1
00:00:00,040 --> 00:00:01,880
So I think the market is 
shifting. 

2
00:00:02,400 --> 00:00:05,600
We're shifting from knowledge 
work being radically transformed

3
00:00:05,800 --> 00:00:09,400
with tools like Claude Code 
being reimagined for knowledge 

4
00:00:09,400 --> 00:00:13,680
workers with Claude Co work. 
And we're now moving into an era

5
00:00:13,680 --> 00:00:17,080
where skill sets such as 
managing AI tools that are doing

6
00:00:17,080 --> 00:00:20,480
work for you is more important 
than just using the tool itself.

7
00:00:20,880 --> 00:00:24,560
And this big change brings with 
it brand new skill sets. 

8
00:00:24,640 --> 00:00:27,000
And I think these skill sets are
going to be incredibly important

9
00:00:27,000 --> 00:00:30,040
for you at work, your family, 
your colleagues. 

10
00:00:30,240 --> 00:00:35,080
Because right now, January 
hiring and firing data in the US

11
00:00:35,080 --> 00:00:38,000
looks quite painful. 
It was the worst since 2009, 

12
00:00:38,560 --> 00:00:41,520
with companies openly saying 
that they're removing and firing

13
00:00:41,520 --> 00:00:44,800
people as a result of AI. 
In today's episode, I'm going to

14
00:00:44,800 --> 00:00:47,680
talk about the exact skill sets 
that I think are critical for 

15
00:00:47,680 --> 00:00:51,440
you to learn. 
Judgement, taste, delegation. 

16
00:00:51,440 --> 00:00:54,000
I'm going to talk about 
frameworks for deciding when to 

17
00:00:54,000 --> 00:00:59,120
use AI or when to hand the work 
off to AI, and ultimately what 

18
00:00:59,120 --> 00:01:01,800
you can practically do about 
this massive change. 

19
00:01:01,960 --> 00:01:03,360
This is in the loop with Jack 
Horton. 

20
00:01:03,840 --> 00:01:17,780
I hope you enjoy the show. 
So let's start with what skills 

21
00:01:17,780 --> 00:01:19,140
I think are now really 
important. 

22
00:01:19,300 --> 00:01:22,700
Now, Ethan Molik did a fantastic
piece on all of this. 

23
00:01:23,260 --> 00:01:26,360
And for those who don't know 
him, Ethan Molik is AAI, a 

24
00:01:26,400 --> 00:01:29,320
researcher and professor that 
puts out fantastic pieces 

25
00:01:29,320 --> 00:01:31,480
online. 
I'd highly recommend to go check

26
00:01:31,480 --> 00:01:34,440
him out. 
But he did a big deep dive into 

27
00:01:34,520 --> 00:01:37,200
important skill sets. 
His answer was fantastic. 

28
00:01:37,840 --> 00:01:40,080
It's very, very simple. 
It's management. 

29
00:01:40,280 --> 00:01:42,760
Not management in the sense of 
sitting in meetings and, you 

30
00:01:42,760 --> 00:01:45,600
know, reviewing slide decks. 
Management in the way of 

31
00:01:45,880 --> 00:01:52,280
describing exactly what you need
to give fantastic and effective 

32
00:01:52,280 --> 00:01:56,200
and clear feedback and design 
new ways of evaluating work you 

33
00:01:56,200 --> 00:01:59,560
get back from in this context, 
agents, if you're good at those 

34
00:01:59,560 --> 00:02:01,800
things, you're going to be able 
to work with agents very well. 

35
00:02:01,920 --> 00:02:04,920
And what I absolutely loved 
about this piece is that he adds

36
00:02:04,920 --> 00:02:08,720
a very important nuance 
Management has always assumed 

37
00:02:08,720 --> 00:02:12,160
scarcity. 
You delegate because you can't 

38
00:02:12,160 --> 00:02:15,000
do everything yourself and 
because talent is very limited 

39
00:02:15,000 --> 00:02:18,160
and very expensive. 
AI completely flips this talent 

40
00:02:18,160 --> 00:02:21,520
is abundant and it's cheap. 
What's scarce and difficult all 

41
00:02:21,520 --> 00:02:24,840
of a sudden is knowing what to 
ask for and how to ask it. 

42
00:02:25,040 --> 00:02:29,880
And for example, senior product 
managers at Google mapped out 

43
00:02:29,880 --> 00:02:33,720
different skill sets that they 
believe are now critical to a 

44
00:02:33,720 --> 00:02:36,120
modern worker. 
The ability to shape a problem, 

45
00:02:36,120 --> 00:02:41,280
so turning vague objectives into
precise, testable subtasks. 

46
00:02:41,360 --> 00:02:45,280
The ability to curate context, 
so deciding what information is 

47
00:02:45,280 --> 00:02:49,640
actually important for this 
subtask and only giving it that 

48
00:02:49,640 --> 00:02:51,640
context so you can get the 
perfect output. 

