Latest / The Starting Block / Exploring the origins of DeFi protocols with Andre Cronje

Transcript

Exploring the origins of DeFi protocols with Andre Cronje · The Starting Block

• 0:08
  Good morning, good afternoon.
• 0:09
  Wherever you are around the world,
• 0:10
  joining us today, welcome to The
• 0:12
  Starting Block. If it's your first
• 0:13
  time here, my name's Gareth
• 0:14
  Jenkinson.
• 0:15
  I'm the host of the show and head of
• 0:16
  multimedia at The Block.
• 0:19
  It's Wednesday, we're midway through
• 0:20
  the week.
• 0:21
  Been an exciting week for
• 0:23
  The Block team.
• 0:24
  We have a few people on the ground
• 0:26
  at the Wyoming Blockchain Symposium,
• 0:29
  producing a lot of content there.
• 0:30
  So if you're looking for some
• 0:32
  big interviews coming out in the
• 0:34
  next couple of days, check out the
• 0:35
  Blocks YouTube channel and our X
• 0:37
  account. If you're not following us,
• 0:40
  please go ahead and do that.
• 0:41
  Follow us on X, subscribe on
• 0:42
  YouTube. If you watching the live
• 0:44
  stream here today, we're gonna have
• 0:45
  some great interviews coming out of
• 0:47
  Salt in Wyoming in the next couple
• 0:49
  of days.
• 0:50
  Of course, you can also look out for
• 0:51
  some updates from some of those
• 0:52
  interviews on our X timeline
• 0:55
  as well.
• 0:56
  Really excited for today's show.
• 0:57
  In a few minutes time, we'll be
• 0:59
  joined by André Cronier, of course,
• 1:00
  a household name in the world of
• 1:02
  DeFi and also a fellow South
• 1:04
  African, so looking forward to
• 1:05
  having a chat with him on
• 1:07
  the hot seat today.
• 1:08
  But of course, before we do that, we
• 1:09
  will be jumping into the Block
• 1:10
  briefing, which is a rundown of the
• 1:12
  top headlines from the Block's
• 1:15
  exceptional news team over the last
• 1:17
  24 hours.
• 1:18
  So without further ado, let's jump
• 1:20
  straight into that today and
• 1:22
  kicking off with our first headline,
• 1:23
  Wall Street Heavyweight City is
• 1:26
  stepping into crypto custody.
• 1:28
  The bank confirmed yesterday it will
• 1:30
  launch a digital asset custody
• 1:31
  service later this year, starting
• 1:33
  with Bitcoin under a new platform
• 1:35
  called Custody Plus.
• 1:36
  Clients will be able to hold
• 1:37
  traditional and crypto assets inside
• 1:39
  the framework.
• 1:40
  It's the latest move in city, steady
• 1:42
  digital asset build out.
• 1:44
  The bank has already piloted
• 1:46
  tokenized deposits with Swift and
• 1:48
  Intercontinental Exchange this
• 1:50
  year. For one of the world's largest
• 1:51
  banks to offer Bitcoin custody
• 1:53
  signals, just how far institutional
• 1:55
  adoption has come in recent
• 1:57
  years.
• 1:58
  Our next headline, Washington's
• 1:59
  gridlock isn't stopping the SEC.
• 2:02
  The regulator on Tuesday pronounced,
• 2:04
  proposed rather a new rule change.
• 2:06
  Regulation crypto assets,
• 2:09
  a tailored offering regime that
• 2:10
  would exempt certain digital asset
• 2:12
  offerings from traditional
• 2:13
  securities law.
• 2:15
  Chair Paul Atkins framed it as a way
• 2:17
  to help entrepreneurs raise capital
• 2:19
  while protecting investors and
• 2:20
  to unsure crypto innovation.
• 2:23
  It builds on guidance the SEC
• 2:25
  floated back in March.
• 2:26
  Crucially, the agency is acting on
• 2:28
  its own as broader market structure
• 2:30
  legislation stalls on
• 2:32
  Capitol Hill, assign the SEC.
• 2:34
  Won't be waiting for Congress to
• 2:36
  move.
• 2:37
  Then our next headline, some good
• 2:38
  news for battered Bitcoin bulls.
• 2:41
  VanX says the roughly 11 month
• 2:43
  correction may be nearing its end.
• 2:45
  The asset managers researchers found
• 2:46
  eight of 12 signals in
• 2:48
  its Bitcoin capitulation check
• 2:50
  are flashing and all 12 have
• 2:52
  hit the capitulation zone at some
• 2:54
  point over the past three months.
• 2:56
  They read here the market is likely
• 2:57
  seeing capitulation and is
• 2:59
  entering an accumulation phase.
• 3:02
  Bitcoin's trading near $64,700
• 3:05
  range bound since June and
• 3:07
  still about 48% below its October
• 3:09
  record of roughly $126,000
• 3:12
  last year.
• 3:14
  Tokenization keeps pulling in Wall
• 3:15
  Street's biggest names.
• 3:16
  Securitize on Tuesday launched the
• 3:18
  Newburger Securitas High Income
• 3:21
  Tokenize Fund.
• 3:22
  The first time Newburger, a global
• 3:24
  manager with $230 billion
• 3:26
  in assets has acted as
• 3:29
  sub advisor to a tokenized fund.
• 3:31
  The fund with the ticker HINC
• 3:33
  targets high yield bonds and
• 3:35
  income producing fixed income.
• 3:37
  And it's launching across four
• 3:38
  blockchains, Avalanche, Ethereum,
• 3:40
  Solana and SWE.
• 3:41
  CEO Carlos Domingo says it brings
• 3:43
  Newberg's established fixed income
• 3:45
  muscle onto public blockchain
• 3:47
  through securitized, regulated
• 3:49
  end-to-end tokenization
• 3:52
  platform.
• 3:53
  And the block team is sitting down
• 3:55
  with an interview with Carlos Dmingo
• 3:57
  while they're in Wyoming.
• 3:58
  So keep an eye out for that.
• 3:59
  In the next couple of days.
• 4:02
  Japan's MetaPlanet is going shopping
• 4:04
  in America. The self-styled strategy
• 4:05
  of Japan is taking a roughly 96%
• 4:08
  stake in NASDAQ listed Super League
• 4:10
  entertainment through $134
• 4:12
  million investment, turning the
• 4:14
  gaming media firm into its US
• 4:16
  Bitcoin treasury platform.
• 4:18
  Super League shares spiked as much
• 4:20
  as 80% yesterday before settling
• 4:22
  around 50.
• 4:23
  Analysts say this deal stands apart.
• 4:25
  Mates Planet is funding it with
• 4:27
  Bitcoin from its own balance sheet,
• 4:28
  not discounted third-party capital
• 4:31
  and its shares carry a
• 4:32
  five-year lockup.
• 4:34
  Our final story,
• 4:36
  Kraken is expanding beyond crypto
• 4:38
  and beyond America.
• 4:39
  The exchange has launched US stock
• 4:41
  trading for eligible customers
• 4:43
  across the European economic area
• 4:45
  following a limited rollout in
• 4:47
  Germany, the Netherlands and France.
• 4:49
  Kraken says that makes it the only
• 4:51
  crypto native platform offering EEA
• 4:53
  customers both traditional.
• 4:55
  U.S.
• 4:56
  Listed shares and tokenized
• 4:57
  versions.
• 4:58
  It's X stocks inside one
• 5:00
  regulated account.
• 5:01
  Rivals like BitPan.crypto.com and
• 5:03
  Robinhood offer one or the
• 5:05
  other in Europe, but Kraken's
• 5:07
  betting that bundling both under
• 5:09
  one roof will give it
• 5:11
  an edge in this market.
• 5:12
  And that is it for the block
• 5:14
  briefing.
• 5:14
  This afternoon or this morning,
• 5:16
  depending on wherever you are around
• 5:17
  the world, we're gonna be shifting
• 5:18
  on straight into the hot seat
• 5:20
  this morning.
• 5:21
  And it's a great pleasure to have
• 5:23
  André Cronier joining us.
• 5:24
  The founder of Flying Tulip
• 5:26
  and many other household names in
• 5:28
  the DeFi space.
• 5:29
  First things first, Andre, thanks so
• 5:30
  much for joining us.
• 5:31
  Oh, thank you for having me.
• 5:32
  And I have to admit,
• 5:35
  I feel almost a little bit bad
• 5:36
  saying this, because it's nothing to
• 5:38
  any of the other interviews
• 5:40
  I've been on. But I mean, as you're
• 5:42
  a South African, you know, my
• 5:43
  surname is fairly hard to pronounce.
• 5:45
  Most people that aren't from South
• 5:46
  Africa can't really get the cronia
• 5:48
  right.
• 5:49
  But when you said
• 5:51
  it originally now, at the start of
• 5:52
  the intro, I was like, oh, wait,
• 5:54
  yeah, that is my surname.
• 5:55
  Because I've almost gotten used to
• 5:56
  the more Americanized
• 5:57
  pronunciations.
• 5:58
  Anyway, thank I don't even want to
• 6:00
  ask.
• 6:01
  How many variations of that
• 6:03
  is? Kronji, Kronje?
• 6:05
  There's a lot.
• 6:05
  There's a lot. I accept everything.
• 6:07
  It's fine.
• 6:09
  It doesn't matter at this point.
• 6:11
  Hi, it's all good.
• 6:12
  Andre, it is a great pleasure to
• 6:13
  have you on the show.
• 6:15
  Obviously, as I alluded to before,
• 6:17
  you really are a household name in
• 6:18
  the name of DeFi.
• 6:19
  You've been around for a very long
• 6:21
  time. You have built some of the
• 6:23
  best technology in the industry.
• 6:25
  For anyone out there that does not
• 6:27
  know a lot about you, can you just
• 6:28
  take us back a few years and
• 6:29
  actually tell us how you found
• 6:31
  yourself edging into
• 6:33
  the crypto space and what
• 6:35
  started you on that journey?
• 6:36
  I know a little bit of people start
• 6:37
  with Bitcoin.
• 6:37
  You obviously were, you know, very
• 6:39
  influential in different DeFi
• 6:41
  protocols and building different
• 6:42
  tech stacks, but did it start
• 6:44
  with Bitcoin or was it more
• 6:45
  Ethereum? Can you, can you take us
• 6:47
  back in time?
• 6:48
  Neither.
