow
people to access both traditional markets and DeFi and offer the plumbing for fast orders, execution,
and high-level trading. Really, really interesting. Can't wait for you guys to hear the conversation
we had about it. Let's talk about what you're building now. What's going on?
Sure. So we started this company, Architect Financial Technologies in January. And at
a high level, Architect is a set of front-end and back-end technologies that are trying
to eliminate market structure complexity, not just in digital assets, but also in regulated
derivatives and similar products.
And this comes from the experience of having been inside of trading firms for more than a decade,
seeing a lot of the clients of FTX struggle with things like connecting to many different
exchanges, the DeFi protocols, to custody platforms, and realizing that all of these
companies have to basically build the exact same infrastructure. And it's wasteful.
And instead, there could just be a company that kind of builds this for everyone to use
so that they can immediately access all these different markets in a unified way,
whether that's through manual trading or API trading.
And so that's what we started building in Architect.
And so does that allow people to hold their own keys, control their own wallets?
Is there any custody side to it at all?
Or is it basically just that plumbing?
Yeah, very good question.
So there are a lot of different kind of competitors out there that do similar things to what we're
describing in terms of just sort of straight connectivity.
And one issue that we have with a lot of those platforms is that they are a hosted service.
And I'm not sure if you would go so far as to call it custody or not, but often you have to kind of give them sensitive information,
like whether it's an exchange API key or a private wallet key of some kind so that they can send transactions on your behalf. And we think that really opens up an attack
vector of what happens if three commas is hacked and all the API keys are leaked and
people start to drain.
It did happen, exactly. So we have a different solution, which is basically to give people
packaged software that they run and install themselves, and they run locally.
And so they can manage their own keys locally without ever having to give them to us.
And otherwise, the software itself is just straight plumbing straight through to the
exchange or the protocol.
So when they're running the software, do they still take the API risk with connecting to
exchange?
I mean, I guess there's always some sort of risk if you're
connecting to an Exchange API that something could happen, but then I guess that's just out of your
hands completely and puts it as their responsibility to handle it and secure their assets.
Right. And so for people with levels of sophistication, what they can do is they
can encrypt their keys on disk at rest on their local servers.
And then they have the full security of their own local setup.
I mean, it is true that if someone hacked their desktop computer where they're running this,
then that might be an issue.
And if they can decrypt the local key, then that's their issue.
But we think that is a much more difficult problem for a hacker to try to solve than to get into a single database on a single server and suddenly get access to hundreds of thousands of keys. Did they ever
really figure out what happened with that three commas hack? Because at the time,
I just saw these random reports where people said, I think my account was drained because
of three commas, but I never really saw
any sort of a conclusion to that. So I think there was a conclusion posted and I don't remember if
it was through social engineering or what, but there was like an employee account, I think that
got compromised. I can't remember exactly what the conclusion was. I'd maybe be confusing that
with a different situation, but I think you're right. Yeah. So with, with what you're building,
it seems to have happened really fast.
I mean, you just mentioned it started in January.
It's only April and you guys are effectively deploying the product already, correct?
Yeah.
So in the coming weeks, we're going to start onboarding our first kind of initial customers
that are going to try out the product in like a limited fashion.
You know, they're not going to have their re have every feature on day one, but they will be able to connect to a couple exchanges and a different DeFi protocol,
try out the initial kind of algorithm simulation platform, the initial GUI.
And we think that will give people a good taste of what it is that we're building and show that
through a single kind of unified front end or or unified API, they can access Coinbase,
they can access Uniswap, they can access CME for regulated futures and options,
kind of all from this kind of singular platform. So I've definitely seen some that talk about
connecting to DeFi and even CeFi, but I don't think I've seen anything that connects to CME
or regulated platforms like that. So how does that work? Because to trade on CME, but I don't think I've seen anything that connects to CME or regulated platforms like that.
So how does that work? Because to trade on CME, obviously you have to have a certain level of
sophistication, licensing, et cetera, right? I mean, most people have to trade on the CME
through an investment advisor or some sort of fund, right?
