1
00:00:05,820 --> 00:00:07,319
Hello everyone, and welcome to

2
00:00:07,320 --> 00:00:08,759
another episode of The Operations

3
00:00:08,760 --> 00:00:10,619
Room, a podcast for CEOs.

4
00:00:10,620 --> 00:00:12,449
I am Brandon Metzinger and I am

5
00:00:12,450 --> 00:00:14,069
joined by my lovely co-host Bethany

6
00:00:14,070 --> 00:00:16,319
Air is having some nice porridge.

7
00:00:16,320 --> 00:00:18,179
Breakfast is what I want.

8
00:00:18,180 --> 00:00:19,199
I'm looking at.

9
00:00:19,200 --> 00:00:20,969
If only it were porridge.

10
00:00:20,970 --> 00:00:22,049
Oh, it's not porridge.

11
00:00:22,050 --> 00:00:23,129
Looks like porridge.

12
00:00:23,130 --> 00:00:25,139
No, this is the chia seed concoction

13
00:00:25,140 --> 00:00:27,119
that, thanks to Zoe, I eat every

14
00:00:27,120 --> 00:00:28,979
day. From chia seeds,

15
00:00:28,980 --> 00:00:32,069
other seeds, kefir,

16
00:00:32,070 --> 00:00:33,809
milk, lots of nuts.

17
00:00:33,810 --> 00:00:35,729
And therefore every

18
00:00:35,730 --> 00:00:36,359
day.

19
00:00:36,360 --> 00:00:38,309
Right. So every trendy ingredient

20
00:00:38,310 --> 00:00:39,719
that I can imagine is on the list.

21
00:00:39,720 --> 00:00:40,679
Pretty much.

22
00:00:40,680 --> 00:00:42,089
So didn't you tell me before that

23
00:00:42,090 --> 00:00:44,099
the taste of your dish

24
00:00:44,100 --> 00:00:46,079
there, the cheesy delight, is not

25
00:00:46,080 --> 00:00:47,879
great or it's rather plain.

26
00:00:47,880 --> 00:00:49,319
Yeah. It's plain.

27
00:00:49,320 --> 00:00:50,669
I have some strawberries in there

28
00:00:50,670 --> 00:00:52,709
that helps the nuts help

29
00:00:52,710 --> 00:00:54,209
it, but it's a bit of a slog.

30
00:00:54,210 --> 00:00:56,849
And also because it's so

31
00:00:56,850 --> 00:00:58,889
chewy, it requires

32
00:00:58,890 --> 00:01:00,749
so much chewing

33
00:01:00,750 --> 00:01:02,459
that it takes me absolutely forever

34
00:01:02,460 --> 00:01:03,479
to eat it.

35
00:01:03,480 --> 00:01:05,128
I don't know how much reading you've

36
00:01:05,129 --> 00:01:07,529
done about ultra processed food.

37
00:01:07,530 --> 00:01:09,389
So one of the things

38
00:01:09,390 --> 00:01:10,829
with ultra processed food is it's

39
00:01:10,830 --> 00:01:12,449
very soft.

40
00:01:12,450 --> 00:01:14,339
And that has made our

41
00:01:14,340 --> 00:01:16,439
jaws stop growing properly.

42
00:01:18,540 --> 00:01:19,619
I've not heard this before.

43
00:01:19,620 --> 00:01:20,849
Is that true?

44
00:01:20,850 --> 00:01:23,219
Yeah. So children

45
00:01:23,220 --> 00:01:25,199
and we were raised on enough ultra

46
00:01:25,200 --> 00:01:26,129
processed food to make the

47
00:01:26,130 --> 00:01:26,969
difference.

48
00:01:26,970 --> 00:01:28,829
All of our jaws are more

49
00:01:28,830 --> 00:01:30,449
recessed, and the reason why we need

50
00:01:30,450 --> 00:01:32,159
our wisdom teeth taken out is.

51
00:01:32,160 --> 00:01:34,019
Apparently your jaw grows

52
00:01:34,020 --> 00:01:36,029
through hard work as

53
00:01:36,030 --> 00:01:37,529
it's forming as a child.

54
00:01:37,530 --> 00:01:39,179
And so it actually comes further

55
00:01:39,180 --> 00:01:41,039
out. And then you have space

56
00:01:41,040 --> 00:01:42,119
for your wisdom teeth.

57
00:01:42,120 --> 00:01:43,469
But with ultra processed food, it's

58
00:01:43,470 --> 00:01:45,179
really soft. So our jaws don't work

59
00:01:45,180 --> 00:01:46,919
anymore. And that's why everybody

60
00:01:46,920 --> 00:01:48,479
needs braces.

61
00:01:48,480 --> 00:01:50,579
Oh my goodness. I did not know this.

62
00:01:50,580 --> 00:01:51,749
Well, I would say this I ate a

63
00:01:51,750 --> 00:01:53,129
tremendous amount of cashews.

64
00:01:53,130 --> 00:01:54,479
And cashews are quite hard, so I

65
00:01:54,480 --> 00:01:56,309
suspect my jaw line is

66
00:01:56,310 --> 00:01:58,089
kind of being exercised.

67
00:01:58,090 --> 00:01:59,019
From the outside, it looks like your

68
00:01:59,020 --> 00:02:00,819
jawline is fine, but I don't know.

69
00:02:00,820 --> 00:02:01,959
Did you take your wisdom teeth out?

70
00:02:01,960 --> 00:02:03,189
Could you have had more of a jaw?

71
00:02:03,190 --> 00:02:04,190
Had you

72
00:02:05,320 --> 00:02:06,939
knocked everyday?

73
00:02:06,940 --> 00:02:07,749
I don't even know.

74
00:02:07,750 --> 00:02:09,159
I literally have no memory or

75
00:02:09,160 --> 00:02:10,508
recollection of having them taken

76
00:02:10,509 --> 00:02:12,189
out, so they must still be on.

77
00:02:12,190 --> 00:02:14,259
Right. And came in in a normal

78
00:02:14,260 --> 00:02:15,549
way. Like they didn't cause any

79
00:02:15,550 --> 00:02:17,379
problems. You didn't need anything.

80
00:02:17,380 --> 00:02:19,509
My superior jawline was

81
00:02:19,510 --> 00:02:21,189
sufficient to house them.

82
00:02:21,190 --> 00:02:23,169
All those cashew nuts and the clean

83
00:02:23,170 --> 00:02:24,699
living of Canada.

84
00:02:24,700 --> 00:02:25,700
So besides

85
00:02:26,710 --> 00:02:27,729
the chia seed.

86
00:02:27,730 --> 00:02:28,869
That's right. The ultra processed

87
00:02:28,870 --> 00:02:29,859
food.

88
00:02:29,860 --> 00:02:31,059
What is happening in Bethany's

89
00:02:31,060 --> 00:02:31,749
world?

90
00:02:31,750 --> 00:02:34,089
Just insanely busy.

91
00:02:34,090 --> 00:02:36,099
So I've been working from home

92
00:02:36,100 --> 00:02:37,809
this week. I managed to leave the

93
00:02:37,810 --> 00:02:39,729
house on Wednesday and

94
00:02:39,730 --> 00:02:41,559
now today it's Friday.

95
00:02:41,560 --> 00:02:43,389
We're doing this at 8 a.m..

96
00:02:43,390 --> 00:02:45,669
I have back to back meetings

97
00:02:45,670 --> 00:02:47,349
like literally back to back in my

98
00:02:47,350 --> 00:02:49,209
calendar. Not a sliver of

99
00:02:49,210 --> 00:02:51,459
space until half six.

100
00:02:51,460 --> 00:02:52,899
So luckily my husband is working

101
00:02:52,900 --> 00:02:54,549
from home today and so he's going to

102
00:02:54,550 --> 00:02:56,829
bring me some lunch around noon

103
00:02:56,830 --> 00:02:58,239
so that I could continue to eat,

104
00:02:58,240 --> 00:02:59,739
although at the rate it takes me to

105
00:02:59,740 --> 00:03:00,999
eat the chia seeds, I might not

106
00:03:01,000 --> 00:03:02,109
actually have finished them by the

107
00:03:02,110 --> 00:03:03,549
time lunch arrives.

108
00:03:03,550 --> 00:03:04,879
Okay, so that's fabulous.

109
00:03:04,880 --> 00:03:06,309
So you have the in-house servant, in

110
00:03:06,310 --> 00:03:07,479
this case being your husband to

111
00:03:07,480 --> 00:03:08,569
deliver your lunch for.

112
00:03:08,570 --> 00:03:09,639
Why not call him that?

113
00:03:13,310 --> 00:03:14,319
That's that's I mean, this is the

114
00:03:14,320 --> 00:03:16,179
give and take in a relationship, so

115
00:03:16,180 --> 00:03:17,559
I don't know either. A quick note, I

116
00:03:17,560 --> 00:03:18,939
just had this small thought this

117
00:03:18,940 --> 00:03:20,589
morning. So I'm working with a

118
00:03:20,590 --> 00:03:21,969
couple individuals right now that

119
00:03:21,970 --> 00:03:23,679
are they do operations.

120
00:03:23,680 --> 00:03:25,779
But really the granular operations

121
00:03:25,780 --> 00:03:27,249
if you want to call it that.

122
00:03:27,250 --> 00:03:29,319
One of the individuals was asking

123
00:03:29,320 --> 00:03:30,699
me for a recommendation for

124
00:03:30,700 --> 00:03:31,929
employment law.

125
00:03:31,930 --> 00:03:33,879
So I just did a quick screenshot

126
00:03:33,880 --> 00:03:35,739
of recommendations coming

127
00:03:35,740 --> 00:03:37,539
out of the Operations Nation slack

128
00:03:37,540 --> 00:03:39,849
channel to her that operations

129
00:03:39,850 --> 00:03:41,259
Nation slack channel is good for

130
00:03:41,260 --> 00:03:43,299
this kind of thing, which is vendor

131
00:03:43,300 --> 00:03:45,069
sourcing recommendations around

132
00:03:45,070 --> 00:03:46,749
products and people and that sort of

133
00:03:46,750 --> 00:03:47,619
thing is fabulous for that.

134
00:03:47,620 --> 00:03:49,449
People are very

135
00:03:49,450 --> 00:03:51,069
proactive. I would say, in terms of

136
00:03:51,070 --> 00:03:52,869
giving thoughts on who to speak to,

137
00:03:52,870 --> 00:03:54,399
who's good to, you know, to work

138
00:03:54,400 --> 00:03:55,689
with these sorts of things.

139
00:03:55,690 --> 00:03:56,679
The other thought was just the

140
00:03:56,680 --> 00:03:57,999
obvious one, which is she's an

141
00:03:58,000 --> 00:03:59,769
operator. She's at that level.

142
00:03:59,770 --> 00:04:01,149
She'd be perfect to be part of the

143
00:04:01,150 --> 00:04:03,009
operations nation, or at least on

144
00:04:03,010 --> 00:04:04,809
the slack channel, is just a good,

145
00:04:04,810 --> 00:04:06,009
useful tool.

146
00:04:06,010 --> 00:04:07,869
When these questions pop up like

147
00:04:07,870 --> 00:04:09,219
this that are quite straightforward,

148
00:04:09,220 --> 00:04:11,349
where people can respond back

149
00:04:11,350 --> 00:04:12,669
and give you the kind of collective

150
00:04:12,670 --> 00:04:14,259
set of good answers.

151
00:04:14,260 --> 00:04:15,519
And it's free.

152
00:04:15,520 --> 00:04:17,169
It's definitely a good community for

153
00:04:17,170 --> 00:04:19,088
like having problems with my pension

154
00:04:19,089 --> 00:04:20,949
provider or

155
00:04:20,950 --> 00:04:22,959
the health insurance quote

156
00:04:22,960 --> 00:04:24,399
has gone up by 40%.

157
00:04:24,400 --> 00:04:25,209
What do I do?

158
00:04:25,210 --> 00:04:26,409
Or is that the same for everyone

159
00:04:26,410 --> 00:04:27,339
else? Like those are the types of

160
00:04:27,340 --> 00:04:28,629
questions you see.

161
00:04:28,630 --> 00:04:30,099
I went to a comedy show on Sunday

162
00:04:30,100 --> 00:04:31,989
night and then a comedy show on

163
00:04:31,990 --> 00:04:33,999
Wednesday night, and I was telling

164
00:04:34,000 --> 00:04:35,349
a friend and he was just like, is

165
00:04:35,350 --> 00:04:36,249
this a new thing?