• 6:50
  I'll try and keep it as brief as I
• 6:51
  can. But it started, I was a
• 6:53
  CTO at a fintech
• 6:55
  new bank in South Africa.
• 6:59
  We were dealing
• 7:01
  with a lot of tech in
• 7:03
  distributed systems, in
• 7:06
  large data processing,
• 7:08
  data lakes back then, Kafka, those
• 7:10
  kinds of tools.
• 7:12
  A colleague who was in our
• 7:14
  business analysis team.
• 7:18
  Couldn't stop talking about Bitcoin.
• 7:19
  This was like 2016,
• 7:22
  early 2017.
• 7:23
  There was the big boom.
• 7:25
  Couldn't stopped talking about it.
• 7:27
  But a lot of the
• 7:30
  frequency of which I was hearing it
• 7:32
  and then as a person interested in
• 7:33
  the technology side of it, I started
• 7:35
  digging in to see, because
• 7:37
  what originally got me is
• 7:40
  this fairly new industry.
• 7:42
  And this is mostly ICOs.
• 7:44
  This has actually got nothing to do
• 7:45
  with a theorem or Bitcoin.
• 7:46
  But ICO's were promising to have
• 7:48
  solved a lot of really
• 7:50
  hard historic problems that have
• 7:51
  been technology issues that
• 7:54
  the industry as a whole has not been
• 7:56
  able to solve for decades.
• 7:57
  And here's a white paper from an
• 7:59
  18-year-old at a high school saying
• 8:00
  they solved it.
• 8:02
  So my original entry into the
• 8:04
  space was actually fairly
• 8:05
  antagonistic in that the first thing
• 8:07
  I started doing was writing what
• 8:09
  later became known as crypto
• 8:11
  code reviews.
• 8:12
  Where I would,
• 8:14
  for myself, as a learning exercise,
• 8:16
  I would go into their get-ups, I
• 8:18
  would see
• 8:20
  if that matches to their claim.
• 8:22
  Often that didn't match to the
• 8:24
  claims.
• 8:26
  That was meant as
• 8:28
  a teaching exercise and
• 8:30
  a documentation exercise for myself
• 8:32
  but because it was public that
• 8:33
  actually blew up quite a bit and
• 8:34
  people started using that sort
• 8:36
  of as, you know, back then everyone
• 8:37
  had these ICO spreadsheets with like
• 8:39
  numeric values that tribute to it.
• 8:41
  And I started becoming a line item
• 8:43
  in that spreadsheet.
• 8:43
  If I had written a review, there
• 8:45
  would be like, hey, what's the
• 8:47
  verdict I gave?
• 8:49
  And that ended up me getting
• 8:50
  involved in crypto briefing, which
• 8:52
  further gave me a lot of access to
• 8:53
  also private repositories.
• 8:55
  And then it almost became a thing
• 8:56
  where people wanted to have their
• 8:57
  code bases reviewed,
• 8:59
  sort of as a approval as
• 9:02
  such.
• 9:03
  But that itself ended up in
• 9:05
  a little bit of a bastardization
• 9:07
  because it's sort of self-selection
• 9:08
  then. So that the people that have
• 9:10
  good code are the ones that want to
• 9:11
  be reviewed. The people that had bad
• 9:12
  code don't want to review.
• 9:14
  And that also kind of just lost the
• 9:15
  purpose of what I was doing because
• 9:16
  it was no longer me
• 9:19
  trying to teach myself and learn
• 9:21
  about the industry.
• 9:23
  So I started walking a little back
• 9:24
  from that.
• 9:25
  But as much
• 9:27
  false claims as exist in
• 9:29
  the industry, was a sliver
• 9:31
  of reality in that there is
• 9:33
  something here.
• 9:34
  There is a real
• 9:36
  solution to
• 9:39
  currency that can exist without of
• 9:41
  specific borders and that is
• 9:43
  powerful.
• 9:44
  I mean a huge problem for SA
• 9:46
  banks. I can't really comment if
• 9:47
  this is for other banks as well, but
• 9:49
  that was the sort of daily
• 9:51
  settlement, which is.
• 9:53
  Just people sending spreadsheets
• 9:55
  across and then needing to go over
• 9:56
  millions of rows of manual
• 9:58
  data to then check, okay,
• 10:00
  what is the balances, and then
• 10:01
  physically need to move actual cash
• 10:03
  from the one to the other one.
• 10:05
  It's a garbage process.
• 10:07
  And I don't know if it's changed by
• 10:08
  now, but I hope
• 10:10
  so. But even that,
• 10:12
  something as simplistic as that
• 10:13
  blockchain could solve, because the
• 10:15
  reason.
• 10:16
  All have to do that is because they
• 10:17
  can't just trust the counterparty.
• 10:20
  But if you have all of that booked
• 10:21
  on the chain,
• 10:23
  even if it's not the actual
• 10:25
  financial transfer, it's just a
• 10:26
  booking entry, then you already
• 10:28
  have an easier matching thing.
• 10:29
  Anyway, from
• 10:32
  the crypto briefing
• 10:34
  days, I'd also started doing some
• 10:36
  advisory. Most notably, I think
• 10:38
  Phantom is probably what a lot of
• 10:39
  people know me for, even though I
• 10:41
  joined there after the fact.
• 10:42
  But they
• 10:44
  They did not have a tech team
• 10:46
  capable of delivering
• 10:48
  their technology, so I got involved
• 10:49
  there.
• 10:51
  I had actually been working on
• 10:53
  the theoretical idea of this
• 10:55
  asynchronous blockchain for a while,
• 10:56
  and then because they didn't have a
• 10:58
  working something, I actually
• 10:59
  offered that to them, and they then
• 11:01
  ended up running with that.
• 11:02
  Because my idea was they have
• 11:04
  money, they don't have tech.
• 11:05
  I have tech, I don't money, so just
• 11:07
  take it.
• 11:09
  But ultimately, layer ones weren't
• 11:11
  really my passion.
• 11:12
  My passion always sat in that sort
• 11:13
  of financial layer because that's
• 11:14
  where I have my own financial
• 11:16
  background in and so I started
• 11:17
  playing around a lot with healed
• 11:19
  automation, which
• 11:21
  back in the day was a lot simpler.
• 11:22
  There were three to four immutable,
• 11:26
  fee-less protocols that
• 11:28
  gave you healed.
• 11:28
  So it was very easy to write
• 11:30
  something that is immutable and
• 11:32
  static and that can just check the
• 11:33
  yield and switch, right?
• 11:34
  Because there's no,
• 11:36
  nowadays, I don't think a product
• 11:37
  like that. Can realistically work
• 11:39
  in the same manner.
• 11:40
  You need active curators, active
• 11:42
  risk teams.
• 11:43
  There's so many counterparties,
• 11:45
  off-chain liability, these kinds of
• 11:47
  things.
• 11:48
  And everything else was just kind of
• 11:51
  a solution to a specific
• 11:52
  requirement we had.
• 11:54
  So with Yearn, as it continued
• 11:56
  growing,
• 11:59
  and for those who don't know, Yearn
• 12:00
  was kind of the very first.
• 12:02
  Faults and automation
• 12:03
  in terms of optimizing your
• 12:05
  yield, which nowadays is mostly done
• 12:07
  by off-chain curators.
• 12:09
  But one of those processes, for
• 12:10
  example, is needing to actively
• 12:12
  switch the yield from
• 12:15
  where it is to where it should be
• 12:16
  going.
• 12:17
  And back then, goal is
• 12:19
  decentralization, goal is
• 12:20
  immutability. You don't want a team
• 12:22
  that needs to maintain it.
• 12:23
  You don't wanna founder because
• 12:24
  there's no fees, there's capital
• 12:27
  extraction. So it has to be really
• 12:28
  neat.
• 12:29
  So, that led to Keeper,
• 12:31
  which I believe was the first
• 12:32
  decentralized Keeper network,
• 12:34
  which allowed random bots
• 12:36
  anywhere in the ether to be able to
• 12:38
  do these work for you, but in
• 12:40
  a way that it is agreed upon
• 12:42
  a ruleset that is codified on-chain,
• 12:44
  and that was that network.
• 12:47
  And then beyond that, I had a bunch
• 12:49
  of prototypes I'd played with.
• 12:50
  I had an insurance prototype.
• 12:52
  I had pay-as-you-go thing.
• 12:53
  I had ultimately DairySwap,
• 12:56
  you know, which was the idea of...
• 12:58
  Combining this capital efficiency
• 12:59
  because one thing that one
• 13:01
  thing why I think DeFi will always
• 13:03
  be a little bit less
• 13:06
  interesting than traditional
• 13:08
  finance is capital efficiency.
• 13:10
  DeFi as a whole has very bad capital
• 13:11
  efficiency and that's because the
• 13:13
  asset can't really just be held as a
• 13:14
  book entry, it has to be physically
• 13:16
  held.
• 13:17
  So capital efficiency has always
• 13:18
  been something I've been obsessing
• 13:19
  with.
• 13:20
  And ultimately everything since then
• 13:22
  has really been trying to address
• 13:23
  this capital efficiency concern.
• 13:26
  VE33 was a new sort
• 13:28
  of tokenomic bootstrapping
• 13:29
  mechanism.
• 13:30
  Again, the requirement at
• 13:32
  that time, that was a
• 13:34
  time in history where there was a L1
• 13:35
  launching almost every 24 hours
• 13:38
  and the biggest problem they all
• 13:40
  faced was bootstraping.
• 13:41
  You need Uniswap, you need Curb, you
• 13:42
  need all of these specific elements
• 13:44
  to get your chain started and so
• 13:46
  that was really a
• 13:49
  three-contract thing, you deploy
• 13:51
  it, you bootstrap all of that.
• 13:55
  And kudos to
• 13:57
  Aerodrome, what they've done with
• 13:58
  that is absolutely phenomenal.
• 13:59
  I mean, there's been plenty more
• 14:01
  of forks that failed.
• 14:02
  So, it's not the code that
• 14:04
  was the contributing success
• 14:06
  factor there, it's really the team.
• 14:08
  So respect to them.
• 14:10
  And then when the kind of sort
• 14:12
  of regulatory clarity changed a
• 14:14
  little bit, I decided, hey, let me
• 14:15
  take all of these ideas up and
• 14:16
  working on that I've wanted to
• 14:18
  deploy. I now finally have.