49
00:02:51,760 --> 00:02:56,480
Creating fantastic documentation
and systems so you can get 90% 

50
00:02:56,680 --> 00:02:58,840
of the way there on a single 
attempt. 

51
00:02:59,000 --> 00:03:01,440
Another really important skill 
they labeled was another 

52
00:03:01,440 --> 00:03:04,400
important skill was actually 
ability to recognise quality. 

53
00:03:04,560 --> 00:03:08,080
The ability to orchestrate many,
many agents, managing sequential

54
00:03:08,080 --> 00:03:10,760
versus, say, parallel tasks. 
We'll get on to that. 

55
00:03:11,920 --> 00:03:15,640
The ability to understand guard 
rails and how to ensure agents 

56
00:03:15,640 --> 00:03:18,000
are performing well. 
Knowing when to intervene with 

57
00:03:18,000 --> 00:03:20,880
an agent or when to start again,
and judgement about the tool 

58
00:03:20,880 --> 00:03:23,280
sets you're using. 
Understanding when to use an 

59
00:03:23,280 --> 00:03:27,240
agent, when to use a simple 
model and when just to do it 

60
00:03:27,240 --> 00:03:29,800
yourself. 
These skills aren't new really. 

61
00:03:30,440 --> 00:03:33,080
Yes, they apply it to AI and 
there's a learning curve there, 

62
00:03:33,080 --> 00:03:36,160
but this is one O 1 management 
skills. 

63
00:03:36,200 --> 00:03:39,320
I think the CEO of Shopify, it's
a fantastic case study here. 

64
00:03:39,480 --> 00:03:42,800
He's GitHub commits, so the 
actual code that he's committed 

65
00:03:42,800 --> 00:03:48,040
to, Shopify's code base has gone
up from 94 commits in 2024 to 

66
00:03:48,040 --> 00:03:51,960
over 900 in 2026 alone. 
You can tell he's clearly not 

67
00:03:51,960 --> 00:03:53,840
writing all that code himself. 
He's the CEO. 

68
00:03:53,840 --> 00:03:57,040
He's probably a busy, busy 
person, but he's orchestrating 

69
00:03:57,040 --> 00:04:00,240
and managing agents to write 
lots of code and get it into the

70
00:04:00,240 --> 00:04:03,280
code base. 
It's turning a CEO of a massive 

71
00:04:03,280 --> 00:04:06,200
company into a builder again, 
because again, talent is now 

72
00:04:06,200 --> 00:04:08,280
cheaper and more readily 
available. 

73
00:04:08,560 --> 00:04:11,120
What matters is your ability to 
instruct it. 

74
00:04:11,720 --> 00:04:14,520
What matters is the skill sets 
that I'm going to go into that 

75
00:04:14,920 --> 00:04:18,320
ensures what you're producing is
actually good quality work. 

76
00:04:18,480 --> 00:04:20,880
We saw this also with Microsoft 
CEO Satya Nadella. 

77
00:04:21,279 --> 00:04:24,560
They used to have, say, a 
designer, a front end engineer, 

78
00:04:24,600 --> 00:04:27,280
a back end, a senior architect, 
a product manager. 

79
00:04:27,480 --> 00:04:29,840
And what they're doing is 
rolling all of that skill set 

80
00:04:29,840 --> 00:04:33,080
into one person and calling them
a full stack builder. 

81
00:04:33,400 --> 00:04:37,640
Someone that can have an idea, 
shape it, understand the problem

82
00:04:37,640 --> 00:04:40,680
they're solving, build a lot of 
it, and get it into a place 

83
00:04:40,680 --> 00:04:44,920
where others can review without 
relying on 18 meetings and 

84
00:04:44,920 --> 00:05:00,280
documents. 
So there's a lot of skill sets 

85
00:05:00,280 --> 00:05:02,440
that are emerging, are becoming 
increasingly important. 

86
00:05:02,880 --> 00:05:05,720
Another one is taste. 
Taste is about selection, 

87
00:05:05,920 --> 00:05:10,160
recognizing quality amongst many
different options or directions.

88
00:05:10,320 --> 00:05:14,440
If AI generates US10 versions of
something, which one do you go 

89
00:05:14,440 --> 00:05:15,920
for? 
Which one is actually good? 

90
00:05:16,040 --> 00:05:18,880
Judgement on the other hand, is 
about weighting trade-offs. 

91
00:05:19,000 --> 00:05:21,600
It's about evaluation, 
especially when you don't have 

92
00:05:21,600 --> 00:05:25,040
complete information, you don't 
know necessarily the end 

93
00:05:25,040 --> 00:05:27,280
direction or an outcome that 
you're going for. 