Right. So the way that regulations work in the US for a CFTC regulated exchange like the CME is that you can't access
CME directly. You can't open a CME account the way you can open a Coinbase account. You have to
go through what's known as a futures commission merchant, which is the sort of CFTC NFA equivalent
of a broker. And you open an account with them, they're the ones who hold your funds. They're
the ones that can extend margin. And then they post margin on your behalf to the exchange clearinghouse. What we can be,
architect, is what's something called an independent software vendor. And what that
means is that we're not a broker. We're not taking customers. We're not onboarding customers
or taking customer funds. We are a pure straight through technology provider. And we've already onboarded as a company with several FCMs. In this
case, we've onboarded with StoneX and with Wedbush Securities. And they can clear and
onboard customer trades on our behalf. So let's say the earliest version of this will look like
the customer wants to trade futures. They'll go, let's say, to StoneX or Wedbush. They'll open
an account with them. Same way they kind of open an account with Coinbase, come back with their
credentials to Architect. Then Architect will actually have a connection to the CME once we
get certified with them to be able to trade directly to the CME. So but on the DeFi and
the Coinbase and those sides, it'll be more direct
to the consumer, correct? So I guess the platform will have some access to those
brokers on your behalf, but also your direct access to the things that you don't need them for.
Exactly. So in the Coinbase case, you will have your Coinbase account, you'll take your API key,
put it into Architect. In the case of CME, you'll have
your Redbush or Stonex or whatever other FCM that we end up working with. You'll have those accounts,
you'll put the credentials into Architect. If you're doing DeFi, you'll either kind of create
a wallet through Architect or you'll write down your private key in an encrypted way that will go
into Architect and that will be able to let you use a hot wallet in an automated fashion to connect to a place like Uniswap. So why is it important for a trader investor
to have access to all of these tools at once or in one place? Is it to execute more sophisticated
hedging strategies? I mean, is this for your everyday person who just wants to buy some Bitcoin
or is this really for institutions or sophisticated investors that are looking to deploy these complicated strategies?
Sure.
So I would say this is for everyone except for the people who want to buy Bitcoin once
a month through a Coinbase type interface.
This isn't for dollar cost averaging.
Yeah.
For the kind of interactive brokers type customers of the world, the people who they don't have
to be a Jane Street or a Winter Mute, but they might be a sophisticated investor who
they want access to good data and visualization.
They might want to run like a VWAP over the course of a day.
They might want to kind of see where the best price is in the multiple different venues.
They might be sophisticated enough to want to swap on some DeFi swap platform.
Having all of that bundled into one application gives them a lot of flexibility.
There's another philosophical point for us, which is that we don't exactly know the direction
that liquidity is heading, especially when you think about the US markets that are such
flux because of the regulatory landscape. Yeah, no liquidity. Yeah, especially when you think about the US markets that are such flux because of the regulatory landscape?
Will the existing centralized crypto exchanges
continue to exist and grow in their current form?
Will crypto volume move almost entirely to DeFi?
Will they move almost entirely to highly regulated exchanges and products like CME Futures?
I mean, you saw, I think it was in January, February, CME had its best months ever for volume in the derivatives products.
I mean, that is directly a result.
I mean, part of it is a result of like the winter starting to thaw a little bit, prices going up.
But a lot of it has to do with people still want exposure to crypto, but they know that the spot side of crypto in the US
is under a lot of scrutiny and they want to figure out how they can do this in a way that's compliant
so they're going to access a venue like CME. And so we want to provide access to all of them
instead of picking a single horse in the race. Trey Lockerbie
Well, it's interesting because for two months in a row, which had certainly never happened before as far as I know, Uniswap volumes have far exceeded those on Coinbase.
So that has to tell you, I don't know if that's an indictment of Coinbase or even a statement
about the popularity of Uniswap.
I think it's more just about, like you said, that fear factor and then the just decrease
in on and off ramps and liquidity in general to centralize exchanges
in the United States. Right. And on top of that, hearing a lot of stuff now about people trying to
create new exchanges in the wake of FTX. And some of the ideas include, and there's a couple of
exchanges that already exist that do this, but there are people who are trying to do off-chain matching with on-chain settlement. So having that sort of the speed of like a
centralized database for matching customer buys and sells that you can have, you know,
transactions that occur in the order of microseconds instead of on the order of
seconds or minutes. At the same time, having on-chain chain settlement so that the exchange itself is never holding on to customer files or not holding it on for too long.
And people get that.
One of the most important aspects of DeFi, which is self-custody of fund and not having to trust an intermediary with holding on to your coins.
And so there's so much flux in the exchange landscape right now that it's really difficult to tell
where it's going to head and it's going to be jurisdiction specific.