• 14:20
  An environment in which I can do
• 14:21
  this.
• 14:23
  So I started my first official
• 14:25
  crypto product, which is Flying
• 14:27
  Tulip. Put all the IP and all the
• 14:28
  brand underneath there.
• 14:29
  Did my first raise, which via
• 14:31
  the put structure, which I'm also
• 14:32
  quite proud of, because I think
• 14:33
  that's a better capital mechanism
• 14:35
  that protects the relevant interest
• 14:37
  of parties.
• 14:38
  And now we're here.
• 14:40
  Yeah I want to talk to you a little
• 14:41
  bit about the evolution of that
• 14:42
  right because as I understand that
• 14:44
  Dairy Swap was an idea you've had
• 14:45
  for many many years but probably at
• 14:47
  the time didn't have
• 14:48
  the ability to bring it all together
• 14:50
  can you talk a little about the
• 14:51
  specifics of you know the
• 14:53
  infrastructure itself and what
• 14:56
  you envision Flying Tulip
• 14:57
  to be and if you can compete with
• 14:59
  some of the you know uh
• 15:01
  the bigger names in in the DeFi
• 15:03
  space I guess you thinks about
• 15:05
  Aave now as the king of
• 15:06
  DeFi, but...
• 15:09
  Things are changing.
• 15:10
  It's a very complicated world in
• 15:12
  which we live in, and maybe you've
• 15:13
  got something better.
• 15:16
  Um, I, I mean, look, I
• 15:17
  don't always, I never liked to
• 15:19
  compare in terms of
• 15:22
  better. Cause I, every, every
• 15:24
  product surface is a very specific
• 15:25
  niche, right? The risk profile with
• 15:27
  something like flying tulip is very
• 15:28
  different, but, but let me touch on
• 15:29
  that second, DairySwap.
• 15:31
  Um, Dairyswap was a very,
• 15:33
  almost naive implementation of
• 15:35
  what flying tulip eventually turned
• 15:37
  into in that Dairieswap's
• 15:39
  idea was very almost
• 15:41
  binary. It was just the idea because
• 15:43
  at the time what a lot of people
• 15:45
  were doing is,
• 15:47
  let's say I wanted to
• 15:49
  earn some extra yield on
• 15:52
  my ETH and my USDC.
• 15:54
  So I would deposit my
• 15:56
  EF and my USDC into Aave.
• 15:58
  Okay, now that's being loaned out
• 15:59
  and I'm earning, but Aave gives me
• 16:01
  a tokenized version of that deposit
• 16:03
  in a EF and a
• 16:05
  USDC I can then go to
• 16:07
  Uniswap and I can create the pair
• 16:09
  for AEF and AUSDC and I
• 16:11
  can LP that.
• 16:12
  So now I'm earning LP fees via
• 16:13
  Uniswa and I'm technically earning
• 16:15
  lending fees via Aave.
• 16:19
  Inversely, I could also
• 16:21
  provide that LP to Unisuap first and
• 16:23
  then use that output token
• 16:25
  as collateral into their market
• 16:27
  because they did have LP markets
• 16:28
  back at the time.
• 16:29
  So the principle was
• 16:31
  really just how do we combine this
• 16:33
  into a singleton so that
• 16:35
  you don't need to hop between all of
• 16:37
  these extra steps and also to
• 16:39
  improve the UX because in the
• 16:40
  example where I'm taking the AVA
• 16:42
  tokens and I'm depositing that into
• 16:44
  Uniswap, You know, Uniswap itself
• 16:46
  doesn't know that.
• 16:47
  That's a wrapper for USTC and
• 16:49
  BTC. So when it quotes a route for
• 16:51
  someone, it often doesn't go through
• 16:52
  that pool because there's also extra
• 16:54
  gas fees involved with needing to
• 16:55
  unwrap the tokens.
• 16:56
  As time went on, aggregators
• 16:58
  kept integrating and got smarter
• 17:00
  and they've started adding these
• 17:02
  routes, but it still doesn't
• 17:04
  mean that the original steps that
• 17:06
  the user has to take to achieve that
• 17:07
  same goal was there.
• 17:09
  So...
• 17:11
  The idea was fairly basic
• 17:13
  in that you use the AMM as the
• 17:15
  primitive, because an AMM by nature
• 17:16
  has an asset B as a reserve
• 17:18
  A and a reserve B, so two assets,
• 17:20
  USDC, WBTC.
• 17:22
  ETH, make it easier.
• 17:24
  Trades are still routed through
• 17:26
  this, but at the same time, it can
• 17:27
  allow a third party to come in,
• 17:29
  use ETH as collateral, and borrow
• 17:31
  some of the USDC or bring USDC and
• 17:33
  then borrow some thieves.
• 17:34
  And that was really it, right?
• 17:36
  Because At that moment,
• 17:38
  you have the better capital
• 17:39
  efficiency. You have the compounding
• 17:41
  fees. These systems are aware of
• 17:42
  each other, which is also important
• 17:44
  because if, for example,
• 17:46
  a trade routes through Uniswap,
• 17:48
  a USDC, a
• 17:50
  EIF, but there isn't available
• 17:52
  liquidity that automatically reverts
• 17:53
  in that as other problems.
• 17:54
  So it's also simpler from an
• 17:56
  accounting perspective, but it's
• 17:58
  still this capital efficiency rule.
• 18:00
  So while I do think at
• 18:02
  the time it probably would have
• 18:04
  worked, I also think it would have
• 18:05
  likely been exploited.
• 18:06
  Because I also think that our
• 18:09
  knowledge in terms of
• 18:11
  a lot of the on-chain accounting,
• 18:12
  a lot or the Oracle dependencies,
• 18:15
  a lot these external factors,
• 18:17
  we're still evolving.
• 18:19
  I think they've gotten to a pretty
• 18:21
  powerful place now, but back
• 18:23
  then, a lot the implementations were
• 18:25
  still quite naive.
• 18:28
  Then, in terms of how to, so
• 18:30
  that was.
• 18:31
  One of the concepts that I wanted to
• 18:33
  incorporate into Flying Tulip.
• 18:36
  And Flying Tulips is really just
• 18:37
  this premise that how do we
• 18:39
  maximize capital efficiency
• 18:41
  on the same capital, but
• 18:43
  without needing to do any
• 18:45
  funky accounting entries or
• 18:47
  that kind of stuff.
• 18:49
  So there are,
• 18:51
  and this is a little bit pitchy, I
• 18:52
  apologize, but there are two
• 18:54
  fundamental parts to Flying Tulop.
• 18:55
  The one is the one that's currently
• 18:56
  available, which I consider
• 18:58
  to be the centralized version.
• 19:00
  And then there's the permissionless,
• 19:02
  well rather permissioned version,
• 19:04
  and then there's a fully
• 19:05
  permissionless version, but
• 19:07
  if one's permissioned and one is
• 19:08
  permissionless there are different
• 19:10
  economic trade-offs.
• 19:12
  So for the permissioned
• 19:14
  version, everything
• 19:16
  really looks like from
• 19:18
  from how you would start with a
• 19:20
  centralized exchange?
• 19:21
  With a centralized exchange, the
• 19:22
  first thing you do is you deposit
• 19:23
  into your account.
• 19:24
  And that account is actually a few
• 19:26
  things. You know, it's a margin
• 19:27
  account.
• 19:28
  It is potentially earning yield
• 19:30
  on your deposits.
• 19:31
  It is potentially allowing you to
• 19:33
  borrow against your deposits, but
• 19:34
  that's really your entry point.
• 19:36
  So. I always consider the
• 19:38
  lending market in a permissioned
• 19:40
  world as the entry point.
• 19:41
  And the reason why it specifically
• 19:42
  needs to be permissioned is
• 19:44
  because someone needs to make
• 19:46
  decisions around what collateral
• 19:48
  is accepted here.
• 19:49
  Because you can't just have anyone
• 19:50
  add any token, because otherwise I
• 19:51
  can create random token, give
• 19:53
  it a bit of Uniswap, wait until I
• 19:55
  have an oracle, then add it as
• 19:56
  collateral, and then drain the
• 19:58
  entire book. So that part has to be
• 19:59
  permissioned.
• 20:00
  But once the assets are in there,
• 20:02
  you want to do a few things.
• 20:03
  You automatically want to start
• 20:04
  allowing people to borrow against it
• 20:06
  so that you have
• 20:08
  yield already coming in just from
• 20:10
  the initial deposit.
• 20:11
  If that's all the user wants, they
• 20:12
  deposit, they earn yield time.
• 20:14
  If they want, that account still
• 20:16
  gives them a credit line that allows
• 20:17
  them to borrow again.
• 20:18
  Whatever other asset is available in
• 20:20
  it.
• 20:21
  If they wanted, they can also opt
• 20:23
  that in into a standard AMLP,
• 20:25
  where they will be taking the IL
• 20:26
  risk, but we have other things to
• 20:28
  cover for that. Or they can do the
• 20:29
  standard order book where they can
• 20:31
  just actually put limit orders on
• 20:32
  the one side.
• 20:34
  And in that case, when you have the
• 20:36
  limit order place, because that's
• 20:37
  another thing, right? In an
• 20:38
  exchange, if I place that limit
• 20:39
  order on a book, it's not
• 20:41
  doing anything. It's just statically
• 20:42
  sitting there. In our version, when
• 20:43
  it's sitting there, it still earning
• 20:44
  the yield because it's still coming
• 20:46
  from that original base premise.
• 20:48
  And then from there, we can also
• 20:49
  offer derivatives on that, you know,
• 20:51
  futures, which is just time-based
• 20:52
  expiry options, which which is just
• 20:54
  binary or perpetuals, which is.
• 20:56
  Actually a category on its own
• 20:58
  because it's a synthetic but it
• 20:59
  still plays against the same book
• 21:01
  using the same criteria.
• 21:03
  So now automatically from
• 21:05
  that same collateral you're
• 21:07
  earning third-party yield, you're
• 21:08
  earnings interest, you are earning
• 21:11
  market making fees, you are earning
• 21:12
  all of those other things.
• 21:13
  So that already gives us a lot more
• 21:15
  capital efficiency on That's all.
• 21:19
  The derivatives have not yet been
• 21:21
  launched. Oh, another very cool
• 21:22
  thing about combining all of these
• 21:24
  things that allows us to do
• 21:25
  something that I haven't seen
• 21:26
  anywhere else is Total Return Swaps.
