Scaling Human Expertise in Crypto and AI

1
00:00:00,040 --> 00:00:04,960
It's one of the things that
drove me to amplify myself.

2
00:00:05,040 --> 00:00:10,320
Once I amplified myself, my work
was better, it was faster, and

3
00:00:10,320 --> 00:00:15,040
then the next Phase I knew that
we needed is not just amplifying

4
00:00:15,040 --> 00:00:18,160
yourself, but scaling yourself.
This is Money Never Sleeps.

5
00:00:18,160 --> 00:00:21,040
Sharp riffs, big ideas and real
insights from smart people.

6
00:00:21,440 --> 00:00:22,840
I'm Pete Townsend.
Let's go money.

7
00:00:25,760 --> 00:00:30,080
Never sleeps, pal.
So Joanie Perovich, founder of

8
00:00:30,080 --> 00:00:33,560
Crystal AOS and The Walking,
living version of The

9
00:00:33,560 --> 00:00:37,240
Innovator's Dilemma.
Disrupt before you're disrupted.

10
00:00:37,640 --> 00:00:40,960
Welcome to the show.
Thanks for having me Pete, good

11
00:00:40,960 --> 00:00:42,960
to be here.
So listen, you pinged me last

12
00:00:42,960 --> 00:00:46,680
week and you said that you'd
like the predictions that

13
00:00:46,680 --> 00:00:50,840
Alejandro and I did where the
headline was.

14
00:00:51,080 --> 00:00:57,000
Crypto is getting boring, OK,
and crypto doesn't get boring

15
00:00:57,000 --> 00:01:00,120
because it failed.
It's a sign of maturity.

16
00:01:00,920 --> 00:01:05,680
So when people hear crypto is
boring, why do you think they

17
00:01:05,680 --> 00:01:08,240
misunderstand what that actually
means?

18
00:01:10,520 --> 00:01:15,480
Well, I think crypto has had so
many highs and lows over the

19
00:01:15,480 --> 00:01:19,240
last decade that it's probably
about time that it got boring.

20
00:01:19,320 --> 00:01:23,120
Totally thinking back to your
life before becoming founder of

21
00:01:23,120 --> 00:01:28,080
Crystal AOS is that you're a
lawyer and still are to this

22
00:01:28,080 --> 00:01:34,280
day, obviously, but in terms of
that expertise that's required,

23
00:01:34,280 --> 00:01:37,480
that doesn't scale.
Were you ever that bottleneck

24
00:01:37,560 --> 00:01:40,360
yourself as a lawyer?
All the time.

25
00:01:40,560 --> 00:01:46,080
And it's one of the things that
drove me to amplify myself.

26
00:01:46,120 --> 00:01:49,640
And actually, when I first
started playing around with AI,

27
00:01:50,280 --> 00:01:56,680
my first AI application to make
to amplify myself was what's now

28
00:01:56,680 --> 00:02:00,280
called Crystal Chat.
And it's a global repository of

29
00:02:00,280 --> 00:02:03,240
all of the crypto laws and
regulations around the world and

30
00:02:03,240 --> 00:02:07,360
policy announcements.
And I could chat with that daily

31
00:02:07,360 --> 00:02:12,760
to check my work to make sure
that I had identified risks and

32
00:02:12,760 --> 00:02:15,960
best practices around the world.
Singapore's doing this on stable

33
00:02:15,960 --> 00:02:17,640
coins.
Hong Kong's doing that.

34
00:02:17,880 --> 00:02:21,720
How does this feed into what the
US is negotiating at the moment?

35
00:02:22,200 --> 00:02:29,240
And even though the profiles of
projects and people that have

36
00:02:29,240 --> 00:02:32,480
come in and out of crypto over
the last decade have changed a

37
00:02:32,480 --> 00:02:35,840
lot, actually being a crypto
lawyer and advising on best

38
00:02:35,840 --> 00:02:40,720
practice standards hasn't.
And I couldn't do that fast

39
00:02:40,720 --> 00:02:44,400
enough without tools like
Crystal Chat.

40
00:02:44,760 --> 00:02:50,760
So once I amplified myself with
Crystal Chat, my work was

41
00:02:50,760 --> 00:02:54,720
better, it was faster, and I was
billing my clients less for

42
00:02:55,040 --> 00:02:58,200
better work.
And then the next Phase I knew

43
00:02:58,200 --> 00:03:04,160
that we needed to move into for
hyper performing advisors is not

44
00:03:04,160 --> 00:03:06,800
just amplifying yourself, but
scaling yourself.