And so what we want to do is position ourselves very well to be there as an infrastructure
provider, kind of regardless of where it heads.
And that's interesting because the thing that's sort of lacked in DeFi, I think, is the robust
orders and order books that you have on centralized exchange.
A lot of people have tried it, but I haven't. I've yet to see anyone perfect it. Some of them
get pretty close. But to be able to actually see an order book and place a trailing stop at the
same time that you have a take profit... Things that have been so obvious on any major, even
retail platform in the past. Really, we've struggled
to get there, I think, in DeFi. So it would be pretty nice to be able to do it in the manner
that you just described. Yeah. I mean, if you're a market maker and you want to quote a few ticks
wide on something like Bitcoin, in order to be able to quote that type, you need to be able to move your order that way quickly.
Well, if you are on a platform like Uniswap, where you're market making in a pool,
and all of a sudden, Bitcoin moves 2% in a minute, and it's going to take you a few minutes just to
get that liquidity order moved in some way because of how slow ethereum is
you're not gonna you're gonna get run over and like i think there's been a lot of studies about
you know market makers generally losing on the platform for some someplace like uniswap
and so there needs to be evolution in the market structure for dexes whereby either
the matching can happen off-chain and be very fast, or the exchanges themselves
can provide mechanics that allow for market makers to employ on-chain strategies, like
automatically move my order if the underlying oracle moves by a certain amount.
Trey Lockerbie Yeah.
And we've seen market makers, I think, struggle in this new environment of low liquidity.
I mean, not my core competency, obviously, but I've read enough on it that there really is not that much profit to be had when these banking rails are cut and people aren't trading as much
on these centralized exchanges. Anecdotally, we've seen major slippage on Coinbase that didn't
seem to exist before. It's becoming really problematic to attempt,
especially if you're trading with size, to do that in the US.
Yeah, a lot of funds in the wake of FTX, but also as volumes have decreased,
as prices have gone down, have either paused or closed up. And it'll be interesting to see what happens over the next three to six months. There might be a lot of shops that are in
building and research mode and they'll
come roaring back once
things are more stable or there's a little bit
less scary regulatory
stuff happening right now.
Or it could be
there's a more permanent dry up in
liquidity on certain kinds of exchanges and
it's all going to move to
DeFi. It's all going to move to DeFi. It's all going
to move to regulated futures. Or there could be other highly regulated players in the security
space that step in to build exchanges in the existing traditional finance paradigm for market
structure and offer a spot that way. And maybe that's the way that we end up with an SEC-compliant spot exchange.
So, we'll have to see.
NASDAQ, Fidelity, et cetera.
We already see.
Yeah.
That's kind of a sad future, I think.
Not that it's not somewhat inevitable,
but listen, you are fundamentally a part
of all of the contagion.
Obviously, last year you suffered it
probably worse than anyone.
But right.
I mean, Voyager, BlockFi, Celsius, FTX, the long list of places that Americans trading
were otherwise just got access to crypto markets has diminished to almost nothing.
I mean, you still have Gemini, I guess, Coinbase.
But now even Coinbase is registering offshore.
And this is, I think it's political theater,
but they're making a very definitive statement
that they're also willing to go offshore.
Their business is here.
But it seems like there's a non-zero chance
that there's just nothing left for us as Americans.
Yeah, it's possible.
It's a bad future for a lot of different reasons. At the same time,
there are little glimmers of hope. There's all this rhetoric in the US surrounding the
market's regulators. And at the same time, was it just yesterday or was it even today
that the European Union voted to approve MICA.
And so ESMA is going to be writing rules for crypto over the next year.
That is going to be in place by 2024.
And typically, the US is the largest capital market center for all trading,
whether that's futures, equities, options.
It happens in the US.
Maybe that's not where crypto heads over the next couple of years
until the administration turns over and we have different people in charge.
It's hard to tell.
The optimistic case would be, well, there's comprehensive regulation in the EU.
Hong Kong is trying to attract crypto companies
and actually seems to be working with China
to be able to attract business there.
Singapore has comprehensive rules.
And all these different jurisdictions are coming up with actual clear guidelines for
how you might register a security token or a token exchange.
And maybe that spills back to the US instead of going the other way around.