94
00:05:27,480 --> 00:05:30,280
So yeah, suddenly decision 
making integrates all these 

95
00:05:30,280 --> 00:05:33,880
different skill sets of taste of
judgement, of evaluation. 

96
00:05:34,160 --> 00:05:37,720
And then management skills is 
the coordination of all these 

97
00:05:37,720 --> 00:05:41,920
different decisions across many 
different agents and tools to 

98
00:05:41,920 --> 00:05:44,360
get more work done than ever 
before. 

99
00:05:44,480 --> 00:05:47,440
Another really important skill 
set here is curiosity, which I 

100
00:05:47,440 --> 00:05:52,200
think is arguably at the moment 
my key most desirable skill set.

101
00:05:52,360 --> 00:05:57,240
You could have extraordinary 
taste, but without any agency 

102
00:05:57,240 --> 00:05:59,920
and curiosity, you won't create 
anything. 

103
00:06:00,080 --> 00:06:02,040
You need to be constantly 
asking, is there another 

104
00:06:02,040 --> 00:06:04,680
direction to take? 
Can I prompt it to explore this 

105
00:06:04,680 --> 00:06:06,200
direction? 
And if it doesn't get me 

106
00:06:06,200 --> 00:06:08,240
anywhere, that's fine. 
But then I've been curious 

107
00:06:08,240 --> 00:06:11,880
enough to explore it. 
And again, that judgement and 

108
00:06:11,880 --> 00:06:15,120
taste to know when it's good 
enough and we're ready to ship. 

109
00:06:15,280 --> 00:06:18,000
So someone could have 
extraordinary taste, but no, 

110
00:06:18,000 --> 00:06:20,480
let's say agency. 
So the ability to make a 

111
00:06:20,480 --> 00:06:22,920
decision, therefore you can't 
create anything. 

112
00:06:24,080 --> 00:06:27,000
You could have lots of agency, 
but awful judgement. 

113
00:06:27,640 --> 00:06:30,680
So you could move really fast, 
but in the wrong direction. 

114
00:06:30,840 --> 00:06:34,120
You could have all the curiosity
in the world, but without the 

115
00:06:34,120 --> 00:06:38,320
ability to follow through and 
take action, nothing happens. 

116
00:06:38,600 --> 00:06:41,120
So all these skills combined 
suddenly become extremely 

117
00:06:41,120 --> 00:06:42,720
important to develop in 
yourself. 

118
00:06:42,880 --> 00:06:45,240
And we've seen these types of 
comments from big leaders all 

119
00:06:45,240 --> 00:06:47,520
over the world right now. 
So the Co founder of Y 

120
00:06:47,520 --> 00:06:50,240
Combinator, probably the most 
famous starter by accelerator in

121
00:06:50,240 --> 00:06:53,160
the entire world, believes 
judgement is now the skill set 

122
00:06:53,160 --> 00:06:57,120
that matters. 
In his words, when anyone can 

123
00:06:57,120 --> 00:07:00,280
make anything, it's the ability 
to decide what to make the 

124
00:07:00,280 --> 00:07:03,000
counts. 
Greg Brockman, the president of 

125
00:07:03,000 --> 00:07:07,360
Open AI, said that taste is now 
his most important skill. 

126
00:07:07,880 --> 00:07:10,440
And the CTO of Cloudflare, so 
the thing that basically makes 

127
00:07:10,440 --> 00:07:14,320
the Internet work, said that 
taste in 2026 is now the 

128
00:07:14,320 --> 00:07:17,360
engineering differentiator. 
So yeah, I've discussed what 

129
00:07:17,360 --> 00:07:19,720
skills are really important now 
and why they're really 

130
00:07:19,720 --> 00:07:21,800
important. 
Because ultimately, as I said, 

131
00:07:22,480 --> 00:07:26,040
you can be a full stop builder 
in whatever career trajectory, 

132
00:07:26,040 --> 00:07:28,520
whatever career path you're 
currently doing, you can be a 

133
00:07:28,520 --> 00:07:31,520
full stop builder, someone who 
can produce more than anyone 

134
00:07:31,520 --> 00:07:34,120
else in the world. 
But it requires you to have 

135
00:07:34,120 --> 00:07:37,720
those important skill sets, to 
be trusted in the organization, 

136
00:07:38,360 --> 00:07:39,800
to actually go and do those 
things. 

137
00:07:52,120 --> 00:07:54,920
So then I think now let's move 
on to, I guess the equation for 

138
00:07:54,920 --> 00:07:57,440
when to delegate to AI. 
Think this is a really important

139
00:07:57,440 --> 00:08:00,040
one to get started with and a 
really good framework of 

140
00:08:00,040 --> 00:08:01,840
thinking about these big 
questions. 