45
00:03:06,920 --> 00:03:08,920
Totally.
I just, you know, when you

46
00:03:08,920 --> 00:03:11,920
mention all the different crypto
regulations that are coming up

47
00:03:11,920 --> 00:03:14,600
in the world and that trying to
cross reference between all of

48
00:03:14,600 --> 00:03:18,680
them and advise your clients on
all of that, it seems like AI is

49
00:03:18,680 --> 00:03:21,720
a survival tool.
It has to be.

50
00:03:22,040 --> 00:03:27,760
It's a non negotiable now.
And so then the next step is how

51
00:03:27,760 --> 00:03:32,080
do we use AI with guardrails?
Because it's probabilistic, it's

52
00:03:32,080 --> 00:03:36,000
not always going to produce the
same outcome as it did the last

53
00:03:36,000 --> 00:03:41,240
two times or the last 200 times.
So how do we step up our systems

54
00:03:41,280 --> 00:03:44,040
to incorporate AI responsibly
and safely?

55
00:03:44,040 --> 00:03:46,840
And that's the problem I've been
working on for the last few

56
00:03:46,840 --> 00:03:49,160
years.
Thinking about things like what

57
00:03:49,160 --> 00:03:53,360
changes when compliance stops
being something that people do

58
00:03:53,360 --> 00:03:56,280
as an add on to their business
and become something that just

59
00:03:56,280 --> 00:03:58,280
runs in the background.
Yeah.

60
00:03:58,280 --> 00:04:02,720
And I actually have bought a
Tesla in the last year.

61
00:04:03,160 --> 00:04:09,000
And the best analogy I can give
is in previous cars, I would get

62
00:04:09,000 --> 00:04:13,200
a sticker to have the car
serviced at 10,000 kilometers

63
00:04:13,200 --> 00:04:16,680
and then 50,000 kilometers and
just these sort of seemingly

64
00:04:16,680 --> 00:04:19,920
arbitrary numbers.
And I didn't know what parts

65
00:04:19,920 --> 00:04:23,080
needed to be changed at each
service or really how much it

66
00:04:23,080 --> 00:04:26,160
was going to cost.
But now with the Tesla, there's

67
00:04:26,160 --> 00:04:29,240
no such thing as a 10,000
kilometer service.

68
00:04:29,360 --> 00:04:36,440
I get notifications to say your
tread in your left rear tire is

69
00:04:36,440 --> 00:04:38,560
low, it's time to get it
changed.

70
00:04:39,160 --> 00:04:41,720
And so this is the perfect
analogy.

71
00:04:41,720 --> 00:04:47,040
I think as our compliance
systems get smarter, we're not

72
00:04:47,320 --> 00:04:52,760
just arbitrarily doing an annual
business risk review or these

73
00:04:52,760 --> 00:04:56,600
check the box type things that
are ad hoc and not in the

74
00:04:56,600 --> 00:04:59,280
moment.
You're actually getting

75
00:04:59,280 --> 00:05:02,000
notifications that are
meaningful that you can respond

76
00:05:02,000 --> 00:05:04,800
to straight away.
And your day-to-day becomes

77
00:05:04,920 --> 00:05:09,040
focused on the quality issues
that are really going to lead to

78
00:05:09,040 --> 00:05:11,640
the business growth and more
choice for the users.

79
00:05:11,920 --> 00:05:14,480
I've got to upgrade my car.
I'm about to spend another

80
00:05:14,600 --> 00:05:19,120
probably 2000 on a catalytic
converter for my 12 year old

81
00:05:19,120 --> 00:05:21,840
car.
But yeah, the warning lights on

82
00:05:21,840 --> 00:05:24,600
that weren't very helpful.
It was just blink, blink, blink.

83
00:05:24,760 --> 00:05:28,960
And that's, you know, kind of
looking at compliance today in

84
00:05:29,080 --> 00:05:32,040
in crypto exchanges, they're
lucky if they get the blink,

85
00:05:32,040 --> 00:05:33,440
blink, blink.
You know.

86
00:05:34,200 --> 00:05:37,840
Never mind a message that comes
to you that says you were

87
00:05:37,840 --> 00:05:39,400
potentially in breach of this
regulation.