And finally, we say, okay, well, maybe what we need to do is adopt a very similar framework to what's
happening with, you know, with Micah and Esna. And that ends up, you know, forming something
that is approachable for crypto companies in the US. Yeah, I think Micah has its problems. And a
lot of people would say that it's suboptimal, but it feels like we're
in an environment in the United States now where even negative clarity would be better than no
clarity at all. I mean, is that your feeling? Yes, absolutely. I mean, if you think about
the CFTC, for example, you know, back in sort of the middle of 2022, there were these articles that were saying
things like, oh, CFTC is the software regulator and things like that.
It's insane.
It's totally untrue.
The rules under the CFTC and the SEC are both extremely strict and comprehensive for registration.
It takes two to three years to register a derivatives exchange in the US
with the CFTC. It takes five years plus to register a clearinghouse with the CFTC. So like
LedgerX, when we acquired them, it took them five years to get their DCO, their derivatives
organization license. So these things take a long time. It is not as easy as starting
another Coinbase or Gemini. However, those are the most liquid futures and options markets in
the world. And so a strict framework that is maybe suboptimal from an innovation perspective can still
be a set of constraints within which people can innovate to the maximum
that they can and provide really great products for consumers and institutions. And I think that
even if we have that, it would be a giant step forward. Having seen all the failures last year,
obviously, being at FTX US, what did you learn from those experiences or seeing, not necessarily the
collapses or just seeing what was lacking in those products or what people were looking for
that's helped contribute to your thought process for what you're building now? I mean, it was
literally only last September you were there. To me, it's just incredible the speed at which
things are developed and happen in the crypto space. I'm sure to you, it feels like it was 10 years ago at this point.
I have a lot more grace than I had back then.
Me too.
So, you know, a couple of high-level learnings.
Obviously, you're going to ignore all of the fraud aspect,
which was just terrible and shocking and it it definitely raises questions
about not how you prevent fraud because criminals will always find a way but more like what kinds of
uh systems can you put in place to be able to have proper oversight like uh you know the people
working on like proof of reserve or proof of solvency protocols, I think that could be
extremely helpful. Having independent agencies looking over centralized entities in the crypto
space. I think taking another step back, DeFi has so much promise because it is a way of providing
truly transparent execution and custody that everyone can see and everyone can audit.
As long as we're going to have a world where there are centralized actors and digital assets
that basically form the function of brokers, they really are. They hold dollars on behalf
of customers and they execute orders on behalf of customers. Those basically should be regulated
and audited and overseen like brokers. And that should be a separate thing from DeFi, but I think that would be a huge help to regaining
trust and confidence in the digital assets world.
Some personal learnings are trying to build products and services that are sort of decentralized.
I don't necessarily mean that in the blockchain sense.
I mean more in that the customer maintains as much control as possible over their own information.
We talked earlier that it was important for us to create a security model by which the clients
of architects can maintain control over their own keys, their own wallet keys, their own API keys,
their own account information at futures exchanges. And that, to me, was very much like
direct learning from what came out of FTX.
Another aspect is that we have nine people on our team. Two people are legal and compliance.
This has to be in place from day one. There needs to be a comprehensive compliance program,
a comprehensive compliance manual. We should start applying for licenses in the US with the
market's regulators as soon as possible. We should not shy away and move offshore.
I think a lot of the rhetoric now is like, okay, well, the only way to successfully build a crypto
company is to move offshore. I think that's giving up. For us, we want to stay in the US for a US
company. And we're going to figure out how to
work within that regulatory regime and do things the right way to whatever extent possible that we
can. And that's been another learning for us. That's interesting, because I think what you do,
you can make compliant and legal, but there's the sort of existential threat that the
people you're serving can't find a way to be compliant
and legal. We haven't really seen, I don't think, a huge increase in rhetoric against DeFi per se,
I don't think yet. But do you worry about the risk that in the US they would try to find a way to
say, listen, you can't even access a Uniswap or DYDX or any of these, and therefore then
your US customers still can't use your your platform
as a practical matter i find it hard to imagine how you could prevent someone from accessing
an open source you know peer-to-peer exchange mechanism on a blockchain that can be
infinitely forced and moved and there's there's dozens of different competing chains
that you can potentially have these mechanisms on.
So I don't see that being possible.
I'm still optimistic that we're going to get to a world
where DeFi and CeFi can coexist in a way that is allowed.