166
00:04:36,250 --> 00:04:37,629
Like, it's going to comedy shows

167
00:04:37,630 --> 00:04:38,630
you're saying?

168
00:04:40,900 --> 00:04:42,249
And said, in that kind of like

169
00:04:42,250 --> 00:04:44,109
snide, dismissive way,

170
00:04:44,110 --> 00:04:45,849
like, what is up with you?

171
00:04:45,850 --> 00:04:47,469
It's not our new thing.

172
00:04:47,470 --> 00:04:49,629
And I actually booked them ages

173
00:04:49,630 --> 00:04:51,639
ago, maybe July, August time.

174
00:04:51,640 --> 00:04:54,009
And it just happened to arrive now.

175
00:04:54,010 --> 00:04:55,329
And there were two very different

176
00:04:55,330 --> 00:04:56,859
comedy shows. So one of them was in

177
00:04:56,860 --> 00:04:58,089
Hackney.

178
00:04:58,090 --> 00:04:59,439
It was very trendy.

179
00:04:59,440 --> 00:05:01,419
The guy is a Sri Lankan

180
00:05:01,420 --> 00:05:03,609
comic, as you do

181
00:05:03,610 --> 00:05:04,809
a Sri Lankan comic that kind of

182
00:05:04,810 --> 00:05:06,159
looks like Jesus, like he is very

183
00:05:06,160 --> 00:05:08,049
curly hair down to his waist.

184
00:05:08,050 --> 00:05:09,369
Very good looking guy.

185
00:05:09,370 --> 00:05:11,319
And he studied

186
00:05:11,320 --> 00:05:12,879
medicine and then decided not to

187
00:05:12,880 --> 00:05:14,829
become a medic, and

188
00:05:14,830 --> 00:05:16,929
then became a developer and

189
00:05:16,930 --> 00:05:18,609
a comic and is trying to be a comic

190
00:05:18,610 --> 00:05:20,439
more than a developer.

191
00:05:20,440 --> 00:05:22,359
So anyhow, I

192
00:05:22,360 --> 00:05:24,339
have never felt like more out

193
00:05:24,340 --> 00:05:25,239
of place.

194
00:05:25,240 --> 00:05:26,559
Everybody was trendy.

195
00:05:26,560 --> 00:05:27,699
Like, I don't think there were any

196
00:05:27,700 --> 00:05:28,749
other women in the room.

197
00:05:28,750 --> 00:05:30,879
They were just non-binary.

198
00:05:30,880 --> 00:05:32,709
I guess we call them women anymore.

199
00:05:32,710 --> 00:05:34,119
Like, I would have chosen to be

200
00:05:34,120 --> 00:05:35,979
non-binary had I had

201
00:05:35,980 --> 00:05:37,029
the option, because you kind of look

202
00:05:37,030 --> 00:05:38,199
at it, you're like, oh, I could be a

203
00:05:38,200 --> 00:05:39,669
woman, and that looks like a really

204
00:05:39,670 --> 00:05:41,499
bad deal, or I can opt

205
00:05:41,500 --> 00:05:42,500
out

206
00:05:43,660 --> 00:05:45,789
and I think I'm going to opt out.

207
00:05:45,790 --> 00:05:46,839
Yeah.

208
00:05:46,840 --> 00:05:47,889
Fair enough. Yeah.

209
00:05:47,890 --> 00:05:50,499
And then every man had a mustache.

210
00:05:50,500 --> 00:05:52,069
Is that. Is that a thing right now?

211
00:05:52,070 --> 00:05:53,659
It really is a thing.

212
00:05:53,660 --> 00:05:55,519
And here we are like middle

213
00:05:55,520 --> 00:05:57,439
class clap mites

214
00:05:57,440 --> 00:05:59,539
nots with mustaches

215
00:05:59,540 --> 00:06:00,559
and combat boots.

216
00:06:00,560 --> 00:06:01,609
And I was just like.

217
00:06:01,610 --> 00:06:03,679
And not non-binary.

218
00:06:03,680 --> 00:06:04,459
Yeah.

219
00:06:04,460 --> 00:06:06,139
And I just never felt so uncool in

220
00:06:06,140 --> 00:06:07,219
my life.

221
00:06:07,220 --> 00:06:08,749
But it was quite a good show other

222
00:06:08,750 --> 00:06:10,759
than just feeling like such

223
00:06:10,760 --> 00:06:11,749
the square.

224
00:06:11,750 --> 00:06:14,179
And then on Wednesday

225
00:06:14,180 --> 00:06:16,669
night, it was a

226
00:06:16,670 --> 00:06:18,709
Irish comedian who's a couple years

227
00:06:18,710 --> 00:06:20,059
older than me. Woman.

228
00:06:20,060 --> 00:06:21,229
And other than the fact that at

229
00:06:21,230 --> 00:06:22,789
least 90% of the audience were

230
00:06:22,790 --> 00:06:24,709
Irish, everybody was a middle

231
00:06:24,710 --> 00:06:26,539
aged woman and maybe

232
00:06:26,540 --> 00:06:28,309
one and four brought their husbands

233
00:06:28,310 --> 00:06:30,229
along. So I felt

234
00:06:30,230 --> 00:06:32,179
totally in

235
00:06:32,180 --> 00:06:34,339
my element. We all belonged.

236
00:06:34,340 --> 00:06:36,199
We were all square, we

237
00:06:36,200 --> 00:06:38,929
were all 50 ish,

238
00:06:38,930 --> 00:06:41,179
and it was quite good, but totally

239
00:06:41,180 --> 00:06:43,009
different sets.

240
00:06:43,010 --> 00:06:44,839
One was around

241
00:06:44,840 --> 00:06:46,699
like his was around not belonging

242
00:06:46,700 --> 00:06:48,559
and racism

243
00:06:48,560 --> 00:06:49,879
and privilege.

244
00:06:49,880 --> 00:06:52,219
And hers is about

245
00:06:52,220 --> 00:06:54,649
menopause and sex.

246
00:06:54,650 --> 00:06:56,449
And husbands and children.

247
00:06:56,450 --> 00:06:57,589
Congratulations though to you.

248
00:06:57,590 --> 00:06:59,509
Because actually booking kind of

249
00:06:59,510 --> 00:07:01,429
not social but like just outings

250
00:07:01,430 --> 00:07:03,019
of fun and enjoyment.

251
00:07:03,020 --> 00:07:04,129
You know, I feel like my calendar

252
00:07:04,130 --> 00:07:05,359
right now is pretty sparse in that

253
00:07:05,360 --> 00:07:06,259
respect.

254
00:07:06,260 --> 00:07:08,299
So I'm always actively looking

255
00:07:08,300 --> 00:07:09,589
for, I don't know, just interesting

256
00:07:09,590 --> 00:07:11,329
things that would be, I don't know,

257
00:07:11,330 --> 00:07:13,309
fun, enjoyable to do and were doable

258
00:07:13,310 --> 00:07:14,779
just given our schedule with the

259
00:07:14,780 --> 00:07:15,919
kids and all that jazz.

260
00:07:15,920 --> 00:07:16,879
But nothing has really popped up

261
00:07:16,880 --> 00:07:18,829
quite recently. That seems useful

262
00:07:18,830 --> 00:07:19,940
enough or good enough to go to.

263
00:07:24,110 --> 00:07:25,969
All right. So we have got

264
00:07:25,970 --> 00:07:27,919
a great topic for today,

265
00:07:27,920 --> 00:07:29,779
which is a returning topic, as

266
00:07:29,780 --> 00:07:31,729
it were, building an AI for

267
00:07:31,730 --> 00:07:33,199
Organization Part two.

268
00:07:33,200 --> 00:07:34,429
We have an amazing returning guest

269
00:07:34,430 --> 00:07:36,229
for this, which is Charlie Cowan.

270
00:07:36,230 --> 00:07:37,549
We had so much fun with Charlie that

271
00:07:37,550 --> 00:07:39,199
we had a second conversation.

272
00:07:39,200 --> 00:07:40,639
And of course, he is an AI

273
00:07:40,640 --> 00:07:42,469
strategist and AI change

274
00:07:42,470 --> 00:07:44,329
agent within organizations

275
00:07:44,330 --> 00:07:46,219
today. So before we get into our

276
00:07:46,220 --> 00:07:48,259
part two with Charlie, Just wanted

277
00:07:48,260 --> 00:07:50,119
to talk about a couple bits and

278
00:07:50,120 --> 00:07:51,379
pieces here.

279
00:07:51,380 --> 00:07:52,369
One of the elements that he spoke

280
00:07:52,370 --> 00:07:54,289
about was leadership.

281
00:07:54,290 --> 00:07:56,629
Using AI themselves

282
00:07:56,630 --> 00:07:58,129
on a daily basis to really

283
00:07:58,130 --> 00:08:00,019
understand what they're talking

284
00:08:00,020 --> 00:08:01,219
about. So if they're going to
encourage

285
00:08:01,220 --> 00:08:02,329
the rest of the organization to

286
00:08:02,330 --> 00:08:04,099
experiment with AI, the starting

287
00:08:04,100 --> 00:08:05,869
point for these things, outside of

288
00:08:05,870 --> 00:08:07,549
speaking about it, is for the

289
00:08:07,550 --> 00:08:08,779
leadership to really understand what

290
00:08:08,780 --> 00:08:09,859
they're actually talking about, or

291
00:08:09,860 --> 00:08:11,569
what value they're actually getting

292
00:08:11,570 --> 00:08:12,739
themselves, to be able to express it

293
00:08:12,740 --> 00:08:13,759
in a more articulate way,

294
00:08:13,760 --> 00:08:14,839
presumably.

295
00:08:14,840 --> 00:08:15,949
What do you make of that? Is that a

296
00:08:15,950 --> 00:08:17,959
hard requirement, do you think?

297
00:08:17,960 --> 00:08:18,739
I don't know if it's a hard

298
00:08:18,740 --> 00:08:20,719
requirement, but like once you start

299
00:08:20,720 --> 00:08:22,549
using it, you can't go back.

300
00:08:22,550 --> 00:08:24,349
So I think it's like the it's more

301
00:08:24,350 --> 00:08:26,209
like in order to understand

302
00:08:26,210 --> 00:08:28,099
what you're talking about, you

303
00:08:28,100 --> 00:08:30,439
should. And also to understand the

304
00:08:30,440 --> 00:08:31,999
what is the constraints, what it's

305
00:08:32,000 --> 00:08:33,168
good at, what it's not.

306
00:08:33,169 --> 00:08:35,329
But it's also just like a really

307
00:08:35,330 --> 00:08:36,499
convenient tool.

308
00:08:36,500 --> 00:08:37,999
So whether or not you're using it

309
00:08:38,000 --> 00:08:39,259
for other people, you should start

310
00:08:39,260 --> 00:08:40,428
to play around with it and use it

311
00:08:40,429 --> 00:08:41,389
for yourself.

312
00:08:41,390 --> 00:08:42,739
It's just so nice.

313
00:08:42,740 --> 00:08:44,689
So we had a new

314
00:08:44,690 --> 00:08:45,729
commission policy.

315
00:08:45,730 --> 00:08:46,819
Somebody had written it.

316
00:08:46,820 --> 00:08:48,919
It was just in legalese,

317
00:08:48,920 --> 00:08:50,599
through and through.

318
00:08:50,600 --> 00:08:52,159
In the old world, I would have had

319
00:08:52,160 --> 00:08:54,079
to rewrite that or have somebody

320
00:08:54,080 --> 00:08:55,589
rewrite it or, you know, give it to

321
00:08:55,590 --> 00:08:57,049
our copywriter to just put it in our

322
00:08:57,050 --> 00:08:58,309
tone of voice or, you know, it'd be

323
00:08:58,310 --> 00:08:59,479
a total waste of time.

324
00:08:59,480 --> 00:09:00,469
And instead, I stuck it in.

325
00:09:00,470 --> 00:09:02,959
Claude said, keep it exactly

326
00:09:02,960 --> 00:09:05,059
the same sections

327
00:09:05,060 --> 00:09:06,979
don't change the content, just

328
00:09:06,980 --> 00:09:08,899
make it sound like our tone of tone,

329
00:09:08,900 --> 00:09:09,769
of voice.

330
00:09:09,770 --> 00:09:11,209
And under a minute later, it was

331
00:09:11,210 --> 00:09:12,169
done.

332
00:09:12,170 --> 00:09:13,189
It was really good.