88
00:05:39,520 --> 00:05:42,840
Looking at your kind of the
wedge that you've built into the

89
00:05:42,840 --> 00:05:49,080
market, Joni, and that you've
got those firms on board that

90
00:05:49,200 --> 00:05:52,320
know that they have to embrace
regulation and they have to move

91
00:05:52,320 --> 00:05:54,800
forward with it and actually can
be a benefit to the business.

92
00:05:54,800 --> 00:05:58,880
Like stablecoins, for example,
that hit 250 billion as an asset

93
00:05:58,880 --> 00:06:02,600
class before regulation.
And look what now regulation

94
00:06:02,600 --> 00:06:05,320
will do to that to open up the
universe even more.

95
00:06:06,520 --> 00:06:09,160
Have you come across yet any
firms that have said we'll

96
00:06:09,160 --> 00:06:11,840
listen, we're staying away from
regulation and do you think they

97
00:06:11,840 --> 00:06:15,760
might hit a wall later?
I think the sums are more mature

98
00:06:15,760 --> 00:06:19,440
now and they're not so much
saying let's stay away from

99
00:06:19,440 --> 00:06:25,560
regulation, but those that are
choosing to incorporate or to

100
00:06:25,560 --> 00:06:31,120
remain in I'd say regulatory and
tax friendly jurisdictions,

101
00:06:31,560 --> 00:06:37,680
those that have very pioneering
innovations that really don't

102
00:06:37,680 --> 00:06:41,160
fit then a square peg, round
hole problem.

103
00:06:41,680 --> 00:06:46,120
Those are the projects that they
have a true heart, a true

104
00:06:46,120 --> 00:06:49,960
innovator's heart to make
change, but they need some

105
00:06:49,960 --> 00:06:53,440
freedom and flexibility in those
regulatory and tax friendly

106
00:06:53,440 --> 00:06:58,000
environments before they can
mature into jurisdictions that

107
00:06:58,000 --> 00:07:01,560
are working on changing their
laws and evolving them for these

108
00:07:01,560 --> 00:07:04,440
idiosyncrasies.
Just look at the US market

109
00:07:04,440 --> 00:07:09,480
structure, bill negotiations and
the carve out with the European

110
00:07:09,480 --> 00:07:11,800
Union for fully decentralized
projects.

111
00:07:12,240 --> 00:07:18,160
There is a mind towards what
that change is, but there's a

112
00:07:18,160 --> 00:07:22,520
journey for the regulators and
the governments to go on on how

113
00:07:22,680 --> 00:07:26,040
to sensibly bring that into
regulatory frameworks.

114
00:07:26,160 --> 00:07:29,920
Thinking about things like
agentic compliance specifically

115
00:07:30,000 --> 00:07:37,000
and agentic legal advice.
Talk to me about how this will

116
00:07:37,000 --> 00:07:41,200
allow you to multiply expertise.
A lot of it revolves around

117
00:07:41,200 --> 00:07:44,160
there are a set of laws that
need to be complied with or a

118
00:07:44,160 --> 00:07:49,280
set of internal policies that
all of the internal staff and

119
00:07:49,440 --> 00:07:53,520
external stakeholders, some to
some degree need to know about

120
00:07:53,520 --> 00:07:56,840
and follow.
And there's a need for evidence

121
00:07:56,840 --> 00:08:00,720
that those laws or internal
policies have been followed.

122
00:08:01,240 --> 00:08:05,000
We need to have guardrails that
say, well, even though the

123
00:08:05,000 --> 00:08:11,000
process is known, the AI output
could still go off track.

124
00:08:11,240 --> 00:08:13,880
And so we need the human in the
loop accountability.

125
00:08:14,400 --> 00:08:18,400
But more than that, we need
another set of AI agents

126
00:08:19,280 --> 00:08:23,880
reviewing and evaluating what
the primary AI or the eugenic

127
00:08:23,880 --> 00:08:27,760
processes are producing.
We can see what the AI is doing.

128
00:08:27,760 --> 00:08:30,720
We could intervene anytime, but
we've got these human

129
00:08:30,720 --> 00:08:35,520
checkpoints and these AI
detectives making sure that it's

130
00:08:35,520 --> 00:08:37,960
all working OK.
Within their day-to-day

131
00:08:37,960 --> 00:08:42,320
business.
Why is that gap so hard to

132
00:08:42,320 --> 00:08:45,680
cross?
It's such a different way of

133
00:08:45,680 --> 00:08:49,400
working.
It's it's like the first time

134
00:08:49,520 --> 00:08:55,320
that a human elevates in their
role from a doer to a manager.