141
00:08:01,920 --> 00:08:04,120
Now Ethan Mollick believes it 
comes down to just three 

142
00:08:04,120 --> 00:08:08,560
variables. 1st, the human 
baseline of time. 

143
00:08:08,800 --> 00:08:11,720
So how long would a task take 
for you to complete? 

144
00:08:11,920 --> 00:08:14,040
Second, the probability of 
success. 

145
00:08:14,160 --> 00:08:18,840
How how likely is AI to give you
a good answer and output? 

146
00:08:19,080 --> 00:08:22,400
And 3rd is AI process time. 
How long would it take you to 

147
00:08:22,400 --> 00:08:25,800
write the prompt, get the answer
and evaluate the output? 

148
00:08:26,040 --> 00:08:28,440
Now, on the face of it, it seems
quite simple, but that requires 

149
00:08:28,440 --> 00:08:31,240
you to have in depth 
understanding of the tools, how 

150
00:08:31,240 --> 00:08:34,840
to manage agents, what it's good
at and bad at, to have stress 

151
00:08:34,840 --> 00:08:36,960
tested all those different 
avenues, and therefore to really

152
00:08:36,960 --> 00:08:40,360
touch yourself into it. 
And the reason I'm emphasizing 

153
00:08:40,360 --> 00:08:42,919
this point so much about these 
new skill sets, the way of 

154
00:08:42,919 --> 00:08:47,200
deciding when to delegate or not
delegate something to AI is 

155
00:08:47,400 --> 00:08:50,400
because of the huge amount of 
layoffs that are starting to 

156
00:08:50,400 --> 00:08:52,520
happen across the US. 
I'm sure the data is going to 

157
00:08:52,520 --> 00:08:55,240
back it up in the UK too. 
You know, US employers announced

158
00:08:55,640 --> 00:08:59,960
over 108,000 layoffs in January,
which is 120% increase year on 

159
00:08:59,960 --> 00:09:01,960
year. 
Hiring intentions fell to their 

160
00:09:01,960 --> 00:09:05,720
lowest level since records began
and AI was cited in over 7 1/2 

161
00:09:05,720 --> 00:09:08,320
thousand job cuts. 
And although that only 

162
00:09:08,320 --> 00:09:11,680
represents 7% of those total 
jobs, I truly believe that it's 

163
00:09:11,680 --> 00:09:14,560
understating the impact AI is 
starting to have on the job 

164
00:09:14,560 --> 00:09:28,240
market. 
So what should you take away 

165
00:09:28,240 --> 00:09:32,160
from this episode? 
Well, ultimately, we're seeing a

166
00:09:32,160 --> 00:09:34,440
shift from execution to 
direction. 

167
00:09:34,440 --> 00:09:37,880
And we're seeing this especially
in the world of programming and 

168
00:09:37,880 --> 00:09:40,840
developers. 
But as I've said a number of 

169
00:09:40,840 --> 00:09:43,320
times on this podcast, it's 
coming for everybody. 

170
00:09:44,520 --> 00:09:48,120
The skills that matter now, our 
taste, our judgement, agency, 

171
00:09:48,120 --> 00:09:51,880
curiosity, and the ability to 
coordinate work across many 

172
00:09:51,880 --> 00:09:53,840
different agents. 
These are the sorts of tasks 

173
00:09:53,840 --> 00:09:57,440
that often people didn't develop
in themselves because they were 

174
00:09:57,440 --> 00:09:59,560
so busy on executing things 
themselves. 

175
00:09:59,560 --> 00:10:01,840
Whereas now everybody is 
becoming a manager. 

176
00:10:02,000 --> 00:10:07,040
And the economic reality of this
change is uneven and in places 

177
00:10:07,040 --> 00:10:10,040
it's going to be really unfair. 
I think layoffs in this next 

178
00:10:10,040 --> 00:10:13,440
period are probably going to be 
structural, not temporary, but 

179
00:10:13,440 --> 00:10:16,960
there are many skills that 
everybody must learn and 

180
00:10:16,960 --> 00:10:20,600
demonstrate if they're going to 
navigate these quite difficult 

181
00:10:20,600 --> 00:10:23,120
times. 
Nobody can fully predict what 

182
00:10:23,120 --> 00:10:25,760
work's going to look like, what 
everybody's a manager, but I 

183
00:10:25,760 --> 00:10:29,800
think those that will thrive are
those that can describe exactly 

184
00:10:29,800 --> 00:10:32,640
what they need and make the 
perfect judgement of when it is 

185
00:10:32,640 --> 00:10:34,880
good enough to ship. 
Anyway, that's a short episode 

186
00:10:34,880 --> 00:10:37,920
this week on important skill 
sets that we're seeing emerge. 

187
00:10:38,040 --> 00:10:41,080
I hope you enjoyed the episode, 
I hope you found it useful and I

188
00:10:41,080 --> 00:10:42,320
will see you next week.