333
00:09:13,190 --> 00:09:14,599
The only annoying thing is Claude

334
00:09:14,600 --> 00:09:16,489
loses formatting, so I

335
00:09:16,490 --> 00:09:17,689
had to go and reformat it.

336
00:09:17,690 --> 00:09:19,639
But like, it's just so nice

337
00:09:19,640 --> 00:09:21,529
to have those options or

338
00:09:21,530 --> 00:09:24,409
handed a legalese documents

339
00:09:24,410 --> 00:09:26,449
to have to understand.

340
00:09:26,450 --> 00:09:27,979
Couldn't understand one paragraph.

341
00:09:27,980 --> 00:09:29,299
No matter how many times I looked at

342
00:09:29,300 --> 00:09:31,399
it, threw it

343
00:09:31,400 --> 00:09:33,289
into ChatGPT to

344
00:09:33,290 --> 00:09:34,669
just explain to me what the

345
00:09:34,670 --> 00:09:36,049
paragraph is talking about and what

346
00:09:36,050 --> 00:09:37,039
I need to worry about.

347
00:09:37,040 --> 00:09:38,899
And it came out and told me

348
00:09:38,900 --> 00:09:39,950
it's really nice.

349
00:09:41,300 --> 00:09:42,739
It just makes life easier.

350
00:09:42,740 --> 00:09:44,039
I mean, do you use it.

351
00:09:44,040 --> 00:09:45,839
For me creating things I need to

352
00:09:45,840 --> 00:09:47,969
distribute? I always end up passing

353
00:09:47,970 --> 00:09:48,929
it through ChatGPT.

354
00:09:48,930 --> 00:09:50,609
Now to kind of do a once over in

355
00:09:50,610 --> 00:09:52,559
terms of making it better.

356
00:09:52,560 --> 00:09:54,179
And all things being equal, I am not

357
00:09:54,180 --> 00:09:55,259
a fabulous writer.

358
00:09:55,260 --> 00:09:56,759
I can write and I can get my message

359
00:09:56,760 --> 00:09:57,959
across, but every time I put it

360
00:09:57,960 --> 00:10:00,059
through ChatGPT and get my output,

361
00:10:00,060 --> 00:10:01,889
it is dramatically better for

362
00:10:01,890 --> 00:10:03,209
the most part. So I'm like, oh,

363
00:10:03,210 --> 00:10:04,739
great. And I might tweak it here and

364
00:10:04,740 --> 00:10:06,419
there, but definitely use it for

365
00:10:06,420 --> 00:10:08,039
that purpose. And then the reverse,

366
00:10:08,040 --> 00:10:09,509
which is where you just pointed out

367
00:10:09,510 --> 00:10:11,399
any kind of incoming document that

368
00:10:11,400 --> 00:10:12,869
is highly annoying for me to

369
00:10:12,870 --> 00:10:14,459
understand and read, passing it

370
00:10:14,460 --> 00:10:16,019
through to you, but again, to either

371
00:10:16,020 --> 00:10:18,059
simplify it or kind of expressing

372
00:10:18,060 --> 00:10:19,049
it in a different way, where I can

373
00:10:19,050 --> 00:10:20,579
actually pull out the key bits that

374
00:10:20,580 --> 00:10:21,959
I need to quickly.

375
00:10:21,960 --> 00:10:23,789
It's very useful for that.

376
00:10:23,790 --> 00:10:25,739
I'm a little bit sometimes

377
00:10:25,740 --> 00:10:27,599
skeptical, but for things that are

378
00:10:27,600 --> 00:10:29,189
rather important where the detail

379
00:10:29,190 --> 00:10:31,379
matters, I'm a little more worried

380
00:10:31,380 --> 00:10:32,759
sometimes that it's not pulling out

381
00:10:32,760 --> 00:10:33,929
all the bits that I actually need,

382
00:10:33,930 --> 00:10:35,429
to be honest. There's some level of

383
00:10:35,430 --> 00:10:37,259
my going back to the source

384
00:10:37,260 --> 00:10:38,789
document just to check things and

385
00:10:38,790 --> 00:10:40,169
verify that at all.

386
00:10:40,170 --> 00:10:41,459
I'm getting what I need, basically.

387
00:10:41,460 --> 00:10:43,379
But my suspicion is, and

388
00:10:43,380 --> 00:10:44,879
this is separate from ChatGPT and

389
00:10:44,880 --> 00:10:45,899
the generic tools right now.

390
00:10:45,900 --> 00:10:47,609
But there's probably again, probably

391
00:10:47,610 --> 00:10:49,439
AI agents out there that

392
00:10:49,440 --> 00:10:50,849
are verticals that probably do a

393
00:10:50,850 --> 00:10:52,529
much better job of this to ensure

394
00:10:52,530 --> 00:10:53,699
that you're getting exactly what you

395
00:10:53,700 --> 00:10:55,019
need with a higher level of

396
00:10:55,020 --> 00:10:56,669
confidence and accuracy than using

397
00:10:56,670 --> 00:10:57,689
the generic tools.

398
00:10:57,690 --> 00:10:59,729
But that aside, your

399
00:10:59,730 --> 00:11:01,649
use cases or my use cases, and I'm

400
00:11:01,650 --> 00:11:03,509
sure for any C-suite leader or

401
00:11:03,510 --> 00:11:05,369
that kind of incoming document

402
00:11:05,370 --> 00:11:07,079
of complexity, output of actual

403
00:11:07,080 --> 00:11:08,559
things that you need to communicate

404
00:11:08,560 --> 00:11:09,869
seems like a no brainer.

405
00:11:09,870 --> 00:11:12,059
And then the other one I use is

406
00:11:12,060 --> 00:11:14,009
if I have writer's block

407
00:11:14,010 --> 00:11:15,299
or I don't quite know how to get

408
00:11:15,300 --> 00:11:17,069
started, and that blank sheet of

409
00:11:17,070 --> 00:11:18,929
paper a moment, rather than

410
00:11:18,930 --> 00:11:20,589
just the willpower of getting past

411
00:11:20,590 --> 00:11:22,139
the blank sheet of paper that I have

412
00:11:22,140 --> 00:11:24,089
used for the last, whatever,

413
00:11:24,090 --> 00:11:26,159
30, 40 years of my life,

414
00:11:26,160 --> 00:11:28,229
I'll go to ChatGPT or Claude

415
00:11:28,230 --> 00:11:30,059
and just say, this is what I have

416
00:11:30,060 --> 00:11:31,349
to do.

417
00:11:31,350 --> 00:11:32,729
How should I get started?

418
00:11:32,730 --> 00:11:34,619
You know, and just kind of what

419
00:11:34,620 --> 00:11:36,209
would be the typical arguments in

420
00:11:36,210 --> 00:11:37,859
this case? Or what would the AI

421
00:11:37,860 --> 00:11:38,939
structure be?

422
00:11:38,940 --> 00:11:40,469
And even if I don't end up using

423
00:11:40,470 --> 00:11:41,669
that structure. I no longer have a

424
00:11:41,670 --> 00:11:43,289
blank sheet of paper, and that just

425
00:11:43,290 --> 00:11:44,669
really helps get me started.

426
00:11:44,670 --> 00:11:46,109
And I don't need that same level of

427
00:11:46,110 --> 00:11:47,849
willpower I used to have.

428
00:11:47,850 --> 00:11:49,469
Yeah, I think you're exactly right.

429
00:11:49,470 --> 00:11:51,359
So I was having to come up with

430
00:11:51,360 --> 00:11:53,369
kind of a interview process for

431
00:11:53,370 --> 00:11:55,319
sales reps, and also the kinds

432
00:11:55,320 --> 00:11:57,419
of questions that I want to ask for

433
00:11:57,420 --> 00:11:58,769
different elements that I was

434
00:11:58,770 --> 00:12:00,029
looking for in terms of being able

435
00:12:00,030 --> 00:12:02,009
to make judgments on the candidates

436
00:12:02,010 --> 00:12:04,109
and ChatGPT for that kind of generic

437
00:12:04,110 --> 00:12:05,999
interview process and questions

438
00:12:06,000 --> 00:12:07,319
was absolutely fabulous.

439
00:12:07,320 --> 00:12:08,759
And to your point, instead of me

440
00:12:08,760 --> 00:12:10,409
having to I mean, I've interviewed

441
00:12:10,410 --> 00:12:11,939
sales reps so many times before with

442
00:12:11,940 --> 00:12:13,259
so many different question types and

443
00:12:13,260 --> 00:12:14,339
processes.

444
00:12:14,340 --> 00:12:15,869
It's all vaguely in the back of my

445
00:12:15,870 --> 00:12:17,789
head, but it requires me

446
00:12:17,790 --> 00:12:19,379
if I need to do net new, to think

447
00:12:19,380 --> 00:12:20,459
about it again and kind of think

448
00:12:20,460 --> 00:12:21,839
through. Okay, what did I do before?

449
00:12:21,840 --> 00:12:22,979
Do I have some previous documents I

450
00:12:22,980 --> 00:12:24,119
can rip off?

451
00:12:24,120 --> 00:12:25,379
All that takes thinking time

452
00:12:25,380 --> 00:12:27,539
basically, whereas now

453
00:12:27,540 --> 00:12:29,219
I can just ignore all that wholesale

454
00:12:29,220 --> 00:12:31,169
and very rapidly get

455
00:12:31,170 --> 00:12:32,789
the process, get the steps, get the

456
00:12:32,790 --> 00:12:34,769
questions and get the

457
00:12:34,770 --> 00:12:36,659
criteria boxes all set up

458
00:12:36,660 --> 00:12:38,519
basically within whatever, half

459
00:12:38,520 --> 00:12:40,439
an hour and then ship it out to Emma

460
00:12:40,440 --> 00:12:41,669
as part of the process and boom,

461
00:12:41,670 --> 00:12:42,779
rocking and rolling.

462
00:12:42,780 --> 00:12:43,889
So I think for that kind of

463
00:12:43,890 --> 00:12:45,329
activity, for stuff that you kind of

464
00:12:45,330 --> 00:12:47,219
know, but it's annoying to have

465
00:12:47,220 --> 00:12:49,469
to re remember everything.

466
00:12:49,470 --> 00:12:51,119
That's a great use case.

467
00:12:51,120 --> 00:12:52,709
I love that like the annoying re

468
00:12:52,710 --> 00:12:53,879
remembering.

469
00:12:53,880 --> 00:12:55,139
That's why we're like yeah we're so

470
00:12:55,140 --> 00:12:56,009
over it.

471
00:12:56,010 --> 00:12:58,019
So Charlie talked several

472
00:12:58,020 --> 00:12:59,489
times about this, which is every

473
00:12:59,490 --> 00:13:01,859
time he has to do something,

474
00:13:01,860 --> 00:13:03,149
he actually doesn't go to the

475
00:13:03,150 --> 00:13:04,979
generic box, as it were.

476
00:13:04,980 --> 00:13:06,719
He always creates a project for

477
00:13:06,720 --> 00:13:08,879
himself and gives context

478
00:13:08,880 --> 00:13:10,889
or the prompts around the particular

479
00:13:10,890 --> 00:13:12,569
subject matter that he's wanting a

480
00:13:12,570 --> 00:13:14,099
response to in this case.

481
00:13:14,100 --> 00:13:15,809
And so his go to effectively is to

482
00:13:15,810 --> 00:13:17,429
create projects, is what I'm saying.

483
00:13:17,430 --> 00:13:19,379
So my curiosity with you,

484
00:13:19,380 --> 00:13:21,389
do you use projects and

485
00:13:21,390 --> 00:13:22,379
is it useful?

486
00:13:22,380 --> 00:13:23,969
I'm very impatient and a little bit

487
00:13:23,970 --> 00:13:24,749
lazy.

488
00:13:24,750 --> 00:13:26,909
And so I'm using gen

489
00:13:26,910 --> 00:13:29,369
AI to make my life easier

490
00:13:29,370 --> 00:13:30,659
really quickly.

491
00:13:30,660 --> 00:13:32,609
And I find projects kind of annoying

492
00:13:32,610 --> 00:13:34,409
because it slows me down.

493
00:13:34,410 --> 00:13:36,269
And I think if I were trying to do

494
00:13:36,270 --> 00:13:38,079
the next level of work, it would be

495
00:13:38,080 --> 00:13:39,129
worth the input.

496
00:13:39,130 --> 00:13:41,139
But for the most part, I find

497
00:13:41,140 --> 00:13:43,059
that without using projects, the

498
00:13:43,060 --> 00:13:44,949
output is good enough that

499
00:13:44,950 --> 00:13:47,259
I don't find it worthwhile

500
00:13:47,260 --> 00:13:48,879
putting in that next layer.