• 21:29
  So TRS, Total Return swaps is
• 21:30
  basically the instrument that most
• 21:32
  traditional financial institutions
• 21:34
  use to get the same kind of
• 21:36
  exposure that a perpetual allows.
• 21:37
  Because it means that I don't have
• 21:39
  to buy the full thing.
• 21:40
  I can still do it at leverage and I
• 21:41
  can sell earn.
• 21:42
  Any positive performance, or
• 21:44
  I will pay if there's any negative
• 21:45
  performance.
• 21:46
  The big difference between
• 21:48
  TRS and PERPs,
• 21:50
  same leverage, same
• 21:52
  capital efficiency, but
• 21:54
  funding is often a lot cheaper
• 21:56
  because PERPs as a mechanism have
• 21:58
  to target sort of 10%
• 22:00
  flat.
• 22:01
  Year, your funding is actually the
• 22:03
  delta of your borrowing cost.
• 22:04
  So if you're using USDC as
• 22:06
  collateral, which is generally
• 22:07
  earning like 3% and you're
• 22:09
  shorting BTC
• 22:11
  for argument.
• 22:12
  Borrowing BTC is normally like 0.1%.
• 22:15
  So your net on that position is
• 22:16
  actually plus 6%.
• 22:18
  In the long counter, you're
• 22:20
  paying the 4% for the USDC
• 22:22
  borrowing. You're getting the
• 22:24
  0. 1% for the BTC collateral.
• 22:27
  So for that short position,
• 22:29
  you know, you are paying minus 4
• 22:30
  versus on most perpetual exchanges
• 22:31
  should be paying close to minus 10.
• 22:33
  And then no ADL, which I
• 22:35
  think is another important thing,
• 22:37
  because because it's a spot based
• 22:39
  solution, it doesn't require
• 22:40
  any kind of auto deleveraging,
• 22:42
  you always have your position as
• 22:44
  long as your account has margin.
• 22:46
  There are a bunch of little micro
• 22:48
  innovations there as well that I'm
• 22:49
  always happy to talk about.
• 22:50
  Our liquidations doesn't use the
• 22:52
  standard cut mechanism.
• 22:53
  We actually just look at it as a
• 22:55
  sale of an asset, so that goes
• 22:56
  through a standard RFQ approach.
• 22:57
  We take the best bid, we settle your
• 22:59
  position, so our liquidations are
• 23:01
  healthier. We also don't use loan to
• 23:02
  value. We don't look
• 23:05
  at it you are providing
• 23:07
  BTC and you are borrowing USDC.
• 23:09
  We look at it as what is your
• 23:11
  account equity and what is our
• 23:13
  maintenance requirements, which is
• 23:14
  the same as most exchanges use.
• 23:16
  Then how does that compare in terms
• 23:18
  of other solutions?
• 23:19
  I think because
• 23:22
  there's a lot more happening,
• 23:24
  there's lot more risk assumptions,
• 23:26
  and ultimately, that means something
• 23:28
  like AVE and a lot of conditions
• 23:30
  will be safer.
• 23:32
  In other times, you
• 23:34
  just want siloed exposure.
• 23:35
  You might not want the
• 23:36
  cross-collateral exposure of AVE,
• 23:38
  and then you go deposit into morpho.
• 23:39
  But then you're taking curators.
• 23:41
  So in each of
• 23:43
  these has a different vertical
• 23:45
  that I think applies to a
• 23:47
  different use case
• 23:49
  where I think
• 23:51
  we stand out, like I said,
• 23:53
  total return swaps, I think is
• 23:55
  definitely one. I think, we offer
• 23:56
  generally better yield than most of
• 23:57
  our competitors can at the same
• 23:59
  size.
• 24:01
  And then lastly is just the
• 24:03
  feature set that you get, you know,
• 24:04
  there's no needing to hop between
• 24:05
  stuff, you automatically get these
• 24:06
  things, even if you're just earning
• 24:08
  the yield, that's fine.
• 24:09
  But if you want to borrow, it's
• 24:10
  there. If you want to just do a swap
• 24:12
  into something else, it is there.
• 24:13
  If you want to do a leverage
• 24:14
  position, it is there and I
• 24:16
  think it's the same reason why
• 24:17
  someone might deposit their
• 24:19
  crypto into an exchange instead of
• 24:21
  having it into DeFi and a custodial
• 24:23
  wallet.
• 24:25
  So on a
• 24:26
  per feature level, I do
• 24:28
  believe we have differentiators that
• 24:30
  do compete with the relevant
• 24:31
  parties, but I also think that's
• 24:33
  something to be said about, that's
• 24:34
  all they offer.
• 24:36
  I mean, it's also a little
• 24:37
  bit like.
• 24:39
  Using a reference between the US and
• 24:41
  South Africa here.
• 24:42
  In South Africa,
• 24:44
  most restaurants do one thing.
• 24:46
  One will do burgers, another one
• 24:47
  will do hot dogs.
• 24:49
  And you go to the one you like.
• 24:51
  You're in the US diners are more of
• 24:52
  a thing. They do a lot of variety.
• 24:55
  And sometimes you want to just get
• 24:56
  that one item that that company
• 24:58
  does because you know it's
• 24:59
  specialized. And other times you're
• 25:00
  not really sure what you want.
• 25:01
  And then it's nice having that extra
• 25:02
  optionality.
• 25:03
  And I think that's probably the best
• 25:05
  comparison. You know, because with
• 25:07
  FT, our goal is...
• 25:09
  Is every blockchain where it's an
• 25:11
  L1, L2, doesn't matter, should look
• 25:13
  and feel and operate like a
• 25:14
  centralized exchange where you
• 25:16
  deposit, you can trade, you can
• 25:17
  swap, you borrow, you can do
• 25:18
  everything you can on the exchange,
• 25:19
  but in a non-custodial
• 25:22
  on-chain way.
• 25:23
  And that's really the premise.
• 25:24
  And that same premise gives a lot of
• 25:26
  capital efficiency that I think,
• 25:27
  that translates to better yield.
• 25:29
  And I ultimately
• 25:31
  think that's the competitor, right?
• 25:32
  It's like, what's the yield?
• 25:35
  What's the risk?
• 25:36
  And what's the available liquidity?
• 25:38
  I think those are the three
• 25:39
  questions people generally ask.
• 25:41
  Beautiful. I mean, just in terms
• 25:43
  of the context of what's happened in
• 25:45
  2026, obviously we had things like
• 25:46
  the KelpDar exploit and I think
• 25:48
  there's been a sort of general,
• 25:50
  I don't want to say in malaise
• 25:51
  because I really don't like to talk
• 25:52
  negatively about the industry and I
• 25:54
  don't think that it serves
• 25:56
  anything to paint
• 25:58
  it in that way, but it's definitely
• 25:59
  not been easy, right, for DeFi.
• 26:01
  And we've had Stanley Kulikov on the
• 26:03
  show and a number of other big names
• 26:04
  in the DeFi industry, but
• 26:06
  doing something a little bit
• 26:07
  different, innovating, trying new
• 26:09
  things, that kind of seems where
• 26:10
  you're taking this.
• 26:11
  And it does seem that there's merits
• 26:13
  in that, but your
• 26:15
  general outlook for DeFi going
• 26:17
  into the end of 2027,
• 26:20
  do you think a lot of people are
• 26:21
  learning a lot of hard lessons, but
• 26:23
  we will still see DeFi
• 26:25
  continue to attract a
• 26:27
  lot of capital, attract a lot of
• 26:29
  investment and still be a future
• 26:31
  infrastructure that's kind of better
• 26:32
  than what TradFi offers us?
• 26:35
  100%. I don't doubt
• 26:37
  that statement for a second.
• 26:41
  Now again,
• 26:43
  where I would like to make a
• 26:45
  qualifier is I think a lot of
• 26:47
  people always have this vision that
• 26:48
  DeFi is somehow supposed to like
• 26:50
  be bigger and or replace traditional
• 26:52
  finance. That has never been my
• 26:53
  premise. I do not think it is true.
• 26:56
  I think
• 27:00
  I prefer calling it on-chain finance
• 27:01
  nowadays. That's a different
• 27:02
  discussion. But I think on-chain
• 27:04
  finance is a
• 27:06
  niche for a
• 27:08
  user that is willing to take on a
• 27:10
  little bit of extra complexity for
• 27:12
  the trade-offs.
• 27:13
  Those trade-off being you control
• 27:15
  the assets, but that comes with the
• 27:17
  responsibility of the fact that you
• 27:18
  control assets.
• 27:19
  There's no one else that can help
• 27:20
  you when something bad happens.
• 27:22
  So my general recommendation is,
• 27:25
  would I let someone's
• 27:27
  parents try and do DeFi?
• 27:30
  No. No, I'd automatically tell them
• 27:31
  no. And like most of even my
• 27:33
  peers that aren't technically
• 27:34
  orientated, I tell them, no, do not
• 27:35
  get into the space, keep your stuff
• 27:37
  in your bank account. Do not come
• 27:38
  here.
• 27:39
  For that technically inclined subset
• 27:41
  that might be frustrated
• 27:43
  with a lot of the sort of
• 27:45
  traditional mechanism, for them, I
• 27:46
  highly recommend it.
• 27:47
  But that's not
• 27:49
  the total addressable market of
• 27:51
  finance. That's a very small subset
• 27:53
  there are.
• 27:54
  Do I still think That's big enough.
• 27:56
  That it still validates all of this?
• 27:58
  Yes, no, a hundred percent.
• 27:59
  There's enough people in there and
• 28:00
  over time people get more and more
• 28:01
  technically implying as well.
• 28:03
  And that also allows them
• 28:04
  to come into this sort of safer
• 28:06
  space, safer.
• 28:09
  So a hundred per cent believe it's
• 28:11
  going to keep growing.
• 28:12
  I do think what's
• 28:15
  often natural in
• 28:17
  cycles is that people
• 28:19
  overcorrect for a specific element.
• 28:21
  And what I think happened, you know,
• 28:22
  20.
• 28:24
  2021, maybe let's say 2022,
• 28:26
  everything was smart contracts
• 28:28
  exploits.
• 28:29
  So we over-corrected where
• 28:30
  all of the focus, all of
• 28:32
  security, everything, was like
• 28:34
  100% smart contracts, smart
• 28:35
  contracts. To the point where
• 28:38
  it almost started neglecting
• 28:40
  the traditional security risks.