And it might be that the architects of the world, and I mean that with a capital A for a company, can get licensed in a way where we can be a licensed front end for services. the requirement to access DeFi, but for the institutions that want to try to access DeFi
in a way that is allowed, we're going to know the customer anyway, because we're going to
hold their funds and provide access to regulated exchanges.
Well, why not also be a regulated portal into DeFi so there can be significant liquidity
provision so that for all the little guys out there who want to take liquidity from all over
the world and these different swap protocols, these DEXs, other marketplaces, they'll be able
to know that they're interacting with sophisticated sources of liquidity.
Well, I think whether people like it or not, at this point, the lesson we have to have learned
is that if you want institutional adoption, it's not going to happen without
regulated rails. And that doesn't take away from DeFi, but if we want DeFi TVLs to be
trillions of dollars, okay, maybe that's a little hyperbolic, but if we want that to be the case
and we want institutions to be there, they're not just getting a MetaMask wallet and going into
Uniswap to trade. It's just not going to happen.
Yeah, there are other ideas too that have been
kicked around like
not preventing
anyone from accessing DeFi.
But if you're someone who
has scrutiny over
which DeFi pools you can access
based on the risk profile of those pools
and by risk profile I don't mean
the price risk profile, I mean't mean like the price risk profile,
I mean more who are the actors in the pool. Yeah, like they're going to get rug pulled
or you're trading with North Korea, right? Yeah.
Exactly. Exactly. Yeah. Then some combination of on-chain analytics tools,
whether it's like a chain analysis, TRM, can tell you sort of like the average risk profile of a pool based on history
of transactions that have resulted in liquidity provision for that pool. I think that could be a
thing where the institution says, okay, I can provide liquidity in that pool because I have
some assurance over the aggregate risk profile, but I'm going to ignore this other one. It doesn't
prevent the other one from existing. It doesn't say we're going to have only regulated DeFi and nothing else, but it does at least
provide a path that we can transition into that world. And I think we're going to need that if
we're going to get to some wider adoption. Yeah. I mean, the individual can still just
open their MetaMask wallet and connect to Uniswap or whatever and do whatever they're going to do.
So it's really just providing the access for those institutions and the security and safety to do so, which kind of begs the next
question. What else are we lacking in DeFi right now that we would really need for them to start
participating in a meaningful way? Is it insurance? I mean, what are the things that we have
in other markets that they're probably waiting for to do this properly?
So one of the best aspects of blockchain-based transfer mechanisms, and also simultaneously one of the most concerning for an institution, is the irreversibility.
Right.
So if you are in any kind of fintech and you take payments through something like ach you know that fraud is one of the greatest things you have to deal with the fact that
within within the weeks someone can reverse an ach transfer or they can reverse a credit card
transfer and so you have to spend a lot of money and time trying to guess whether you think someone's
going to reverse a transaction because you might not have it in a little while. What's so nice with USDC is they send it to you, it's yours,
there's basically zero fraud risk. However, humans make mistakes.
Right.
And if you're an institution and you send like 100 million to the wrong
To the wrong address.
Wow. And this has probably happened many, many, many times.
I think that's something that's a little bit, it's a hard pill to swallow.
And I've written about this before, but it feels like there can be fully on-chain mechanisms to
have conflict resolution when it comes to potentially like reversing a transaction. And I don't mean
taking blocks off of the record and reordering them. But I do mean like having some kind of,
you know, whether it's like an escrow with approval for transactions, or some kind of
like voting mechanism that allows for like, you know, kind of multi-sig transfer back from another
wallet, that would
allow institutions to become a lot more comfortable with the fact that, okay, 99% of the time
they don't need to need this, they will need this, but in like the absolute worst case
scenario they make a mistake or there's some kind of conflict resolution that needs to
happen, it's possible.
And that could be facilitated only between people who effectively sign on to utilize
that process.
Exactly.
Because I know it's a controversial idea to your average DeFi person to be like,
a transaction should be reversible.
But if it's two institutions that are transacting together and both
sign off or utilize some smart contract for escrow or for some sort of process,
that's those people's freedom and choice to do so.
It's not forced upon them.