501
00:13:48,880 --> 00:13:51,039
I'm not like Charlie, who

502
00:13:51,040 --> 00:13:53,169
has created an entire project

503
00:13:53,170 --> 00:13:55,389
without, you know, to tie a product

504
00:13:55,390 --> 00:13:57,309
and code without using

505
00:13:57,310 --> 00:13:58,539
anybody but him.

506
00:13:58,540 --> 00:14:00,429
I just want to know

507
00:14:00,430 --> 00:14:02,559
what would be some good agenda items

508
00:14:02,560 --> 00:14:04,929
for a sales kickoff.

509
00:14:04,930 --> 00:14:06,669
As Charlie described, some of the

510
00:14:06,670 --> 00:14:08,139
projects that he does, and some of

511
00:14:08,140 --> 00:14:09,879
the prompts that he sets up are

512
00:14:09,880 --> 00:14:11,289
super complex.

513
00:14:11,290 --> 00:14:12,819
Like, like, you know, he described

514
00:14:12,820 --> 00:14:14,769
like doing a five page prompt.

515
00:14:14,770 --> 00:14:16,899
I think it was a 20 page prompt.

516
00:14:16,900 --> 00:14:18,249
Yeah. Yeah, that sounds like a lot

517
00:14:18,250 --> 00:14:19,669
of thinking.

518
00:14:19,670 --> 00:14:21,789
Seems like so the subset

519
00:14:21,790 --> 00:14:23,919
of that is he has created

520
00:14:23,920 --> 00:14:26,079
what he called a CEO pod

521
00:14:26,080 --> 00:14:27,759
with pre-built prompts for

522
00:14:27,760 --> 00:14:29,589
interesting topics, and I thought

523
00:14:29,590 --> 00:14:31,089
that seemed very interesting to me

524
00:14:31,090 --> 00:14:32,379
as a senior executive.

525
00:14:32,380 --> 00:14:33,459
So essentially, what he's done is

526
00:14:33,460 --> 00:14:35,199
taking that product trumpet over the

527
00:14:35,200 --> 00:14:37,239
trumpet before It's usually used for

528
00:14:37,240 --> 00:14:39,249
like sales enablement for a buyer

529
00:14:39,250 --> 00:14:40,329
to go in there with all the

530
00:14:40,330 --> 00:14:41,929
documentation around the product,

531
00:14:41,930 --> 00:14:43,959
this and the other is used it

532
00:14:43,960 --> 00:14:46,089
to basically be a place for

533
00:14:46,090 --> 00:14:48,159
anybody that goes to his LinkedIn

534
00:14:48,160 --> 00:14:50,139
feed to go into that little

535
00:14:50,140 --> 00:14:52,269
box world, and in that box world

536
00:14:52,270 --> 00:14:54,039
that is called the CEO pod.

537
00:14:54,040 --> 00:14:55,599
He has three prompts that are

538
00:14:55,600 --> 00:14:56,589
sitting there as defaults.

539
00:14:56,590 --> 00:14:58,449
One prompt is for

540
00:14:58,450 --> 00:15:00,849
all hands prep and

541
00:15:00,850 --> 00:15:02,469
the definition or the.

542
00:15:02,470 --> 00:15:03,729
I guess the objective of that prompt

543
00:15:03,730 --> 00:15:05,709
is to develop your messaging

544
00:15:05,710 --> 00:15:07,329
and talk track to convey your vision

545
00:15:07,330 --> 00:15:08,829
and inspire your employees.

546
00:15:08,830 --> 00:15:11,019
So it's kind of a drop in prompt for

547
00:15:11,020 --> 00:15:12,849
a project, presumably for all

548
00:15:12,850 --> 00:15:13,839
hands that you could use going

549
00:15:13,840 --> 00:15:16,059
forward. So my suspicion

550
00:15:16,060 --> 00:15:18,399
is for people like us where

551
00:15:18,400 --> 00:15:20,499
we don't do spending hours building

552
00:15:20,500 --> 00:15:22,479
prompts of five pages, that

553
00:15:22,480 --> 00:15:23,799
there should be good, healthy,

554
00:15:23,800 --> 00:15:25,599
generic prompts for things like all

555
00:15:25,600 --> 00:15:27,399
hands or this then the other, that

556
00:15:27,400 --> 00:15:29,379
we can simply take it and drop in

557
00:15:29,380 --> 00:15:32,019
and use it, where we actually get

558
00:15:32,020 --> 00:15:33,279
a better answer than what we would

559
00:15:33,280 --> 00:15:35,469
have got generically to the box, but

560
00:15:35,470 --> 00:15:37,419
I don't have to think about

561
00:15:37,420 --> 00:15:38,769
creating that problem myself in this

562
00:15:38,770 --> 00:15:39,909
case.

563
00:15:39,910 --> 00:15:42,759
Yeah. And then the question is

564
00:15:42,760 --> 00:15:43,759
who's going to provide it?

565
00:15:43,760 --> 00:15:45,819
Like is it going to be the rise of

566
00:15:45,820 --> 00:15:48,039
a bunch of apps that are

567
00:15:48,040 --> 00:15:49,869
basically rap arounds of the

568
00:15:49,870 --> 00:15:51,009
labs?

569
00:15:51,010 --> 00:15:52,869
Is it going to be the

570
00:15:52,870 --> 00:15:54,699
labs themselves providing it,

571
00:15:54,700 --> 00:15:56,589
or is it going to be consultants, or

572
00:15:56,590 --> 00:15:57,759
is it going to be like your really

573
00:15:57,760 --> 00:15:59,649
talented 22

574
00:15:59,650 --> 00:16:01,119
year old who comes into the business

575
00:16:01,120 --> 00:16:02,649
is like, you know that going to be

576
00:16:02,650 --> 00:16:03,909
their future, but I don't think it's

577
00:16:03,910 --> 00:16:05,259
gonna be the future for very long,

578
00:16:05,260 --> 00:16:07,179
because ChatGPT

579
00:16:07,180 --> 00:16:08,799
is just going to create a better

580
00:16:08,800 --> 00:16:10,659
user interface and eat up

581
00:16:10,660 --> 00:16:12,129
all of these other businesses.

582
00:16:12,130 --> 00:16:14,019
Another one that I hear is not doing

583
00:16:14,020 --> 00:16:15,519
well is 11 X.

584
00:16:15,520 --> 00:16:17,469
Like the first automated

585
00:16:17,470 --> 00:16:19,569
one of these like SDR offerings,

586
00:16:19,570 --> 00:16:21,969
they got to 10 million super fast.

587
00:16:21,970 --> 00:16:24,129
For what I hear, they're about to

588
00:16:24,130 --> 00:16:25,959
that 50% churn

589
00:16:25,960 --> 00:16:28,179
and wasn't a good product.

590
00:16:28,180 --> 00:16:29,799
Not worth it after people gave it a

591
00:16:29,800 --> 00:16:31,689
go. And then also just loads

592
00:16:31,690 --> 00:16:32,839
and loads of people are now building

593
00:16:32,840 --> 00:16:34,489
their own automated stores.

594
00:16:34,490 --> 00:16:36,799
And so that's another example of

595
00:16:36,800 --> 00:16:37,939
there's going to be a lot of crashing

596
00:16:37,940 --> 00:16:39,679
and burning while figuring out where

597
00:16:39,680 --> 00:16:41,689
the real value is, and that

598
00:16:41,690 --> 00:16:43,549
apps layer is definitely at

599
00:16:43,550 --> 00:16:45,409
risk and probably

600
00:16:45,410 --> 00:16:47,269
like how specialized

601
00:16:47,270 --> 00:16:48,079
can you go?

602
00:16:48,080 --> 00:16:50,209
So when you talk about like ones

603
00:16:50,210 --> 00:16:52,579
for lawyers, one's for

604
00:16:52,580 --> 00:16:54,409
doctors, like, you know, the very

605
00:16:54,410 --> 00:16:56,269
specific horizontals are probably

606
00:16:56,270 --> 00:16:57,109
safe.

607
00:16:57,110 --> 00:16:58,489
But let's say.

608
00:16:58,490 --> 00:16:59,749
I somehow feel it's like the hype

609
00:16:59,750 --> 00:17:01,609
curve where the companies come out

610
00:17:01,610 --> 00:17:03,109
of nowhere. They create first

611
00:17:03,110 --> 00:17:04,909
versions of the product around stars

612
00:17:04,910 --> 00:17:05,989
or whatever people are.

613
00:17:05,990 --> 00:17:06,979
Budgets are excited.

614
00:17:06,980 --> 00:17:08,239
They spend it, they use it.

615
00:17:08,240 --> 00:17:10,009
It's very disappointing for all

616
00:17:10,010 --> 00:17:11,088
sorts of reasons.

617
00:17:11,089 --> 00:17:13,129
Then we go through that phase where

618
00:17:13,130 --> 00:17:14,989
all these companies and products

619
00:17:14,990 --> 00:17:16,459
go to the trough of like

620
00:17:16,460 --> 00:17:18,348
disillusionment and like not getting

621
00:17:18,349 --> 00:17:19,759
the revenues, high churn, this, that

622
00:17:19,760 --> 00:17:20,598
and the other.

623
00:17:20,599 --> 00:17:22,699
But slowly but surely they figure

624
00:17:22,700 --> 00:17:24,019
it out and they figure out how to

625
00:17:24,020 --> 00:17:25,519
build real value and they come back

626
00:17:25,520 --> 00:17:26,868
up the curve eventually.

627
00:17:26,869 --> 00:17:29,119
So it feels like on balance,

628
00:17:29,120 --> 00:17:31,609
I would imagine this will happen in

629
00:17:31,610 --> 00:17:33,529
this sector, but there is

630
00:17:33,530 --> 00:17:35,209
that kind of risk that sits there in

631
00:17:35,210 --> 00:17:37,369
terms of open AI ultimately eating

632
00:17:37,370 --> 00:17:39,169
into these stocks in a way where

633
00:17:39,170 --> 00:17:41,089
somehow it becomes so easy

634
00:17:41,090 --> 00:17:42,829
for a single individual organization

635
00:17:42,830 --> 00:17:43,969
to do it all themselves.

636
00:17:43,970 --> 00:17:45,229
They don't actually need to go to

637
00:17:45,230 --> 00:17:46,369
these companies to buy these

638
00:17:46,370 --> 00:17:47,689
products, to your point.

639
00:17:47,690 --> 00:17:49,489
And it's all happening so fast.

640
00:17:49,490 --> 00:17:51,709
So you have these rises and falls

641
00:17:51,710 --> 00:17:53,539
within months, not

642
00:17:53,540 --> 00:17:54,589
years.

643
00:17:54,590 --> 00:17:55,729
Yeah, exactly.

644
00:17:55,730 --> 00:17:57,649
So what I wanted to ask you was

645
00:17:57,650 --> 00:17:59,929
how to communicate the importance

646
00:17:59,930 --> 00:18:01,969
of experimenting

647
00:18:01,970 --> 00:18:04,789
with ChatGPT and Claude

648
00:18:04,790 --> 00:18:07,129
and know PLM and Gemini

649
00:18:07,130 --> 00:18:08,659
across the organization to get

650
00:18:08,660 --> 00:18:10,849
people that dropped people's minds

651
00:18:10,850 --> 00:18:12,559
around the fact that this is very

652
00:18:12,560 --> 00:18:13,729
important for businesses going

653
00:18:13,730 --> 00:18:15,679
forward. And we need to start

654
00:18:15,680 --> 00:18:16,819
experimenting. We need to start

655
00:18:16,820 --> 00:18:18,019
doing things. And obviously, from an

656
00:18:18,020 --> 00:18:19,189
operator standpoint, there's all

657
00:18:19,190 --> 00:18:21,199
sorts of things you can do to make

658
00:18:21,200 --> 00:18:22,069
that happen.

659
00:18:22,070 --> 00:18:23,599
But purely in terms of just straight

660
00:18:23,600 --> 00:18:25,669
up communication to the company,

661
00:18:25,670 --> 00:18:27,229
what are the kind of like key

662
00:18:27,230 --> 00:18:29,119
messages that are

663
00:18:29,120 --> 00:18:30,949
the most powerful to just get into

664
00:18:30,950 --> 00:18:33,139
people's minds that this matters.

665
00:18:33,140 --> 00:18:34,339
And we need to do something about

666
00:18:34,340 --> 00:18:34,879
it.