• 28:42
  Because if we look at almost all of
• 28:43
  the exploits that have happened over
• 28:45
  the last year, most of them weren't
• 28:46
  smart contracts.
• 28:47
  The vast majority were third-party
• 28:50
  infra that wasn't secured or keys
• 28:52
  that were actually committed
• 28:53
  somewhere or a specific individual
• 28:55
  that was compromised.
• 28:56
  And that's not blockchain.
• 28:58
  That's just general opsec
• 29:00
  security that companies need to go
• 29:02
  through.
• 29:03
  And you can clearly see
• 29:05
  everyone's getting a little bit of a
• 29:06
  wake-up call that, hey, the
• 29:08
  blockchain is a very important
• 29:09
  component because that's obviously
• 29:10
  where the assets are out.
• 29:12
  But as we're, and this is
• 29:14
  also a bit of a natural evolution
• 29:15
  because, and, this actually ties
• 29:17
  into why I prefer calling it
• 29:18
  on-chain finance and not
• 29:19
  decentralized finance, because, you
• 29:20
  know, decentralized finance had the
• 29:22
  promise of, it's you and the
• 29:23
  contract, no one else.
• 29:25
  There's no further counterparty.
• 29:27
  For it to be decentralized, it has
• 29:29
  to be only you and that contract.
• 29:31
  It can't be upgraded,
• 29:33
  rules can't be changed, it
• 29:35
  needs to be static, it needs be
• 29:37
  immutable, and there can't be a
• 29:38
  third party. So that's decentralized
• 29:40
  finance.
• 29:41
  What we have today is on-chain
• 29:43
  finance. There's a company,
• 29:45
  there's employees, they can upgrade,
• 29:47
  they can change, they could modify,
• 29:48
  they often set the rules.
• 29:52
  It is something else.
• 29:54
  And I think we often still try and
• 29:55
  push it into that decentralized
• 29:56
  finance bucket.
• 29:57
  And to be fair, for decentralized
• 29:59
  finance, the only counterparty is
• 30:00
  the contract. So the only thing that
• 30:02
  matters is the contracts security.
• 30:04
  And that's also maybe why we
• 30:05
  overcorrect it. But nowadays where
• 30:06
  you have upgradable contracts, you
• 30:08
  have off-chain readings, you have
• 30:10
  people making risk-based decisions,
• 30:11
  you can't call that that anymore and
• 30:13
  that is why the contract is no
• 30:15
  longer the most important component.
• 30:17
  It's still probably top
• 30:19
  one in terms of the list but it's
• 30:21
  not the only thing you can look at
• 30:23
  anymore.
• 30:25
  I mean, it's obviously horrible
• 30:27
  that people are losing money, and I
• 30:28
  wish that was never the case.
• 30:30
  But I also, unfortunately,
• 30:31
  think it is a natural part of the
• 30:33
  evolution of the process.
• 30:36
  When it comes to designing some
• 30:38
  of the elements of Flying Chula,
• 30:40
  you've got FTUSD,
• 30:42
  which is a stablecoin with yield.
• 30:45
  What kind of lessons have you
• 30:46
  taken from the past?
• 30:48
  Not to draw comparisons to
• 30:50
  things like
• 30:53
  the collapse of DoKuan's
• 30:55
  stablecoin project and that kind of
• 30:57
  thing, but people will look at these
• 30:58
  things in the past and Like, have
• 31:01
  we got the right infrastructure here
• 31:03
  to...
• 31:04
  Do these things. And then the second
• 31:05
  point to that question is,
• 31:07
  now that we have very good
• 31:09
  legislation,
• 31:11
  the Genius Act and obviously
• 31:12
  Clarity, hopefully getting passed
• 31:14
  in September, does it make it a
• 31:16
  lot easier to offer these kind of
• 31:18
  services to people?
• 31:21
  I mean, in terms of the regulation,
• 31:23
  I'd say it makes it harder, but
• 31:24
  that's not a bad thing, you know, it
• 31:26
  just means that there are certain
• 31:28
  checks and balances that need to be
• 31:29
  adhered to. And I think that
• 31:30
  ultimately helps, right?
• 31:32
  Because what
• 31:34
  I think people care about, and
• 31:36
  it's a little bit unfair to put this
• 31:38
  on the institutions, actually,
• 31:39
  because they're not, they're
• 31:41
  more of an enforcement entity than
• 31:43
  they are a stamp
• 31:45
  of approval entity.
• 31:46
  But you know, why
• 31:48
  do people deposit into
• 31:50
  an FDIC insured account over
• 31:52
  a non-insured account?
• 31:53
  It's because that stamp gives you a
• 31:55
  little bit of like comfort.
• 31:58
  The same reason why technically,
• 32:00
  you know all of us exchanges, myself
• 32:02
  included, need to actually have an
• 32:03
  ATS license, you know, because that
• 32:06
  means we've gone through that
• 32:08
  rigor. And I'd love to, by the way,
• 32:09
  I'm very excited about the new
• 32:11
  proposals that came out yesterday
• 32:12
  because that finally gives a path.
• 32:14
  Towards it, which is absolutely
• 32:15
  awesome.
• 32:17
  The Hester Pearson, the new SEC are
• 32:18
  absolutely crushing it.
• 32:21
  So There are
• 32:23
  people that want those things.
• 32:25
  And we understand that for us to
• 32:26
  get those things, we need to jump
• 32:28
  through certain hoops.
• 32:29
  Like to be clarity compliant, there
• 32:31
  are going to be additional hoops
• 32:33
  that we have to jump though.
• 32:34
  We have to be able to prove we are
• 32:35
  compliant. And then we have be able
• 32:36
  to continuously prove we're
• 32:38
  compliant. That does increase
• 32:40
  overhead. So that cancels out a lot
• 32:42
  of the sort of old, I mean, most of
• 32:43
  the stuff I bought in
• 32:45
  the original DeFi era wouldn't
• 32:48
  be able because...
• 32:50
  You need millions of dollars of
• 32:51
  fundings, unfortunately, to be able
• 32:53
  to get through this stuff.
• 32:54
  It is expensive.
• 32:55
  But the expense is part of almost
• 32:57
  the security in that you don't
• 32:59
  want someone that can't afford those
• 33:01
  things to really be offering these
• 33:02
  solutions.
• 33:03
  That's why there's minimum
• 33:05
  requirements to banking license,
• 33:07
  EMA license, finance license.
• 33:08
  There's a minimum amount you need to
• 33:09
  keep somewhere in a bank account for
• 33:10
  these people to be like, cool, they
• 33:11
  have the money if something goes
• 33:12
  wrong.
• 33:14
  And so it makes it harder, but
• 33:16
  it makes better.
• 33:17
  Because now you as a user.
• 33:19
  We'll be able to see, hey, OK, this
• 33:21
  thing is actually Clarity compliant.
• 33:23
  And if you are a user and you see a
• 33:24
  website saying it's Clarity-compliant,
• 33:25
  please just go check that on
• 33:27
  the actual Clarity website.
• 33:28
  Because there's going to be a lot of
• 33:29
  websites that say they're compliant,
• 33:30
  and they're not.
• 33:33
  And that in itself is already a
• 33:35
  better step, right?
• 33:36
  I think a big problem that
• 33:38
  a lot of founders,
• 33:40
  myself included, had with the
• 33:41
  previous SEC is they're telling us
• 33:45
  that they were giving a little bit
• 33:46
  of conflicting messages, because
• 33:47
  they were like, you have to comply
• 33:48
  to do this, but there's also no way
• 33:50
  to comply.
• 33:51
  And that's just an impossible
• 33:52
  situation to be in, versus the new
• 33:54
  one is saying, we've
• 33:55
  really looked at this.
• 33:57
  Here are the rules.
• 33:58
  They are stringent.
• 33:59
  They're not going to be easy, but it
• 34:00
  is an actual path, right?
• 34:02
  Because previously, to get the ATS
• 34:04
  license, and I went through this
• 34:05
  before launching FT,
• 34:07
  to the Ats license, almost 99%
• 34:10
  of the stuff is focused on the
• 34:11
  broker-dealer.
• 34:12
  But in these solutions, you don't
• 34:14
  have a broker-deal. The contracts
• 34:15
  are the broker, and the contracts
• 34:16
  can't be compliant.
• 34:17
  And if we do a third party,
• 34:19
  they're not comfortable with the
• 34:20
  contracts. So there was still this
• 34:22
  tension.
• 34:23
  With the new proposals, and I hope
• 34:25
  they all pass, most
• 34:27
  of these tensions are removed.
• 34:29
  There are still some, don't get me
• 34:30
  wrong. But I think friction is
• 34:32
  a natural part of
• 34:34
  security and defensiveness from
• 34:35
  users as well.
• 34:36
  Because if it was
• 34:38
  too easy, then every
• 34:41
  team just falls into form, and
• 34:43
  they're all of a sudden compliant.
• 34:44
  So there also has to be a little bit
• 34:45
  of a friction, which is a testing
• 34:46
  process.
• 34:48
  Anyway, I think I've gone off on an
• 34:49
  entire different tangent to your
• 34:50
  original question.
• 34:53
  Well, then I'm glad, because I'll
• 34:55
  have to pull myself back a little
• 34:57
  bit, because this is definitely
• 34:57
  another topic that I can dig into
• 34:59
  a bit too much.
• 35:01
  No, no, no. That was perfect.
• 35:02
  I think you answered it quite well
• 35:03
  that, you know, there is from
• 35:05
  a, from a regulatory perspective,
• 35:07
  there's two sides of the coin there.
• 35:09
  And I think that this
• 35:11
  is going definitely in the right
• 35:13
  direction. And we do hope that a lot
• 35:14
  of the things put down on paper, you
• 35:16
  know, serve, serve that purpose.
• 35:20
  Overall, are you, I mean,
• 35:22
  can you talk to us a little bit
• 35:22
  about moving to the U.S.
• 35:24
  As opposed to just doing this
• 35:25
  globally anywhere and why you
• 35:27
  actually wanted to have FTE?
• 35:29
  I mean, as you said, you're based in
• 35:31
  New York, you are there now.
• 35:33
  Is it just because of the general
• 35:35
  change in the regulatory
• 35:37
  landscape there and also just
• 35:39
  access to capital the size of the
• 35:40
  market there?