135
00:08:56,160 --> 00:09:00,360
You have to let go and you have
to learn the new skill of

136
00:09:00,360 --> 00:09:05,520
instructing the junior team
members how to do what, yes, it

137
00:09:05,520 --> 00:09:09,000
could be easier if you do it
yourself, but if you're the one

138
00:09:09,000 --> 00:09:12,240
still doing all of that work,
you're still the player and

139
00:09:12,240 --> 00:09:14,960
you're not the coach.
And so you never have time to

140
00:09:15,000 --> 00:09:20,240
scale yourself and the business.
And that mindset change that

141
00:09:20,240 --> 00:09:25,280
hurts people to go through it.
And and so on a larger scale,

142
00:09:25,280 --> 00:09:30,560
when an enterprise is saying we
want to lift everybody up to

143
00:09:30,560 --> 00:09:34,720
manage out or to coach so that
you haven't a workforce of

144
00:09:35,080 --> 00:09:40,440
agentic workflows and AI agents
working for you, that that is

145
00:09:40,440 --> 00:09:44,000
going to hurt.
And, and so first of all, it's

146
00:09:44,000 --> 00:09:48,320
identifying that we are moving
this way whether we like it or

147
00:09:48,320 --> 00:09:51,280
not.
And that's actually part of why

148
00:09:51,280 --> 00:09:53,920
just the innovator's dilemma,
why I started it.

149
00:09:53,920 --> 00:09:58,200
I could see that the traditional
delivery of legal and compliance

150
00:09:58,200 --> 00:10:03,560
services was going to change
with AI, and if somebody was

151
00:10:03,560 --> 00:10:08,920
going to change that, then I
wanted to be at the forefront

152
00:10:08,920 --> 00:10:12,120
and doing it sensibly and
responsibly from all of the

153
00:10:12,120 --> 00:10:15,240
experience of actually being an
advisor.

154
00:10:15,840 --> 00:10:18,720
And that's what I've done.
That's what I'm doing.

155
00:10:19,360 --> 00:10:21,200
You absolutely have.
You absolutely have.

156
00:10:21,680 --> 00:10:24,600
If someone's still waiting for
perfect clarity before engaging

157
00:10:24,800 --> 00:10:28,400
with crypto or AI, what's the
risk of waiting too long?

158
00:10:29,360 --> 00:10:33,600
Well, I say in the same way that
perfect is the enemy of goods,

159
00:10:34,160 --> 00:10:37,000
waiting is the enemy of ongoing
relevance.

160
00:10:37,640 --> 00:10:42,080
If if you're waiting, you're
actually not experimenting and

161
00:10:42,080 --> 00:10:47,520
learning, and you could possibly
be missing out on the skills

162
00:10:47,520 --> 00:10:52,200
that are going to carry you
through the next shift to the AI

163
00:10:52,200 --> 00:10:54,680
era.
So start now.

164
00:10:54,800 --> 00:10:58,400
Stop waiting.
So true, so true.

165
00:10:58,480 --> 00:11:01,520
And Joni, for people who want to
keep following your thinking,

166
00:11:01,640 --> 00:11:03,520
what's the best way to get in
touch with you?

167
00:11:05,000 --> 00:11:09,480
I'm mostly on LinkedIn at
Journey Perovich and they can

168
00:11:09,480 --> 00:11:15,160
follow the Crystal AOS page.
That's where I share a lot of AI

169
00:11:15,240 --> 00:11:17,840
developments, how it's being
used in legal risk and

170
00:11:17,840 --> 00:11:22,160
compliance, as well as global
crypto policy developments.

171
00:11:22,720 --> 00:11:27,840
But otherwise, on Telegram at
Journey Perovich, I'm responsive

172
00:11:27,840 --> 00:11:30,480
there.
Fantastic, yes, and I know that

173
00:11:30,480 --> 00:11:33,960
for a fact you are so.
It's thank you.

174
00:11:33,960 --> 00:11:36,920
It's the crypto expectation.
Respond immediately.

175
00:11:37,840 --> 00:11:39,440
Absolutely.
Thank you, Joni.

176
00:11:39,680 --> 00:11:40,120
Awesome.
Check.

177
00:11:40,120 --> 00:11:40,600
Thanks.
Thanks.

178
00:11:41,040 --> 00:11:45,480
No, great to be here.
He never sleeps, pal.