667
00:18:34,880 --> 00:18:36,919
So I'm just stuck on your mention

668
00:18:36,920 --> 00:18:38,629
of Gemini, and I'm using Gemini

669
00:18:38,630 --> 00:18:39,979
right now, and I can't really tell

670
00:18:39,980 --> 00:18:41,209
if I'm using the paid version of the

671
00:18:41,210 --> 00:18:42,769
free version, but whatever I'm using

672
00:18:42,770 --> 00:18:44,029
just sucks.

673
00:18:44,030 --> 00:18:45,349
So first of all, I just want to add

674
00:18:45,350 --> 00:18:47,329
that in like it is

675
00:18:47,330 --> 00:18:49,039
massively underwhelming.

676
00:18:49,040 --> 00:18:51,409
It doesn't help me in any way.

677
00:18:51,410 --> 00:18:52,759
It doesn't do any of the things I

678
00:18:52,760 --> 00:18:54,439
would like it to do, and its answers

679
00:18:54,440 --> 00:18:55,440
are stupid.

680
00:18:58,160 --> 00:19:00,139
Slam on Gemini.

681
00:19:00,140 --> 00:19:01,969
Basically, everywhere I

682
00:19:01,970 --> 00:19:04,399
go on the Google suite,

683
00:19:04,400 --> 00:19:05,779
there's a little like, how can I

684
00:19:05,780 --> 00:19:06,919
help you box?

685
00:19:06,920 --> 00:19:08,599
And then I ask it to help me and it

686
00:19:08,600 --> 00:19:10,849
just does not do anything helpful.

687
00:19:10,850 --> 00:19:12,709
And so perfect example

688
00:19:12,710 --> 00:19:14,569
is I took

689
00:19:14,570 --> 00:19:16,729
the commission policy,

690
00:19:16,730 --> 00:19:18,409
ran it through Claude, made it

691
00:19:18,410 --> 00:19:20,869
friendly and pleasant, stuck

692
00:19:20,870 --> 00:19:23,269
it in a Google doc.

693
00:19:23,270 --> 00:19:25,459
Then I asked Google to format

694
00:19:25,460 --> 00:19:27,919
it like the above

695
00:19:27,920 --> 00:19:29,959
policy, and it just rewrote

696
00:19:29,960 --> 00:19:31,849
the policy for me in

697
00:19:31,850 --> 00:19:32,989
a stupid way.

698
00:19:32,990 --> 00:19:34,369
And it didn't reformat anything.

699
00:19:34,370 --> 00:19:35,929
And it only does stuff in the chat

700
00:19:35,930 --> 00:19:36,799
box.

701
00:19:36,800 --> 00:19:38,269
But like, I don't care about the

702
00:19:38,270 --> 00:19:40,099
chat box. I would think that if

703
00:19:40,100 --> 00:19:42,169
you are Google and it's integrated,

704
00:19:42,170 --> 00:19:43,879
it should do things in my document

705
00:19:43,880 --> 00:19:45,559
and not just produce me.

706
00:19:45,560 --> 00:19:47,329
Like all it is, is an integrated

707
00:19:47,330 --> 00:19:48,439
chat box.

708
00:19:48,440 --> 00:19:50,179
That's not the power of Google

709
00:19:50,180 --> 00:19:51,319
having it.

710
00:19:51,320 --> 00:19:53,339
They need to get to that next level.

711
00:19:53,340 --> 00:19:54,709
When you think about canvas within

712
00:19:54,710 --> 00:19:56,659
chat GPT, that is genius.

713
00:19:56,660 --> 00:19:58,189
That works so well in that respect.

714
00:19:58,190 --> 00:19:59,359
And you can imagine from a Google

715
00:19:59,360 --> 00:20:00,889
doc point of view, they should do

716
00:20:00,890 --> 00:20:02,179
exactly the same thing, and it

717
00:20:02,180 --> 00:20:04,129
should be way better given as Google

718
00:20:04,130 --> 00:20:06,049
Docs as opposed to some flimsy

719
00:20:06,050 --> 00:20:07,759
canvas thing that ChatGPT just

720
00:20:07,760 --> 00:20:08,839
created.

721
00:20:08,840 --> 00:20:09,949
So anyhow, that wasn't your

722
00:20:09,950 --> 00:20:10,939
question.

723
00:20:10,940 --> 00:20:12,829
So your question was what are

724
00:20:12,830 --> 00:20:15,229
the reasons why?

725
00:20:15,230 --> 00:20:16,729
What's the message to the company?

726
00:20:16,730 --> 00:20:18,259
You know, we need have this idea

727
00:20:18,260 --> 00:20:20,209
that for employees that

728
00:20:20,210 --> 00:20:21,919
this is the future, this is how we

729
00:20:21,920 --> 00:20:23,779
win. Part of that is to take

730
00:20:23,780 --> 00:20:25,709
A.I., truly embrace it,

731
00:20:25,710 --> 00:20:27,569
Activate ourselves to use it to help

732
00:20:27,570 --> 00:20:29,399
us with what we're doing to to get

733
00:20:29,400 --> 00:20:31,289
to the promised land effectively,

734
00:20:31,290 --> 00:20:33,749
and that this is a winning mindset

735
00:20:33,750 --> 00:20:35,789
and don't see it as cheating.

736
00:20:35,790 --> 00:20:37,769
We should be using it and

737
00:20:37,770 --> 00:20:39,119
it's very sensible.

738
00:20:39,120 --> 00:20:41,309
So we have for

739
00:20:41,310 --> 00:20:43,589
engineering, we've thought through

740
00:20:43,590 --> 00:20:45,599
the workflows and the tools to

741
00:20:45,600 --> 00:20:47,549
be used and have some

742
00:20:47,550 --> 00:20:49,499
guidance is quite specific.

743
00:20:49,500 --> 00:20:50,759
And then for the rest of the

744
00:20:50,760 --> 00:20:53,309
business, what we're doing is

745
00:20:53,310 --> 00:20:55,409
in every single weekly

746
00:20:55,410 --> 00:20:57,449
stand up, somebody does a show

747
00:20:57,450 --> 00:20:59,489
and tell on what they've done

748
00:20:59,490 --> 00:21:01,409
with Jenny that week

749
00:21:01,410 --> 00:21:02,729
and what they've learned and what

750
00:21:02,730 --> 00:21:03,989
worked and what didn't work.

751
00:21:03,990 --> 00:21:05,909
And so it can be really high

752
00:21:05,910 --> 00:21:07,439
level stuff for what marketing's

753
00:21:07,440 --> 00:21:09,329
doing. Or this week's

754
00:21:09,330 --> 00:21:11,699
was somebody in our R&D

755
00:21:11,700 --> 00:21:14,189
team going into

756
00:21:14,190 --> 00:21:16,769
huge amounts of detail with,

757
00:21:16,770 --> 00:21:17,819
I don't know, I think it was

758
00:21:17,820 --> 00:21:19,679
Claude's coding

759
00:21:19,680 --> 00:21:21,719
program and

760
00:21:21,720 --> 00:21:23,429
it's now becoming a genetic.

761
00:21:23,430 --> 00:21:25,109
And so like how it was solving the

762
00:21:25,110 --> 00:21:27,749
problem and working on

763
00:21:27,750 --> 00:21:29,279
multiple versions.

764
00:21:29,280 --> 00:21:31,199
And when it got itself into a death

765
00:21:31,200 --> 00:21:32,789
loop and how they got out of that

766
00:21:32,790 --> 00:21:35,369
death loop. So it was like super

767
00:21:35,370 --> 00:21:37,559
detailed and technical,

768
00:21:37,560 --> 00:21:39,959
but it wasn't a set training

769
00:21:39,960 --> 00:21:40,979
program.

770
00:21:40,980 --> 00:21:43,379
But we are giving people

771
00:21:43,380 --> 00:21:45,419
guidance and

772
00:21:45,420 --> 00:21:47,279
thinking through more than

773
00:21:47,280 --> 00:21:49,319
just a policy on how to use

774
00:21:49,320 --> 00:21:51,149
it and what

775
00:21:51,150 --> 00:21:52,379
to get out of it.

776
00:21:52,380 --> 00:21:53,729
And then the other thing is we do

777
00:21:53,730 --> 00:21:55,139
have budgets to experiment.

778
00:21:55,140 --> 00:21:56,969
So people are looking at windsurf

779
00:21:56,970 --> 00:21:59,669
and cursor and the ChatGPT

780
00:21:59,670 --> 00:22:02,009
and the Claude and

781
00:22:02,010 --> 00:22:03,839
Copilot,

782
00:22:03,840 --> 00:22:06,539
GitHub Copilot and

783
00:22:06,540 --> 00:22:08,519
Amazon Q so we're looking

784
00:22:08,520 --> 00:22:11,399
at like all of the different coding

785
00:22:11,400 --> 00:22:13,259
tools and not

786
00:22:13,260 --> 00:22:14,399
embedding them, but just

787
00:22:14,400 --> 00:22:16,829
experimenting and seeing

788
00:22:16,830 --> 00:22:18,239
which ones are good and which ones

789
00:22:18,240 --> 00:22:21,569
are shit. And just as another FYI,

790
00:22:21,570 --> 00:22:23,279
nobody seems to like copilot very

791
00:22:23,280 --> 00:22:25,259
much. It sounds like copilot

792
00:22:25,260 --> 00:22:27,149
might be a bit like Gemini,

793
00:22:27,150 --> 00:22:28,769
where it's just not there.

794
00:22:28,770 --> 00:22:30,449
So Charlie has services that he

795
00:22:30,450 --> 00:22:31,919
provides.

796
00:22:31,920 --> 00:22:34,259
The primary one is his AI

797
00:22:34,260 --> 00:22:36,119
Inspiration workshop, where he goes

798
00:22:36,120 --> 00:22:38,549
into an organization and

799
00:22:38,550 --> 00:22:40,529
gets them really not just excited by

800
00:22:40,530 --> 00:22:42,239
the possibilities around Cod and

801
00:22:42,240 --> 00:22:44,189
ChatGPT, but they kind

802
00:22:44,190 --> 00:22:45,659
of work hands on.

803
00:22:45,660 --> 00:22:47,009
What kind of use cases might be

804
00:22:47,010 --> 00:22:48,329
interesting for that particular

805
00:22:48,330 --> 00:22:49,769
group? And I think he referenced

806
00:22:49,770 --> 00:22:51,689
working in a organization for this

807
00:22:51,690 --> 00:22:53,759
workshop with the finance team

808
00:22:53,760 --> 00:22:55,139
and with that finance team.

809
00:22:55,140 --> 00:22:57,059
They outline all sorts of possible

810
00:22:57,060 --> 00:22:58,889
ways to use it within finance.

811
00:22:58,890 --> 00:23:00,299
That's part of that workshop where

812
00:23:00,300 --> 00:23:01,829
they got those finance individuals

813
00:23:01,830 --> 00:23:03,659
on their pathway to take

814
00:23:03,660 --> 00:23:05,309
those projects and start building

815
00:23:05,310 --> 00:23:07,289
them out, basically with projects

816
00:23:07,290 --> 00:23:09,239
or custom chat gifts or what have

817
00:23:09,240 --> 00:23:10,199
you, I suppose.

818
00:23:10,200 --> 00:23:12,119
So I was kind of wondering, what

819
00:23:12,120 --> 00:23:13,469
do you make of that activation of

820
00:23:13,470 --> 00:23:15,779
people getting things to happen,

821
00:23:15,780 --> 00:23:17,219
workshops like this?

822
00:23:17,220 --> 00:23:18,179
Yeah, I think it's a really good

823
00:23:18,180 --> 00:23:20,159
idea. And also like

824
00:23:20,160 --> 00:23:21,839
finance might be a bit risk averse

825
00:23:21,840 --> 00:23:23,099
and not wanting to change, but they

826
00:23:23,100 --> 00:23:24,509
have some of the most boring,

827
00:23:24,510 --> 00:23:26,939
tedious, repetitive jobs out there.