• 35:41
  Why would you want to be based in
• 35:42
  the US to get this off the ground
• 35:44
  properly?
• 35:45
  I mean, I think crypto
• 35:48
  as all can probably agree, even
• 35:50
  if we didn't like it historically,
• 35:52
  that the US has become the
• 35:53
  unofficial sort of
• 35:56
  watchdog.
• 35:57
  So that means
• 35:59
  CFTC, ACC,
• 36:01
  that's where you need to start
• 36:02
  having these conversations and
• 36:04
  that's why they're based.
• 36:05
  Um, and originally that was my
• 36:07
  entire premise and, and, and kudos
• 36:09
  to, you know, those teams because
• 36:11
  they, they, they were truly
• 36:13
  informed. Don't get me wrong.
• 36:14
  They had, they knew, they understood
• 36:16
  the space intimately when I started
• 36:17
  having my original conversations,
• 36:19
  but they also understood the gap
• 36:21
  between what it was and what the
• 36:22
  legislation allowed.
• 36:24
  And, and that gap,
• 36:26
  the informal advice I actually
• 36:28
  received and which I'm very
• 36:30
  appreciative of is I
• 36:32
  was told not to apply at that
• 36:34
  point in time.
• 36:35
  Because if we were
• 36:37
  to apply, given the current rule
• 36:39
  set, they have to decline.
• 36:40
  And then if we launched,
• 36:43
  much bigger problem.
• 36:45
  Because then we're going in direct
• 36:46
  contrast to what the
• 36:48
  statement is.
• 36:49
  So the informal advice
• 36:51
  was, go for now.
• 36:53
  But as soon as the new updated stuff
• 36:55
  comes out, let's restart this
• 36:56
  conversation. And I'm very
• 36:57
  appreciative of that.
• 36:59
  But ultimately, that was it.
• 37:00
  You know, the capital and our
• 37:02
  funding, that was still mostly
• 37:03
  global, predominantly
• 37:05
  out of the EU, actually.
• 37:09
  In terms of access and networking,
• 37:11
  I mean, this is also the place to
• 37:13
  be. Almost all of the big companies
• 37:14
  are here.
• 37:15
  I am
• 37:17
  antisocial to the point where I'm
• 37:18
  basically a hermit, so that's not
• 37:20
  really beneficial for me.
• 37:21
  But in theory, it's very
• 37:22
  beneficial if I ever actually leave
• 37:25
  the computer alone, which
• 37:27
  realistically isn't going to happen.
• 37:30
  But in theory, that was another good
• 37:32
  reason.
• 37:34
  So ultimately, if you want to start
• 37:36
  a crypto-based company, I'd say
• 37:37
  the US is still the place to be.
• 37:39
  And for the reasons you said,
• 37:40
  capital formation, access to
• 37:42
  the networks that are here already
• 37:44
  and existing, and then
• 37:46
  to be able to comply with
• 37:47
  the regulatory requirements, which I
• 37:49
  think ultimately is going
• 37:51
  to happen.
• 37:52
  Yeah, yeah, we've we've had some
• 37:54
  interesting conversations with with
• 37:56
  different voices from the industry
• 37:57
  in the past couple of weeks You
• 37:58
  know, obviously clarity got pushed
• 37:59
  out to next month you
• 38:01
  know people are hopeful that the the
• 38:03
  Senate passes that but
• 38:05
  You know as GSR's
• 38:07
  chief chief of legal told us
• 38:09
  a little bit earlier this week It's
• 38:10
  it's very much just a political
• 38:11
  conversation and my mind kind of
• 38:13
  wonders, you know if in two years
• 38:16
  You know things change drastically
• 38:18
  in the US from a government at a
• 38:19
  governmental level what
• 38:22
  would be the possibility that a lot
• 38:23
  of the hard work to drive
• 38:25
  forward innovation and help the
• 38:27
  industry thrive, you know,
• 38:30
  starts getting undone?
• 38:31
  Do you think there's any possibility
• 38:32
  that that happens or do you think
• 38:34
  that pragmatism wins after and
• 38:35
  the foundation has been laid for
• 38:37
  basically the future of the industry
• 38:39
  indefinitely and a lot this can't be
• 38:41
  undone.
• 38:43
  I'd actually say that's more true
• 38:45
  for the EU currently than I would
• 38:46
  say for the US, in that the
• 38:48
  EU's laws are already laws, you
• 38:50
  know, so there you know MICA etc
• 38:52
  are pretty strong.
• 38:55
  Year, it really does all
• 38:57
  depend if it gets passed to
• 38:58
  legislation, if it gets passed the
• 39:00
  legislation it'll stick, it'll stay.
• 39:02
  If it stays a voting topic
• 39:04
  and there's a
• 39:05
  a pre-current SEC administration
• 39:08
  shift, then I think it's dead.
• 39:12
  I mean, again, I do
• 39:14
  hate bringing things back to
• 39:16
  politics, but if
• 39:19
  the House or the Senate currently
• 39:21
  changes and these things get blocked
• 39:23
  up even more for the next two years
• 39:25
  and nothing passes,
• 39:27
  that's maybe the worst case
• 39:28
  scenario.
• 39:29
  It still obviously depends on who
• 39:31
  the next administration is,
• 39:34
  because they might share something
• 39:35
  similar. If nothing else, I think
• 39:37
  there has been...
• 39:41
  I wanted to say there was a positive
• 39:42
  shift towards crypto, but I have to
• 39:44
  admit, if I look at the new from
• 39:46
  institutions, I definitely think
• 39:47
  there's a positive ship from from
• 39:48
  retail sentiment. I definitely think
• 39:49
  that's a more negative shift because
• 39:51
  they they now more associated
• 39:53
  with grifting and scams and
• 39:54
  exploitation.
• 39:56
  But I mean, if we go back
• 39:57
  to your new segment, almost all
• 39:59
  those items were about
• 40:00
  institutionals doing something with
• 40:02
  crypto. So I think they've seen
• 40:04
  it, they like it, and they want to
• 40:05
  use it.
• 40:06
  It's another avenue for them, it's
• 40:08
  another source of volume for them.
• 40:09
  It is another source of
• 40:10
  diversification for them so I think
• 40:16
  If they keep backing it
• 40:18
  because it is more profitable for
• 40:19
  them, I think we're in a good point.
• 40:21
  I think if the legislation
• 40:23
  doesn't pass and
• 40:26
  new administration wants to
• 40:28
  be sent, well, not new, but
• 40:29
  potential administration wants, to
• 40:31
  send a very strong signal over we
• 40:32
  won't tolerate some of the stuff
• 40:34
  we've seen over the last four years,
• 40:35
  then I think, we're back to square
• 40:36
  one.
• 40:38
  But you can't really make your
• 40:40
  decisions on that, you know, as a
• 40:42
  builder. You kind of just have to
• 40:43
  bold it and then.
• 40:45
  As any builder or founder
• 40:47
  knows, there's a new unpredicted
• 40:48
  event every week that you need to
• 40:50
  deal with. So you just need to
• 40:52
  keep up to date, try
• 40:54
  and be as compliant
• 40:56
  and forthcoming
• 40:58
  as you can
• 41:00
  and then deal with whatever happens.
• 41:03
  Yeah, I want to pull it
• 41:05
  back to Flying Tulip and I guess
• 41:06
  just the DeFi ecosystem in
• 41:08
  particular and how you go about
• 41:10
  attracting users to your platform.
• 41:12
  Obviously, there's a lot of big
• 41:13
  incumbents and most people feel safe
• 41:15
  to use what they know and I
• 41:17
  was just looking at one of your
• 41:18
  tweets, I think, yeah, from,
• 41:20
  is it from earlier today?
• 41:23
  No, a couple of days ago.
• 41:24
  So it says here FT lend over $15
• 41:27
  million, FT USD just
• 41:29
  under $5 million,
• 41:31
  fees are Revenues up, buy back.
• 41:32
  Over $1,000 a day,
• 41:34
  USD earning at 9.29%.
• 41:37
  When it comes to the actual yield
• 41:39
  rates, is this like
• 41:41
  make or break in DeFi?
• 41:43
  And do you think that's a real
• 41:45
  kicker when it comes getting people
• 41:47
  to actually use your platform as
• 41:49
  opposed to deploying capital in
• 41:51
  other places?
• 41:53
  My personal opinion is that time
• 41:55
  and market is the most important
• 41:56
  thing. What I mean by that is the
• 41:59
  risk assumption for flying tulip
• 42:00
  right now are incredibly high.
• 42:02
  9% healed is probably not
• 42:04
  enough to offset the risk
• 42:06
  because you're talking about a
• 42:07
  protocol that's six months old.
• 42:09
  Obviously, we take every proactive
• 42:11
  step we can towards security, but
• 42:12
  it's six-months-old.
• 42:13
  We've seen bigger protocols
• 42:15
  that have been in the market for
• 42:16
  longer that also got exploited.
• 42:20
  I mean, there's a reason why people
• 42:21
  are fine with 3% on Aave, but they
• 42:23
  might want, you know, 7%
• 42:25
  in a specific morpho-curated faults,
• 42:27
  right? Because that
• 42:29
  differential, that R,
• 42:31
  that is, it's not really an R
• 42:33
  because it's very different risk
• 42:34
  assumptions versus for
• 42:36
  us, I think that's even more.
• 42:38
  I mean realistically, that number
• 42:40
  to offset
• 42:42
  our risk profile would probably need
• 42:44
  to closer to like 30 percent.
• 42:46
  We're not going to get to 30%
• 42:48
  because then you start doing, you
• 42:49
  know, dodgy, fugazi stuff in the
• 42:51
  background. Like I'm proud of the
• 42:52
  APYs we can offer given that
• 42:54
  everything is on chain, there's no
• 42:55
  third party, you, know, it's
• 42:58
  there verifiable, there's
• 43:00
  no one else that you have to trust
• 43:01
  with some credit or something.
• 43:03
  Anyway, so I'm proud of that number
• 43:04
  but I also recognize that that
• 43:06
  number is too low for how long
• 43:08
  we've been around and for how
• 43:10
  much liquidity we have but that
• 43:12
  is something that only gets with
• 43:13
  more time. So now it's about being
• 43:16
  consistent with that number and
• 43:17
  being consistent with our security
• 43:19
  and that
• 43:21
  brings the comfort over time that
• 43:23
  so in in two years from now
• 43:25
  and if our AUM
• 43:27
  is you know instead of 20
• 43:29
  it's two billion then
• 43:31
  nine percent's a fantastic number
• 43:33
  and that's ultimately the
• 43:35
  difference that you have.