Yeah, let's make this very clear. I am not suggesting we change ERC-20 to require
multi-sig on all transactions and reversibility. But I think that there can be opt-in mechanisms
that, again, would allow these larger players to start participating in a way that would make them
feel safe and comfortable in doing so. That makes a ton of sense. So beyond just the plumbing of being able to connect the APIs and
the obvious able to trade, you sort of hinted at the fact that there would also be data,
potentially charts, things like that. So how much of that sort of actionable information and data
are you going to be providing as part of the platform? Does it really become a one-stop shop
where you don't even need a trading view or wherever
you get your data and you can do everything on the singular platform?
Yeah.
So we are really hoping this is a one-stop shop.
We are licensing TradingView.
So the TradingView widget will be within the platform.
You have to.
Yes.
And it really is the best.
We built a kind of like Bloomberg Terminal-like application for a desktop that allows you to
really quickly build dashboards and view your portfolio. We're experimenting with a couple of
interesting features relating to AI, including having various GPT-powered features directly within the platform.
So for example, you're a new Alva developer,
and you want to get started writing some strategy,
and you say, okay, I'd like to arb Bitcoin on Coinbase and Kraken,
and then having something just spit out code for you
that's a great skeleton that works specifically trained on our api so
they're working on the architect's apis we think would be a really great way of getting people
into more sophisticated forms of trading that otherwise you would need a lot of time and
experience to be able to create um thinking a lot about how to like use ai for kind of news
analysis and sentiment analysis in the platform.
So yeah, we're really hoping to build a comprehensive platform over the next three
to six months that everyone can try. So I'm kind of shaking my head here because
my next question was, were you considering doing any of that in January when you started
the AI side? Because it's happening so fast.
No, I think, you know, uh, so some of my background, so in, in college, like my, um,
like bachelor's and master's work was in AI. And specifically it was like, I was trying to do move
prediction in the game of go. And this was, to DeepMind, AlphaGo, all that stuff.
And at that time, I was like, you know, when I wrote my thesis, I said, okay, well,
you know, we're probably many years off from AI being able to beat anywhere near the top
professionals in Go. And then a couple years later, that was like completely blown out of the water. And we went in such a great leap in AI ability at
that point. And now it's like we are totally parabolic and it's something that we've been
following. But now that it's available, you can access it via API, it works incredibly well
across so many different fields. It feels like something that's impossible to ignore.
And I don't mean in sort of like the cynical way of like,
oh, like add AI to your thing
and then try to get money from VCs, right?
Like this is like, we really like think
that there are complimentary aspects to our platform,
probably to many others,
that with like a little bit of integration
with these new AI tools,
we can really give users an amazing experience that they otherwise couldn't have had a few months ago.
And with time, that will be totally abstracted away anyways, and we'll just be typing into a
search engine on the platform. It's not going to be like people are going to have to even understand
that it's AI or understand how to
utilize AI at all. They'll just probably ask your platform a question and it'll give them the proper
answer. Yeah. There's a lot of talk to your data stuff happening with GPT and other LLMs right now
where, let's say you have a... I actually tried this out. So if you go to GPT and say, okay, I'm long a call option on this strike, and I'm short a put
option on this other strike, and I also am long stock this amount, describe my portfolio.
It actually does a very good job of saying, oh, you have this particular spread on, which has
this risk. It's actually pretty amazing that you just like put in your own data and ask it questions and get reasonable natural
language responses. And I think it's simple, but really powerful for something like a trading
front end. Does it take the next step and say, Hey, listen, like you're not that zero,
your risk is completely like off weight. And this is you need to do to uh get back to your
proper strategy i haven't pushed it there but i would be surprised if that wasn't possible with
that if not immediately then with a little bit of fine tuning it's so incredible it's so incredible
so for you like you said you wrote your thesis on this i mean i guess you were and humans are
this way exceptionally good at predicting what was coming, but the timeline is really, really
difficult, right? Well, yeah, I'm not sure I was good at predicting where we were going, but it
felt like this was an exciting field to sit, to research and do something in. And, uh, man,
if it was exciting, then it's, it's probably one of the most exciting things now.
So, I mean, obviously you have a pretty deep understanding of that side.
You've been in crypto now for many, many years.
Are there other things where you see these two coming together and it being a perfect marriage?
I think another probably clear use case is just blockchain analytics in general.
There's a lot of data across a lot of different chains.
There's a huge graph, a web of transactions and different wallets in the system.
And you have tools like Dune, which lets you kind of create little dashboards of analytics on top of blockchain data, I think being able to ask questions about what's happening
across these disparate ecosystems is something that AI will do a very good job of, purely because I
think one of the best things you can do with AI is infer patterns from large, somewhat noisy
distributions. And that feels like a very good application of that.