828
00:23:26,940 --> 00:23:28,349
If I were the person who had to,

829
00:23:28,350 --> 00:23:30,449
like, handle all

830
00:23:30,450 --> 00:23:33,059
of the expenses

831
00:23:33,060 --> 00:23:35,639
and accounts, I would be crying

832
00:23:35,640 --> 00:23:37,619
out and experimenting to see how

833
00:23:37,620 --> 00:23:39,359
I can make my job less boring and

834
00:23:39,360 --> 00:23:40,949
let the machines do all of the shit

835
00:23:40,950 --> 00:23:42,719
work. And I think that that's kind

836
00:23:42,720 --> 00:23:44,669
of the inspiration is

837
00:23:44,670 --> 00:23:46,559
what are the things

838
00:23:46,560 --> 00:23:49,229
that are mind numbingly boring

839
00:23:49,230 --> 00:23:51,959
that you have to do all the time,

840
00:23:51,960 --> 00:23:53,879
and then it's worth taking the

841
00:23:53,880 --> 00:23:55,619
time to create the prompts and to

842
00:23:55,620 --> 00:23:57,479
create the agent that just does

843
00:23:57,480 --> 00:23:59,309
the shit work for you.

844
00:23:59,310 --> 00:24:00,449
And that's kind of like my

845
00:24:00,450 --> 00:24:01,949
inspiration, you know, it's like all

846
00:24:01,950 --> 00:24:04,049
of those back office roles

847
00:24:04,050 --> 00:24:05,249
that are horrible.

848
00:24:05,250 --> 00:24:06,659
They're the first place to look

849
00:24:06,660 --> 00:24:08,099
because you just have happier people

850
00:24:08,100 --> 00:24:10,169
who aren't just doing

851
00:24:10,170 --> 00:24:12,209
awful stuff, and they

852
00:24:12,210 --> 00:24:13,199
don't have to do awful stuff the

853
00:24:13,200 --> 00:24:15,089
whole time. But like months end,

854
00:24:15,090 --> 00:24:16,619
nobody looks forward to month end in

855
00:24:16,620 --> 00:24:17,699
finance.

856
00:24:17,700 --> 00:24:20,019
Yeah, it's a lot of pressure.

857
00:24:20,020 --> 00:24:21,489
A lot of pressure and a lot of

858
00:24:21,490 --> 00:24:22,989
boring shit.

859
00:24:22,990 --> 00:24:24,849
So I will report on this at

860
00:24:24,850 --> 00:24:26,019
some point in the future, but I'm

861
00:24:26,020 --> 00:24:27,699
very keen in an organization I'm

862
00:24:27,700 --> 00:24:29,889
currently working with to take

863
00:24:29,890 --> 00:24:31,749
all the policies and to

864
00:24:31,750 --> 00:24:33,219
stick them into notebook alarm

865
00:24:33,220 --> 00:24:35,319
specifically. So this company uses

866
00:24:35,320 --> 00:24:36,189
the Google Suite.

867
00:24:36,190 --> 00:24:38,229
So notebook alum, you can share

868
00:24:38,230 --> 00:24:39,159
your notebooks across your

869
00:24:39,160 --> 00:24:40,479
organization seamlessly.

870
00:24:40,480 --> 00:24:42,369
And notebook alum seems very well

871
00:24:42,370 --> 00:24:44,409
suited for this task which is

872
00:24:44,410 --> 00:24:46,839
taking documents inserting them,

873
00:24:46,840 --> 00:24:48,549
not using the extended world of

874
00:24:48,550 --> 00:24:50,379
alums, whereby gets confused between

875
00:24:50,380 --> 00:24:51,969
your documentation and the greater

876
00:24:51,970 --> 00:24:53,289
world of documentation around those

877
00:24:53,290 --> 00:24:55,119
policies, and

878
00:24:55,120 --> 00:24:57,339
also with that podcast

879
00:24:57,340 --> 00:24:58,479
interface and different ways to

880
00:24:58,480 --> 00:24:59,769
access the information.

881
00:24:59,770 --> 00:25:01,269
And it seems so much more obvious.

882
00:25:01,270 --> 00:25:02,439
I wouldn't say fun, but just like

883
00:25:02,440 --> 00:25:04,329
different ways to like allow

884
00:25:04,330 --> 00:25:06,639
the employee to get information

885
00:25:06,640 --> 00:25:08,889
in a fast, efficient, timely basis.

886
00:25:08,890 --> 00:25:10,509
So I think that as an actual kind of

887
00:25:10,510 --> 00:25:12,729
low hanging fruit as a CEO,

888
00:25:12,730 --> 00:25:14,619
that seems like just a fun

889
00:25:14,620 --> 00:25:16,359
thing to start with.

890
00:25:16,360 --> 00:25:17,469
Yeah, I agree.

891
00:25:17,470 --> 00:25:18,819
Let me know how it goes.

892
00:25:18,820 --> 00:25:20,019
So let's park it here and let's get

893
00:25:20,020 --> 00:25:21,069
on to our conversation.

894
00:25:21,070 --> 00:25:22,749
Part two with Mr. Charlie Callum.

895
00:25:26,710 --> 00:25:28,719
The more that I learn about AI,

896
00:25:28,720 --> 00:25:30,579
the more I realize you don't need

897
00:25:30,580 --> 00:25:32,559
to know anything about AI to

898
00:25:32,560 --> 00:25:34,449
embed AI and benefit

899
00:25:34,450 --> 00:25:36,429
from AI, and you

900
00:25:36,430 --> 00:25:38,379
need to know about neural networks

901
00:25:38,380 --> 00:25:39,789
to benefit from AI.

902
00:25:39,790 --> 00:25:41,499
In the same way, you need to know

903
00:25:41,500 --> 00:25:43,599
about nuclear fission to benefit

904
00:25:43,600 --> 00:25:45,069
from electricity.

905
00:25:45,070 --> 00:25:47,079
You don't. You just need to plug in

906
00:25:47,080 --> 00:25:48,939
to the wall and consume the

907
00:25:48,940 --> 00:25:50,079
electricity.

908
00:25:50,080 --> 00:25:52,119
How it gets to your plug

909
00:25:52,120 --> 00:25:54,309
socket is really I have no interest

910
00:25:54,310 --> 00:25:56,289
to you, and we

911
00:25:56,290 --> 00:25:59,019
very much seen this with AI

912
00:25:59,020 --> 00:26:00,879
over the last, you know, sort of 10

913
00:26:00,880 --> 00:26:02,709
or 20 years where it's been, oh,

914
00:26:02,710 --> 00:26:04,209
you need to be a data scientist.

915
00:26:04,210 --> 00:26:05,499
You need to have a machine learning

916
00:26:05,500 --> 00:26:07,449
degree to be able to understand

917
00:26:07,450 --> 00:26:08,859
what's going on.

918
00:26:08,860 --> 00:26:10,899
But November 22nd,

919
00:26:10,900 --> 00:26:12,939
with the arrival of ChatGPT,

920
00:26:12,940 --> 00:26:14,049
that is the plug socket.

921
00:26:14,050 --> 00:26:16,059
The plug socket is the chat input

922
00:26:16,060 --> 00:26:16,959
window.

923
00:26:16,960 --> 00:26:18,879
That is all you need to know what

924
00:26:18,880 --> 00:26:19,929
happens behind it.

925
00:26:19,930 --> 00:26:21,279
Neural networks.

926
00:26:21,280 --> 00:26:22,659
Reinforcement learning.

927
00:26:22,660 --> 00:26:24,129
You don't need to know that to

928
00:26:24,130 --> 00:26:25,389
benefit from it.

929
00:26:25,390 --> 00:26:28,689
We are all

930
00:26:28,690 --> 00:26:30,789
CEOs of organizations

931
00:26:30,790 --> 00:26:32,379
that don't want to be left behind.

932
00:26:32,380 --> 00:26:35,049
We understand the importance of AI.

933
00:26:35,050 --> 00:26:36,579
How do we get started?

934
00:26:36,580 --> 00:26:38,199
There's a few things that I would

935
00:26:38,200 --> 00:26:40,149
think about as a sort

936
00:26:40,150 --> 00:26:41,709
of setting the scenes as an

937
00:26:41,710 --> 00:26:43,659
executive, as a CEO, as part of that

938
00:26:43,660 --> 00:26:44,679
leadership team.

939
00:26:44,680 --> 00:26:46,779
And that is firstly, just

940
00:26:46,780 --> 00:26:48,639
understanding what is our

941
00:26:48,640 --> 00:26:50,289
approach here.

942
00:26:50,290 --> 00:26:52,569
You can be an AI

943
00:26:52,570 --> 00:26:54,939
pessimist, which is where

944
00:26:54,940 --> 00:26:56,349
you're like, oh, you know, I'm not

945
00:26:56,350 --> 00:26:57,609
really sure about this thing.

946
00:26:57,610 --> 00:26:59,019
And if we are going to use it, I'm

947
00:26:59,020 --> 00:27:00,609
going to be thinking about how do we

948
00:27:00,610 --> 00:27:02,529
strip out costs from

949
00:27:02,530 --> 00:27:03,849
our business.

950
00:27:03,850 --> 00:27:05,919
How do we become more

951
00:27:05,920 --> 00:27:07,569
efficient and do things with less

952
00:27:07,570 --> 00:27:08,379
people?

953
00:27:08,380 --> 00:27:09,969
I think of this if you're a company

954
00:27:09,970 --> 00:27:11,949
of a thousand people, well,

955
00:27:11,950 --> 00:27:13,809
we can get to the same destination

956
00:27:13,810 --> 00:27:15,229
now with only 500.

957
00:27:15,230 --> 00:27:17,419
So that's the pessimistic view.

958
00:27:17,420 --> 00:27:19,489
The alternative is to be an

959
00:27:19,490 --> 00:27:21,259
AI optimist and go right, we're

960
00:27:21,260 --> 00:27:23,149
going to lean into this and

961
00:27:23,150 --> 00:27:25,109
we're a company of what do I say, a

962
00:27:25,110 --> 00:27:26,509
thousand people and we're now over

963
00:27:26,510 --> 00:27:27,949
500 people. We're not going to act

964
00:27:27,950 --> 00:27:30,229
like a company of 5000 people.

965
00:27:30,230 --> 00:27:32,359
We're going to be able to go further

966
00:27:32,360 --> 00:27:35,029
with the same amount of resource.

967
00:27:35,030 --> 00:27:36,739
And that kind of, are we leaning

968
00:27:36,740 --> 00:27:38,989
into this thing or are we leaning

969
00:27:38,990 --> 00:27:40,999
out of it is a real

970
00:27:41,000 --> 00:27:42,469
precursor to everything that you're

971
00:27:42,470 --> 00:27:44,269
going to do afterwards.

972
00:27:44,270 --> 00:27:46,099
So having defined right, we're going

973
00:27:46,100 --> 00:27:47,419
to lean into this and we're going to

974
00:27:47,420 --> 00:27:48,979
figure out how do we act like a much

975
00:27:48,980 --> 00:27:50,899
bigger company by using

976
00:27:50,900 --> 00:27:52,129
these tools.

977
00:27:52,130 --> 00:27:53,599
Next thing you probably want to do

978
00:27:53,600 --> 00:27:55,729
is very quickly codify

979
00:27:55,730 --> 00:27:57,559
that into some form of

980
00:27:57,560 --> 00:27:59,209
AI policy.

981
00:27:59,210 --> 00:28:00,859
Now I'm very nervous about

982
00:28:00,860 --> 00:28:02,809
mentioning AI policies

983
00:28:02,810 --> 00:28:04,339
because it's like, oh my goodness.

984
00:28:04,340 --> 00:28:05,929
Like, this is just like, you know,

985
00:28:05,930 --> 00:28:06,979
who's going to read it?

986
00:28:06,980 --> 00:28:08,329
What's the point.

987
00:28:08,330 --> 00:28:10,519
Where CEOs, we love a good policy.

988
00:28:10,520 --> 00:28:12,049
You don't need to say that policies

989
00:28:12,050 --> 00:28:13,279
are a problem here.

990
00:28:13,280 --> 00:28:15,529
One of the common things that we see

991
00:28:15,530 --> 00:28:18,199
is that if you ask companies,

992
00:28:18,200 --> 00:28:19,039
you know, what are you doing with

993
00:28:19,040 --> 00:28:19,819
AI?

994
00:28:19,820 --> 00:28:20,989
You know, what benefits are you

995
00:28:20,990 --> 00:28:23,209
seeing? You often see quite

996
00:28:23,210 --> 00:28:25,099
muted responses.

997
00:28:25,100 --> 00:28:26,119
We're not really seeing it.

998
00:28:26,120 --> 00:28:27,469
We've got some pieces.

999
00:28:27,470 --> 00:28:28,549
There's nothing really that's out

1000
00:28:28,550 --> 00:28:30,079
there in production.

1001
00:28:30,080 --> 00:28:32,329
If you go and speak to the people

1002
00:28:32,330 --> 00:28:34,609
one by one, are you using

1003
00:28:34,610 --> 00:28:36,739
ChatGPT every day?