• 43:36
  So to me now I'm
• 43:38
  not trying to push this on anyone,
• 43:40
  I'm not trying, to tell them hey
• 43:41
  come here you know people need to
• 43:43
  understand what the risk profile is.
• 43:46
  Um, anecdotally, if you're curious,
• 43:48
  that's actually where tested product
• 43:50
  originally came from.
• 43:50
  It wasn't, it wasn't some call
• 43:52
  to action for degenerates and it
• 43:54
  wasn't some yolo it
• 43:56
  was actually there'd be dragons
• 43:58
  here. There's risk, please be
• 44:00
  careful. Like these are new
• 44:02
  primitives that we are playing with.
• 44:04
  Like this is basically as
• 44:06
  close to tasting and prod as you
• 44:08
  can, even though it's a production
• 44:10
  product, you know, um,
• 44:13
  And it was actually originally meant
• 44:14
  as a warning.
• 44:15
  It actually annoys me a little bit
• 44:17
  that it ended up becoming basically
• 44:18
  a YOLO
• 44:21
  data site.
• 44:23
  So to me, it's always been important
• 44:24
  that people understand the risk that
• 44:26
  comes, the risk profile, the
• 44:27
  dangers.
• 44:28
  And that's why we've also been
• 44:31
  very cautious in our reporting.
• 44:33
  That's why don't want to
• 44:35
  set out any specific dates or times
• 44:37
  that we're doing stuff.
• 44:39
  I acknowledge that stop speculation
• 44:41
  and that probably does annoy a lot
• 44:42
  of our investors.
• 44:44
  I accept that and I wish I could
• 44:45
  do something for them, but at the
• 44:47
  same time, I'd rather take this very
• 44:49
  pragmatic, slow, secured approach
• 44:50
  where we're focused on what
• 44:52
  do we look like in two years from
• 44:54
  now and not what do you look like in
• 44:55
  one month from now? Because I do
• 44:56
  think a lot protocols get stuck in
• 44:58
  that.
• 44:59
  And I can't blame them, right?
• 45:00
  Because the social sentiment is if
• 45:02
  you launch and you're not at a
• 45:04
  billion dollars, you're a failure.
• 45:05
  But I mean, that's a completely
• 45:06
  ridiculous premise to
• 45:08
  start off with. So a lot of
• 45:10
  teams try and optimize for that with
• 45:12
  their launch, with their listings,
• 45:14
  with their points, with all of these
• 45:15
  other things.
• 45:17
  But you're setting yourself up
• 45:19
  for failure from day one then,
• 45:21
  because then you're always trying to
• 45:22
  chase that chart that will
• 45:24
  decrease over time as your system
• 45:25
  matures, instead of having a very
• 45:27
  small one that matures over time.
• 45:29
  Yeah, so
• 45:31
  Again, I don't know if that
• 45:32
  appropriately answers your question,
• 45:33
  but we're not really trying to go
• 45:35
  out and push this stuff, we're
• 45:37
  trying to now prove consistency.
• 45:40
  Yeah, I think you know your
• 45:42
  answer says a lot about you and how
• 45:43
  long you've been here because there
• 45:44
  are a lot of people That will be
• 45:46
  shouting from the rooftop what they
• 45:48
  you know What yields they're
• 45:49
  offering and telling everyone to
• 45:50
  come running to your platform and I
• 45:52
  also appreciate you You
• 45:54
  know explaining the test and prod
• 45:55
  thing because it was something I
• 45:56
  wanted to ask you about and you know
• 45:59
  I always looked at it as you
• 46:00
  literally saying to everyone just be
• 46:02
  careful out there you know, but it
• 46:03
  did become this kind of internet
• 46:05
  meme that you were known for but
• 46:08
  All these years later, as you say,
• 46:09
  you're trying to be pragmatic and
• 46:11
  build something very slowly and
• 46:13
  carefully.
• 46:15
  And look, I know we've been chatting for
• 46:16
  a long time, so we'll probably start
• 46:17
  winding down the conversation in the
• 46:18
  next couple of minutes. But I did
• 46:19
  just want to chat to you very
• 46:20
  quickly about the AI tools
• 46:22
  that we now have at our disposal.
• 46:24
  It's changed so quickly in the last
• 46:26
  year.
• 46:27
  You've built so many great sort of
• 46:29
  D5 primitives, I think,
• 46:31
  starting with Yearn and all these
• 46:32
  other things. How
• 46:34
  much How many
• 46:36
  of these AI tools and frontier
• 46:38
  models are you using?
• 46:39
  Has it helped your life a lot?
• 46:41
  Is it helping you figure out some
• 46:42
  things that maybe in the past you
• 46:43
  weren't able to?
• 46:45
  Can you talk to us a bit about how
• 46:46
  AI is helping you?
• 46:48
  Well, I
• 46:51
  find its creative process is still
• 46:52
  quite limited.
• 46:53
  So it's not that great at
• 46:55
  giving ideas or giving insights.
• 46:57
  It'll just regurgitate for you
• 46:59
  what's kind of available there.
• 47:00
  So when it comes to the purely
• 47:02
  innovation creative side,
• 47:04
  it doesn't exist.
• 47:06
  That being said,
• 47:09
  I distinctly remember
• 47:11
  this journey, because I've been
• 47:12
  using these tools for almost
• 47:13
  two years And I, it used
• 47:15
  to be a.
• 47:16
  Once a month, I would log in,
• 47:18
  I'd give it like a simple task,
• 47:20
  coding task or something to do.
• 47:22
  It would hallucinate or be
• 47:23
  absolutely garbage and I'd stop.
• 47:25
  And then next month, I'd revisit and
• 47:26
  it will get a little bit better, but
• 47:27
  not really much.
• 47:29
  And that process,
• 47:31
  probably even up until we launched,
• 47:33
  you know, 100% of our
• 47:35
  code was human written.
• 47:37
  And that was what?
• 47:39
  That was January of this year.
• 47:40
  So, you know, it's not that long
• 47:41
  ago. And at that point in time,
• 47:44
  none of the models were good enough
• 47:45
  to really be able to deliver on
• 47:47
  our stuff.
• 47:49
  Fast forward to around,
• 47:51
  I want to say February, March,
• 47:53
  and all of a sudden, these tools
• 47:55
  still aren't writing code, but
• 47:56
  they're helping us with reviews,
• 47:58
  with test cases,
• 48:00
  with checking for interdependencies
• 48:02
  with different systems, and they're
• 48:04
  starting to get really bloody good
• 48:05
  at it. To give an example,
• 48:07
  when we used to ship off to an
• 48:09
  audit, it might come back with
• 48:11
  let's say 15 findings in total,
• 48:14
  split between five medium,
• 48:16
  maybe five low, five informational.
• 48:19
  Post all of these extra tools
• 48:21
  integrated into all of our
• 48:22
  pipelines, our testing, our CI, all
• 48:24
  of those things, that goes down to
• 48:26
  between one to two informational
• 48:27
  findings. Now that's already a huge
• 48:29
  difference.
• 48:30
  Fast forward another month
• 48:32
  and all of a sudden they're starting
• 48:33
  to contribute probably like five to
• 48:35
  10% of the code.
• 48:37
  Fast-forward to where we are now,
• 48:38
  I'm reviewing them.
• 48:39
  I like like their code is better
• 48:41
  than I accept AI currently
• 48:43
  is a better coder than I am and
• 48:45
  probably better than i ever was I
• 48:48
  this took me a while just
• 48:49
  psychologically to kind of make
• 48:50
  peace with because you're talking
• 48:52
  about decades of experience that
• 48:54
  this thing in a few months
• 48:56
  caught up to and and
• 48:58
  it's scary it's terrifying but at
• 48:59
  the same time it's soon.
• 49:01
  We now have an internal policy that
• 49:03
  almost everyone has to vet their
• 49:04
  stuff against specific models.
• 49:06
  I will say the gap between models
• 49:08
  and even in a specific model
• 49:09
  the gap between their different
• 49:11
  settings is still huge.
• 49:14
  The difference between
• 49:16
  what I'm going to use as an example
• 49:17
  codecs because that's one of our
• 49:18
  streams.
• 49:19
  The difference between a 5.3
• 49:21
  model versus a 5,5 is
• 49:23
  worlds apart. And then even a 5
• 49:25
  point 5 instant versus a max is
• 49:27
  worlds a part.
• 49:28
  So we have very strict policies on
• 49:29
  which models on what settings can be
• 49:31
  used to do things.
• 49:33
  But if you use those right,
• 49:35
  it enables.
• 49:37
  I can now be working on five
• 49:39
  different work streams at the same
• 49:40
  time and then review the output post
• 49:42
  that and then agree.
• 49:43
  And if I'm being honest, they often
• 49:45
  catch things that I missed in my
• 49:46
  original designs and views and
• 49:48
  things.
• 49:50
  So no, the change has been
• 49:51
  astronomical in
• 49:53
  both good and bad directions
• 49:55
  because it obviously also, I
• 49:57
  think we will still see a lot of
• 49:59
  kind of vibe coded bugs coming
• 50:00
  through because a
• 50:02
  lot people are skipping that
• 50:04
  human review process, which I do
• 50:05
  still think is very important.
• 50:06
  It's boring, but it's very
• 50:08
  important, and I will admit AI code
• 50:09
  is difficult to read sometimes.
• 50:10
  They don't
• 50:13
  write for a human reader, you know,
• 50:15
  so often
• 50:17
  it's harder to
• 50:19
  to code review or peer review what
• 50:21
  a bot did than what another human
• 50:23
  did.
• 50:24
  But again, these things are changing
• 50:25
  as time is going on.
• 50:27
  But I mean, hopefully that gives you
• 50:28
  a picture of just how quickly the
• 50:30
  last eight months changed in terms
• 50:31
  of these models.
• 50:32
  Yeah, it has.
• 50:33
  And I mean, uh, both my brothers are
• 50:35
  developers, one front end, one back
• 50:37
  end. And I think, you know, the
• 50:38
  conversations I had with them
• 50:39
  offline a lot was, you know,
• 50:41
  you guys should try this and try
• 50:42
  that. And they're like, you're very
• 50:43
  hesitant for a long time.