I've never even thought about this till this moment,
but it seems that this AI would be either a great compliment
or threat to existing oracles.
I think one thing I'm not too...
It's not always accurate, so I guess there's that.
I'm not super confident about LLMs in particular.
It's price prediction,
but probably directional prediction.
Yes.
I think there was like a recent headline that GPT does like a better job
in existing algorithms on trying to deduce directional sentiments from like
that comments and from other news headlines,
but actual price prediction. I mean, that's so,
it's very much random noise. I mean,
I meant more for like verification of data, you know, in the smart contract,
kind of like link, you know, chain link or something like that, that, but,
I also, you know, we're not there yet. I mean,
I think yesterday I asked chat gpt just jokingly like
to summarize the gary gensler congressional hearing and it gave me effectively like a boilerplate
summary of what the sec should be doing that was completely irrelevant to the fact that he got
grilled on the floor and was like two years old because you know they're probably using it 2021
it was like we should have a robust regulation and it was completely inaccurate to summarize what had actually happened so well what you should try is get the transcript and then right enter that
it'll summarize it what do you think the summary is and i bet it would do a pretty good job well
yeah i i actually so i use chat gpt on a daily basis to help me summarize articles and do those
um for my newsletter right just it's made my workflow 10x.
And I think that that's a lot of people are worried about AI taking everyone's job
and crushing the industry, but I think that that's what it does.
It just makes you ten times more productive.
I use it for programming on a daily basis.
And again, it's not going to replace a programmer, but it is like, okay, I can
look through seven Stack Overflow articles that are of various
degree of quality and staleness.
Or I can ask one question of like, give me some examples of how I can achieve this thing.
A very targeted thing in a particular language, a particular library, and it will usually
give me the right answer.
Yeah.
And I've heard that it sort of catches up, you know, like a 1x programmer to the 10x
guys, and then it makes the 10x guys just supercharged. What you
would have taken a month to do, you can do three times in a day. Right.
So that seems really appealing because I think it probably gets rid of a lot of the minutiae
and annoying tasks that you have to do to get to where you want. Just eliminating the entire
workflow, like you said, of doing that research.
That's really interesting to see where it's going to end up landing in crypto.
I think we're going to see
some pretty crazy integrations
where the blockchain sort of helps
to verify what's coming out.
But anything else, any final thoughts
before I let you go
that we might've missed here
that you guys are going to be looking to build maybe
after the beta testing is done and something we haven't talked about?
Yeah, I think maybe the one thing to end on is we realized as we build out this platform that
digital assets are the most modern markets to build infrastructure for,
but there's a lot of modernization that is still wanting in
traditional assets. And so we see no reason over time to stop at digital assets, especially
because... So let's say we support cash treasuries in six months. Well, at some point, fixed income
products might be tokenized. And so the lines between what is a
digital asset, fixed income product versus a traditional asset, fixed income product are
going to be blurred. And again, we want to be prepared to handle that transition. So we're
going to look to do more than just digital assets and digital asset derivatives very soon.
Yeah. Other than that, just a plug that if you are a great engineer, either on the front end or back end and looking for work, we'd love to have you,
but so would everyone. Just hire the AI, man. Come on.
Just hire the AI for that job. But that's really interesting. So you guys are effectively building
plumbing that also can be ahead of the tokenize everything trend.
We think so. Yeah. Absolutely. Awesome. Where can everybody follow you and then check
out? Do you need beta testers as well? I mean, who are you going to get to try this thing out at
first? Sure. So I guess first I'm primarily on Twitter at Brett Harrison, 88. You can also
follow architect at architect underscore XYZ. If you go to architect.xyz, the website on the
product page, there's a signup link where
you can enter for the wait list, or you can email beta at architect.xyz.
And looking forward to the future launch.
Always a pleasure, Brett.
I love catching up and I love that seemingly you've landed so well.
I know for you, it probably wasn't that exciting for a while there or for any of us.
I'm still a Voyager creditor.
So, you know, I think everybody felt.
I'm still an F2F creditor.
So I feel your pain.
I'm quite sure you are, man.
Well, thank you so much for your time and looking forward to see what you built and
catching up down the road.
Thanks a lot.
Thank you. Let's go.