1004
00:28:36,740 --> 00:28:38,179
Every day? I'm using it all the

1005
00:28:38,180 --> 00:28:39,979
time. You know, it's where I go to

1006
00:28:39,980 --> 00:28:42,079
ask any task that I've got.

1007
00:28:42,080 --> 00:28:43,969
So where is this discrepancy

1008
00:28:43,970 --> 00:28:44,809
coming from?

1009
00:28:44,810 --> 00:28:46,489
People are saying they're using it

1010
00:28:46,490 --> 00:28:48,349
all the time, and yet companies

1011
00:28:48,350 --> 00:28:50,479
are saying we're not really using it

1012
00:28:50,480 --> 00:28:52,399
after this because people are not

1013
00:28:52,400 --> 00:28:54,259
telling their manager, their

1014
00:28:54,260 --> 00:28:56,359
leaders, their colleagues when

1015
00:28:56,360 --> 00:28:57,889
they're using it. It's kind of under

1016
00:28:57,890 --> 00:28:59,389
the radar a little bit.

1017
00:28:59,390 --> 00:29:00,829
You know, oh, I'm just going to ask

1018
00:29:00,830 --> 00:29:02,569
ChatGPT or Claude.

1019
00:29:02,570 --> 00:29:04,939
It might be seen as cheating.

1020
00:29:04,940 --> 00:29:07,039
And often this is because certainly

1021
00:29:07,040 --> 00:29:08,899
when I'm speaking to teams, I'm

1022
00:29:08,900 --> 00:29:10,369
not sure what the rule is.

1023
00:29:10,370 --> 00:29:12,199
I'm not sure whether I'm allowed

1024
00:29:12,200 --> 00:29:13,849
to use it, so I'm definitely doing

1025
00:29:13,850 --> 00:29:15,289
it because I know that it gets me

1026
00:29:15,290 --> 00:29:17,239
where I want to go faster, but

1027
00:29:17,240 --> 00:29:19,339
I'm not being very public about it.

1028
00:29:19,340 --> 00:29:21,109
So one of the things that you can do

1029
00:29:21,110 --> 00:29:23,359
with creating a short,

1030
00:29:23,360 --> 00:29:25,459
pragmatic, positive

1031
00:29:25,460 --> 00:29:27,679
AI policy is

1032
00:29:27,680 --> 00:29:29,179
we are leading into AI.

1033
00:29:29,180 --> 00:29:30,859
We encourage your use.

1034
00:29:30,860 --> 00:29:32,599
We encourage you to experiment with

1035
00:29:32,600 --> 00:29:34,429
new tools, but be

1036
00:29:34,430 --> 00:29:36,649
sensible. Here are some guardrails

1037
00:29:36,650 --> 00:29:38,659
that you might want to follow so

1038
00:29:38,660 --> 00:29:40,129
that you know we're not sharing

1039
00:29:40,130 --> 00:29:42,049
confidential data or so on,

1040
00:29:42,050 --> 00:29:44,539
but having that policy where we're

1041
00:29:44,540 --> 00:29:46,099
setting what's right and what's

1042
00:29:46,100 --> 00:29:47,929
wrong, but reinforcing to everyone

1043
00:29:47,930 --> 00:29:49,609
we are leaning into this and we want

1044
00:29:49,610 --> 00:29:51,469
you to lean into it is

1045
00:29:51,470 --> 00:29:53,269
it's a really good approach.

1046
00:29:53,270 --> 00:29:54,289
And I can't remember if I asked this

1047
00:29:54,290 --> 00:29:55,609
question before, but it's something

1048
00:29:55,610 --> 00:29:56,869
I really want to know about.

1049
00:29:56,870 --> 00:29:58,489
And it's probably like the next

1050
00:29:58,490 --> 00:30:00,499
layer of depth rather than like,

1051
00:30:00,500 --> 00:30:03,349
should you have a AI policy?

1052
00:30:03,350 --> 00:30:05,229
But being in tech, everybody says

1053
00:30:05,230 --> 00:30:07,069
that 2025 is the year of

1054
00:30:07,070 --> 00:30:08,209
a genetic.

1055
00:30:08,210 --> 00:30:10,069
It's all about a genetic AI.

1056
00:30:10,070 --> 00:30:11,719
And yet we mostly just talk about

1057
00:30:11,720 --> 00:30:14,209
the chat bot and using ChatGPT

1058
00:30:14,210 --> 00:30:15,529
to do a bit of thinking for us,

1059
00:30:15,530 --> 00:30:17,449
rather than to actually start to

1060
00:30:17,450 --> 00:30:19,279
automate our lives and be

1061
00:30:19,280 --> 00:30:20,539
proper agents.

1062
00:30:20,540 --> 00:30:21,919
I really want to understand about a

1063
00:30:21,920 --> 00:30:23,899
genetic, because it's

1064
00:30:23,900 --> 00:30:25,879
like the year of people creating

1065
00:30:25,880 --> 00:30:27,889
agents to sell to other people,

1066
00:30:27,890 --> 00:30:29,569
but if you're in a business and

1067
00:30:29,570 --> 00:30:31,669
don't want to buy 500 million

1068
00:30:31,670 --> 00:30:34,159
different vertical agents,

1069
00:30:34,160 --> 00:30:35,959
how do you start to actually

1070
00:30:35,960 --> 00:30:37,669
automate with agents and are there

1071
00:30:37,670 --> 00:30:39,499
any good platforms to do that?

1072
00:30:39,500 --> 00:30:40,969
Agent says the password.

1073
00:30:40,970 --> 00:30:42,949
So first let's define what

1074
00:30:42,950 --> 00:30:44,449
an agent is.

1075
00:30:44,450 --> 00:30:46,009
And as everyone's talking about it,

1076
00:30:46,010 --> 00:30:47,989
suddenly it becomes just like this

1077
00:30:47,990 --> 00:30:50,089
fake. No one's quite sure.

1078
00:30:50,090 --> 00:30:52,519
For an agent to have agency,

1079
00:30:52,520 --> 00:30:55,219
it has to have the ability to choose

1080
00:30:55,220 --> 00:30:57,229
what tools, if any,

1081
00:30:57,230 --> 00:30:59,149
it is going to use to complete

1082
00:30:59,150 --> 00:31:00,169
the task.

1083
00:31:00,170 --> 00:31:01,849
The opposite of that is what we

1084
00:31:01,850 --> 00:31:03,709
might think of a workflow

1085
00:31:03,710 --> 00:31:05,149
or an automation.

1086
00:31:05,150 --> 00:31:06,619
So if people are building things

1087
00:31:06,620 --> 00:31:07,649
with it's happier if they're

1088
00:31:07,650 --> 00:31:09,179
building things would make if they

1089
00:31:09,180 --> 00:31:11,039
built integrations with Boomi

1090
00:31:11,040 --> 00:31:12,479
or Informatica.

1091
00:31:12,480 --> 00:31:14,759
If those things are predetermined,

1092
00:31:14,760 --> 00:31:16,859
even if it is an alarm call

1093
00:31:16,860 --> 00:31:18,899
and then another alarm call

1094
00:31:18,900 --> 00:31:20,459
that is still more of a workflow or

1095
00:31:20,460 --> 00:31:22,529
an automation because whatever

1096
00:31:22,530 --> 00:31:24,869
happens has been predetermined.

1097
00:31:24,870 --> 00:31:26,849
The true agent is

1098
00:31:26,850 --> 00:31:28,709
that I've got this alarm

1099
00:31:28,710 --> 00:31:30,539
tool and it is able to

1100
00:31:30,540 --> 00:31:32,369
decide now I'm going to

1101
00:31:32,370 --> 00:31:34,379
go and check a stock price.

1102
00:31:34,380 --> 00:31:36,359
Now I'm going to create a document.

1103
00:31:36,360 --> 00:31:37,919
Now I'm going to go and ask, you

1104
00:31:37,920 --> 00:31:39,419
know, another alarm for for

1105
00:31:39,420 --> 00:31:40,529
something else.

1106
00:31:40,530 --> 00:31:42,359
And I'm going to decide because

1107
00:31:42,360 --> 00:31:44,789
I've got the agency about

1108
00:31:44,790 --> 00:31:47,039
what to use and when.

1109
00:31:47,040 --> 00:31:49,379
And so for any company that is

1110
00:31:49,380 --> 00:31:51,539
thinking of building agents

1111
00:31:51,540 --> 00:31:53,969
or is being sold an agent

1112
00:31:53,970 --> 00:31:55,379
by someone else, that's the

1113
00:31:55,380 --> 00:31:56,999
important thing to dig in.

1114
00:31:57,000 --> 00:31:58,889
Have we really got an agent which

1115
00:31:58,890 --> 00:32:00,779
is able to make decisions, or

1116
00:32:00,780 --> 00:32:02,429
we just got a workflow?

1117
00:32:02,430 --> 00:32:04,259
And why that's important is that

1118
00:32:04,260 --> 00:32:06,209
if we're saying 2025 is the year

1119
00:32:06,210 --> 00:32:07,499
of the agents.

1120
00:32:07,500 --> 00:32:08,459
What does that mean?

1121
00:32:08,460 --> 00:32:11,009
We start to think about true

1122
00:32:11,010 --> 00:32:12,899
digital workers,

1123
00:32:12,900 --> 00:32:14,849
digital employees that

1124
00:32:14,850 --> 00:32:16,289
can make decisions.

1125
00:32:16,290 --> 00:32:18,119
So if we I don't

1126
00:32:18,120 --> 00:32:19,979
know if someone working in HR as

1127
00:32:19,980 --> 00:32:22,079
an HR administrator,

1128
00:32:22,080 --> 00:32:24,359
I'm able to as a digital

1129
00:32:24,360 --> 00:32:26,609
agent, I'm receiving a

1130
00:32:26,610 --> 00:32:28,649
request or a job posting

1131
00:32:28,650 --> 00:32:30,209
request from a manager.

1132
00:32:30,210 --> 00:32:32,309
I'm then able to go and

1133
00:32:32,310 --> 00:32:34,049
use another tool to go and write up

1134
00:32:34,050 --> 00:32:35,759
the job description.

1135
00:32:35,760 --> 00:32:37,619
Then able to go and decide if

1136
00:32:37,620 --> 00:32:38,879
we're going to have that budget and

1137
00:32:38,880 --> 00:32:40,409
when we're going to post it, I'm

1138
00:32:40,410 --> 00:32:42,539
then going to choose which

1139
00:32:42,540 --> 00:32:43,859
job boards we're going to post it

1140
00:32:43,860 --> 00:32:45,749
on. Then maybe if and I'm going to

1141
00:32:45,750 --> 00:32:47,669
choose which candidate I'm

1142
00:32:47,670 --> 00:32:49,559
going to shortlist and get invited

1143
00:32:49,560 --> 00:32:51,389
in for an invite, that would

1144
00:32:51,390 --> 00:32:53,759
be one or a series of agents

1145
00:32:53,760 --> 00:32:55,709
that are making decisions as

1146
00:32:55,710 --> 00:32:57,959
opposed to a true workflow.

1147
00:32:57,960 --> 00:32:59,729
But that's when you're starting to

1148
00:32:59,730 --> 00:33:01,919
really replace humans.

1149
00:33:01,920 --> 00:33:03,899
So that's a great definition

1150
00:33:03,900 --> 00:33:05,219
of agents.

1151
00:33:05,220 --> 00:33:06,539
This year is supposed to be the year

1152
00:33:06,540 --> 00:33:07,679
of a genetic AI.

1153
00:33:07,680 --> 00:33:09,299
By the way, genetic is a word that's

1154
00:33:09,300 --> 00:33:10,559
been basically made up.

1155
00:33:10,560 --> 00:33:12,659
That means agent, IC

1156
00:33:12,660 --> 00:33:14,819
agent. Like now everybody

1157
00:33:14,820 --> 00:33:16,529
in here, like I think I saw a

1158
00:33:16,530 --> 00:33:17,639
genetic for the first time about

1159
00:33:17,640 --> 00:33:18,659
nine months ago. And I was like,

1160
00:33:18,660 --> 00:33:20,159
what is this word? Is it a typo?

1161
00:33:20,160 --> 00:33:21,839
And now, like my entire life is

1162
00:33:21,840 --> 00:33:22,829
about a genetic.

1163
00:33:22,830 --> 00:33:25,109
It sounds great, I want some

1164
00:33:25,110 --> 00:33:26,399
if this is the year of a genetic, I

1165
00:33:26,400 --> 00:33:28,169
said, this is a year that all

1166
00:33:28,170 --> 00:33:30,539
software companies create agents

1167
00:33:30,540 --> 00:33:32,339
that other people can buy.