• 50:44
  And then, you, the last few months
• 50:46
  has kind of been like the same sort
• 50:47
  of. Yeah.
• 50:49
  You know, capitulation and well, I
• 50:50
  don't want to say capitulation, but
• 50:51
  like, You know this acceptance that,
• 50:53
  okay, this, this stuff is pretty
• 50:54
  good. But I would say the
• 50:56
  advantage that you have over a
• 50:58
  layman like me, like if I vibe coded
• 51:00
  some sort of an app or something
• 51:01
  like that.
• 51:02
  I have no technical capability
• 51:04
  to review what it's done versus
• 51:06
  you who does and can actually look
• 51:08
  at it and be like, okay, well, this
• 51:09
  is great, but actually XYZ
• 51:11
  is not good enough and we need to go
• 51:13
  back.
• 51:14
  But also, to be specific
• 51:16
  on that statement,
• 51:18
  our code deals with money, so it
• 51:20
  needs quite a bit of extra scrutiny.
• 51:22
  For the average person making a
• 51:23
  static website or a game or
• 51:24
  something that won't negatively
• 51:26
  impact the user, the
• 51:28
  worst thing that can happen is the
• 51:29
  website doesn't lose or the app
• 51:31
  crashes. Go for it.
• 51:33
  There's no reason you need technical
• 51:34
  ability there.
• 51:36
  There, you can just go with the
• 51:37
  vibes. But obviously, if it
• 51:39
  touches money or there's a credit
• 51:41
  card involved, you know, whether
• 51:42
  it's crypto or TradFly.
• 51:43
  Different story.
• 51:45
  Don't go down that route at all
• 51:46
  then.
• 51:47
  Yep. Yeah, exactly.
• 51:48
  Yeah, you really have to be humble,
• 51:50
  go ahead. There is one comment I
• 51:51
  want to make though, this is not
• 51:52
  crypto related, this is AI related,
• 51:55
  is if you are using these tools
• 51:57
  please maximize and
• 51:59
  enjoy them now because these $200
• 52:01
  subscriptions or whatever it is will
• 52:03
  go away.
• 52:06
  The cost of what
• 52:08
  these models are doing for us
• 52:09
  currently versus what we're paying
• 52:11
  is completely removed from
• 52:12
  each other. We should be paying...
• 52:15
  Salaries of full-time employees.
• 52:18
  It should be hundreds of thousands
• 52:19
  of dollars.
• 52:20
  Meanwhile, it's like a $200
• 52:22
  check.
• 52:23
  And I refuse to believe that's
• 52:25
  going to keep being the status quo.
• 52:27
  So we're actually working off of the
• 52:29
  assumption that, and that's why
• 52:30
  we're still trying to add a lot of
• 52:32
  human elements into our processes,
• 52:33
  because my assumption is
• 52:35
  that we are going to be using less
• 52:37
  of this somewhere in the future.
• 52:38
  It's going get throttled.
• 52:40
  So we don't want to make ourselves
• 52:41
  too dependent now on that process
• 52:43
  because then the transition or
• 52:45
  period is going to be worse.
• 52:47
  So the general instruction is, you
• 52:48
  know, the team still has to have
• 52:49
  ownership of anything that ships,
• 52:51
  but use it as
• 52:53
  much for now to expedite your
• 52:55
  processes until it goes away.
• 52:57
  I know, I'm not going to name this
• 52:58
  person, but they might be listening
• 53:00
  to the show and they have like,
• 53:02
  I think three or four Claude Max
• 53:03
  subscriptions because the
• 53:06
  value for money for the tokens that
• 53:08
  you're getting is actually better if
• 53:09
  you just do it that way.
• 53:11
  And I do think you're right.
• 53:12
  We're going to get throttled at some
• 53:13
  point. And to be honest, I found
• 53:15
  it very interesting online.
• 53:16
  There were some people saying that
• 53:18
  OpenAI and Anthropic
• 53:20
  have already kind of nerfed the
• 53:22
  models that they're giving us access
• 53:23
  to. And you'll see like the Bitcoin
• 53:26
  red team that's been reviewing
• 53:29
  all of the Bitcoin open source code
• 53:31
  and disclosing
• 53:33
  those vulnerabilities to the
• 53:35
  associated teams to patch have been
• 53:36
  using all
• 53:38
  the frontier Chinese models
• 53:40
  because they actually allow you to
• 53:41
  go and find all these bugs and
• 53:43
  stuff. And those aren't nerfed yet,
• 53:44
  but at what point does everything
• 53:46
  get nerfed and then we're just left
• 53:47
  with the average stuff and all
• 53:49
  the military grade, you know.
• 53:51
  AI models are kept for
• 53:54
  the select elite few.
• 53:55
  I worry about a future like that.
• 53:59
  I don't think it's a worry, I think
• 54:01
  it is a guarantee.
• 54:03
  We see it on all technologies, like
• 54:05
  the GPS we use versus the GPS that
• 54:07
  the military uses are fundamentally
• 54:09
  different in design and capability
• 54:11
  and feature set.
• 54:12
  So there will always be kind of the
• 54:14
  this is okay for the general public,
• 54:16
  this is too dangerous.
• 54:19
  And AI is too dangerous, like I
• 54:21
  don't think anyone doubts
• 54:23
  that statement.
• 54:24
  You
• 54:26
  know, they're...
• 54:29
  AI is a little bit different
• 54:30
  because, you know, the GPS example
• 54:32
  needs a controlled satellite
• 54:34
  environment, which is very different
• 54:36
  in requirements. So the average user
• 54:38
  can't get access to that.
• 54:39
  I think AI theoretically, the
• 54:41
  average, user can get access to
• 54:43
  these kind of unrestricted models.
• 54:47
  And there will have to be guidelines
• 54:48
  around that, right?
• 54:49
  Like, it's the same reason you don't
• 54:51
  want your average user to have
• 54:52
  weapons-grade plutonium.
• 54:54
  Yeah. You so you're going
• 54:56
  to have to make rules that kind of
• 54:58
  prevents that because As these
• 54:59
  models are getting better...
• 55:04
  I'm not actually as worried about
• 55:06
  crypto, if I'm being honest.
• 55:07
  I think the crypto scare was a
• 55:08
  little bit over because we've always
• 55:09
  worked from that the smart contract
• 55:11
  is the most important point
• 55:12
  fundamentals, and we've secured that
• 55:14
  to the point where
• 55:17
  the scarier part for
• 55:19
  AI is actually the off-chain
• 55:20
  infrastructure part, you know,
• 55:21
  compromising nodes, all of those
• 55:22
  things. That's not to say that we're
• 55:24
  not going to see some node-based
• 55:25
  zero-day or those kinds of things.
• 55:26
  I do think they're coming, but
• 55:27
  that's not actually the bigger worry
• 55:28
  for me. I think the bigger worry is
• 55:31
  when it comes to traditional
• 55:32
  institutions, you know, because At
• 55:33
  some point, they get so large, they
• 55:35
  don't know what ports are open
• 55:36
  anymore, they don't what kind of
• 55:38
  new UDP package can cause some kind
• 55:40
  of shell system escape
• 55:42
  there and then give privilege access
• 55:44
  to some bot.
• 55:45
  And these bots, you
• 55:47
  know, they can just be hitting these
• 55:48
  things 24-7.
• 55:51
  And people underestimate the speed
• 55:53
  at which they can operate on
• 55:55
  the internet versus how quickly
• 55:56
  humans can respond or do something.
• 55:58
  You know, they can do things at the
• 56:00
  speed of light.
• 56:01
  Which us as humans, I don't think
• 56:03
  have really conceptualized or
• 56:04
  comprehended how dangerous that
• 56:06
  statement is.
• 56:07
  Because even if you have a response
• 56:09
  team now, they can't do anything.
• 56:11
  Because it will adapt to your new
• 56:12
  strategies before you can implement
• 56:14
  them.
• 56:15
  So I mean, look, there's definitely
• 56:16
  a real doomsday scenario
• 56:18
  when it comes to comes to
• 56:20
  these bots.
• 56:22
  Hopefully, you know, it's dooms day
• 56:24
  internet and not doomsday nuclear
• 56:26
  weapons.
• 56:27
  Because if the internet goes down
• 56:28
  and we forever, for always, all
• 56:30
  humanity loses access to it,
• 56:32
  I'm not so sure that's a bad thing.
• 56:34
  I might motivate that's all right.
• 56:36
  Like I go back to living in huts.
• 56:39
  I tend to think that's an okay
• 56:41
  scenario. If they wipe out
• 56:42
  electricity, also in a case
• 56:43
  scenario. I'm cool with caves and
• 56:45
  fires.
• 56:47
  If they get access to the weapons,
• 56:48
  different story.
• 56:49
  Yeah, yeah, for sure.
• 56:51
  Look, Andre, thank you so much for
• 56:52
  taking the time to chat to us today.
• 56:54
  You know, you are an OG in DeFi and
• 56:55
  it's really great to actually see
• 56:57
  you finally like launch something
• 56:59
  yourself and have full ownership of
• 57:00
  that and be very excited about what
• 57:02
  it is and, you know, be able to
• 57:03
  innovate on that.
• 57:05
  We'll be watching with keen
• 57:06
  interest. So I think if and when you
• 57:07
  have some big updates to tell us
• 57:09
  around Flying Tulip or
• 57:11
  if we want to have any, you know,
• 57:13
  in-depth discussions around DeFi or
• 57:15
  anything, we'll be knocking on your
• 57:16
  door once again.
• 57:17
  But thank you, so much, for joining
• 57:18
  us.
• 57:19
  Appreciate it, man, and I'm
• 57:20
  available. You can always ask me for
• 57:22
  my unwanted opinion.
• 57:26
  Thank you so much for joining us and
• 57:27
  thank you for watching the Starting
• 57:29
  Block today. We will be back
• 57:30
  tomorrow with another show.
• 57:33
  So yeah, set a reminder on your
• 57:34
  clocks, we broadcast 8.30am Eastern
• 57:36
  Time, 2.30pm
• 57:38
  Central European Time if you're this
• 57:39
  side of the world.
• 57:40
  Come and join us. I promise you'll
• 57:41
  have a good time.
• 57:42
  Until then, we'll catch you
• 57:43
  tomorrow.