1168
00:33:32,340 --> 00:33:34,289
Or is this the year that I can

1169
00:33:34,290 --> 00:33:36,149
suddenly have a million agents in

1170
00:33:36,150 --> 00:33:38,099
my business by

1171
00:33:38,100 --> 00:33:39,839
using ChatGPT?

1172
00:33:39,840 --> 00:33:41,669
I'll answer that by talking

1173
00:33:41,670 --> 00:33:43,499
to maybe some of the steps that

1174
00:33:43,500 --> 00:33:45,419
a CEO or a company might

1175
00:33:45,420 --> 00:33:47,339
want to go through when

1176
00:33:47,340 --> 00:33:48,659
embarking on trying to make that

1177
00:33:48,660 --> 00:33:51,089
decision, both on anthropic

1178
00:33:52,230 --> 00:33:54,989
docs, website documentation,

1179
00:33:54,990 --> 00:33:57,419
and also on eyes.

1180
00:33:57,420 --> 00:33:59,249
They talk about passing the the

1181
00:33:59,250 --> 00:34:01,439
intern test, which is

1182
00:34:01,440 --> 00:34:03,449
you should think of I as

1183
00:34:03,450 --> 00:34:06,119
a brand new, eager,

1184
00:34:06,120 --> 00:34:08,488
but poorly informed intern.

1185
00:34:08,489 --> 00:34:10,829
And if you provide that intern

1186
00:34:10,830 --> 00:34:13,229
with a badly defined

1187
00:34:13,230 --> 00:34:15,178
instructions, you will get

1188
00:34:15,179 --> 00:34:17,638
what you deserve in return.

1189
00:34:17,639 --> 00:34:19,559
The original request

1190
00:34:19,560 --> 00:34:21,779
is quite poorly defined,

1191
00:34:21,780 --> 00:34:23,879
and that's often because

1192
00:34:23,880 --> 00:34:25,198
the person that's doing the job

1193
00:34:25,199 --> 00:34:27,178
today has been doing it for a while.

1194
00:34:27,179 --> 00:34:28,198
They kind of know it.

1195
00:34:28,199 --> 00:34:30,448
There's lots of sort of assumptions

1196
00:34:30,449 --> 00:34:31,619
or well, I'm sure you know what that

1197
00:34:31,620 --> 00:34:32,698
means.

1198
00:34:32,699 --> 00:34:35,069
And this is often why

1199
00:34:35,070 --> 00:34:36,749
you see in companies that when

1200
00:34:36,750 --> 00:34:38,669
someone asks, you know, how do I

1201
00:34:38,670 --> 00:34:40,109
do this? Or, you know, need and you

1202
00:34:40,110 --> 00:34:40,979
can speak to Gary.

1203
00:34:40,980 --> 00:34:42,448
He's been here for 20 years and he

1204
00:34:42,449 --> 00:34:44,249
knows how that works.

1205
00:34:44,250 --> 00:34:46,319
You can't take what's in Gary's

1206
00:34:46,320 --> 00:34:48,689
head and just give that to

1207
00:34:48,690 --> 00:34:50,579
the LM, because there's so much

1208
00:34:50,580 --> 00:34:52,439
of Gary's experience and knowledge

1209
00:34:52,440 --> 00:34:53,669
that's wrapped up in there.

1210
00:34:53,670 --> 00:34:56,009
So I find myself writing

1211
00:34:56,010 --> 00:34:58,889
a lot of long, prompt documents

1212
00:34:58,890 --> 00:35:00,579
that are maybe anything from 10 to

1213
00:35:00,580 --> 00:35:02,709
20 pages long, which is first

1214
00:35:02,710 --> 00:35:04,629
we do this, then we do that.

1215
00:35:04,630 --> 00:35:06,099
Then we do that.

1216
00:35:06,100 --> 00:35:08,079
This is what that means

1217
00:35:08,080 --> 00:35:09,639
to pass the internal test.

1218
00:35:09,640 --> 00:35:11,679
If I gave that Google doc to

1219
00:35:11,680 --> 00:35:13,659
an intern, could they do

1220
00:35:13,660 --> 00:35:15,249
what I'm asked them to do and I

1221
00:35:15,250 --> 00:35:17,019
would get the output?

1222
00:35:17,020 --> 00:35:19,419
And with anthropic and OpenAI,

1223
00:35:19,420 --> 00:35:21,489
when they say pass the intern test.

1224
00:35:21,490 --> 00:35:23,379
If the answer to that is no, that

1225
00:35:23,380 --> 00:35:24,849
if I gave that document to an

1226
00:35:24,850 --> 00:35:26,799
intern, would I get the response

1227
00:35:26,800 --> 00:35:27,759
I wanted?

1228
00:35:27,760 --> 00:35:29,919
Then how is an El Alam going

1229
00:35:29,920 --> 00:35:31,029
to do that?

1230
00:35:31,030 --> 00:35:32,319
It's not.

1231
00:35:32,320 --> 00:35:34,419
So I'd say a large proportion

1232
00:35:34,420 --> 00:35:36,069
of the time that I spend is in

1233
00:35:36,070 --> 00:35:39,009
getting the ask correct.

1234
00:35:39,010 --> 00:35:41,049
Once you've got the ask correct,

1235
00:35:41,050 --> 00:35:42,879
then creating whether as a

1236
00:35:42,880 --> 00:35:44,859
custom GPT or whether it is

1237
00:35:44,860 --> 00:35:46,689
building out some form of agent that

1238
00:35:46,690 --> 00:35:48,699
is able to access different function

1239
00:35:48,700 --> 00:35:50,739
calls, that becomes relatively

1240
00:35:50,740 --> 00:35:52,329
easy, because what you've said to

1241
00:35:52,330 --> 00:35:54,639
the intern is, let's do that HR

1242
00:35:54,640 --> 00:35:56,169
hiring example, right.

1243
00:35:56,170 --> 00:35:57,069
The first thing that you're going to

1244
00:35:57,070 --> 00:35:58,359
do is you're going to be on the

1245
00:35:58,360 --> 00:36:00,729
lookout for job requests

1246
00:36:00,730 --> 00:36:03,729
that come from hiring managers.

1247
00:36:03,730 --> 00:36:04,809
The second thing that you're going

1248
00:36:04,810 --> 00:36:06,339
to do is you're going to take all

1249
00:36:06,340 --> 00:36:08,229
the details of that job request, and

1250
00:36:08,230 --> 00:36:09,519
then you're going to start drafting

1251
00:36:09,520 --> 00:36:11,619
up a more detailed job description.

1252
00:36:11,620 --> 00:36:13,389
This is what our job descriptions

1253
00:36:13,390 --> 00:36:14,919
should look like, and all of the

1254
00:36:14,920 --> 00:36:16,269
details of what should be in each

1255
00:36:16,270 --> 00:36:17,199
section.

1256
00:36:17,200 --> 00:36:19,029
And then you're going through all of

1257
00:36:19,030 --> 00:36:21,309
that. We use some examples

1258
00:36:21,310 --> 00:36:22,839
of how that H.R.

1259
00:36:22,840 --> 00:36:24,759
Hiring person would have

1260
00:36:24,760 --> 00:36:26,589
the choice of choosing which job

1261
00:36:26,590 --> 00:36:27,549
boards to post on.

1262
00:36:27,550 --> 00:36:29,319
Well, how do we make that decision

1263
00:36:29,320 --> 00:36:30,729
and how do we access those job

1264
00:36:30,730 --> 00:36:31,599
boards.

1265
00:36:31,600 --> 00:36:33,699
And that's really the instructions

1266
00:36:33,700 --> 00:36:35,079
that you're going to end up providing

1267
00:36:35,080 --> 00:36:36,309
to the AI.

1268
00:36:36,310 --> 00:36:37,809
Now, if you don't need to access any

1269
00:36:37,810 --> 00:36:39,459
tools because the human wouldn't

1270
00:36:39,460 --> 00:36:41,349
access any tools, then you're

1271
00:36:41,350 --> 00:36:43,539
thinking more of just a simpler

1272
00:36:43,540 --> 00:36:44,739
implementation.

1273
00:36:44,740 --> 00:36:46,119
If you are looking at accessing

1274
00:36:46,120 --> 00:36:48,279
tools or you need to access your HR

1275
00:36:48,280 --> 00:36:50,199
system or your payroll, then

1276
00:36:50,200 --> 00:36:51,519
you're thinking more of, I need to

1277
00:36:51,520 --> 00:36:52,929
create some kind of agent that can

1278
00:36:52,930 --> 00:36:55,539
choose when to use these tools.

1279
00:36:55,540 --> 00:36:57,579
And how do you create the agent?

1280
00:36:57,580 --> 00:36:59,079
That's my question.

1281
00:36:59,080 --> 00:37:01,089
My recommendation is a tool

1282
00:37:01,090 --> 00:37:03,069
called the vessel

1283
00:37:03,070 --> 00:37:04,719
AI SDK.

1284
00:37:04,720 --> 00:37:07,719
So vessel VR CEO

1285
00:37:07,720 --> 00:37:10,029
and vessel is a web

1286
00:37:10,030 --> 00:37:12,429
hosting and deployment company.

1287
00:37:12,430 --> 00:37:14,739
So anyone that's an AI product

1288
00:37:14,740 --> 00:37:16,989
management product development role

1289
00:37:16,990 --> 00:37:18,729
may have come across V0.

1290
00:37:18,730 --> 00:37:20,619
So that's a v not dev.

1291
00:37:20,620 --> 00:37:23,049
That also comes from vessel.

1292
00:37:23,050 --> 00:37:25,149
And that is a AI

1293
00:37:25,150 --> 00:37:26,499
generative UI.

1294
00:37:26,500 --> 00:37:29,799
So you can design your applications.

1295
00:37:29,800 --> 00:37:32,499
But there's vessel AI SDK

1296
00:37:32,500 --> 00:37:34,479
is a really neat way

1297
00:37:34,480 --> 00:37:36,519
of plugging in to open AI

1298
00:37:36,520 --> 00:37:38,439
and to anthropic and to Gemini

1299
00:37:38,440 --> 00:37:40,569
and to any of your preferred

1300
00:37:40,570 --> 00:37:42,399
LMS, and it

1301
00:37:42,400 --> 00:37:44,259
gives you the ability to create

1302
00:37:44,260 --> 00:37:46,659
these calls out, these tool calls,

1303
00:37:46,660 --> 00:37:49,209
which is what makes an agent.

1304
00:37:49,210 --> 00:37:51,189
Well, because I guess if I step

1305
00:37:51,190 --> 00:37:53,079
back to what I'm thinking about

1306
00:37:53,080 --> 00:37:55,089
and maybe it can help is in

1307
00:37:55,090 --> 00:37:57,249
the past, and particularly

1308
00:37:57,250 --> 00:37:59,109
as my experiences as Crow, which

1309
00:37:59,110 --> 00:38:00,609
I complain about bitterly.

1310
00:38:00,610 --> 00:38:02,649
Not the crow part, but the number

1311
00:38:02,650 --> 00:38:04,239
of tools I had to buy.

1312
00:38:04,240 --> 00:38:05,799
Because you had this tech stack, the

1313
00:38:05,800 --> 00:38:07,719
revenue tech stack and

1314
00:38:07,720 --> 00:38:09,579
everything costs like 20

1315
00:38:09,580 --> 00:38:11,469
K other than Salesforce, which

1316
00:38:11,470 --> 00:38:13,449
of course costs shitloads more.

1317
00:38:13,450 --> 00:38:14,979
But it'd be like, okay, now it's 20

1318
00:38:14,980 --> 00:38:16,869
K for this and 20 K for

1319
00:38:16,870 --> 00:38:18,429
this and 20 K for that, and you add

1320
00:38:18,430 --> 00:38:20,679
it all in, and then suddenly

1321
00:38:20,680 --> 00:38:22,569
you had a tech stack

1322
00:38:22,570 --> 00:38:23,889
of, I don't know, 20 different

1323
00:38:23,890 --> 00:38:25,719
things and half 1 million pounds.

1324
00:38:25,720 --> 00:38:28,359
What I don't want to do

1325
00:38:28,360 --> 00:38:30,429
when I'm looking at my

1326
00:38:30,430 --> 00:38:32,829
IT stack for the future,

1327
00:38:32,830 --> 00:38:35,139
is to have that 20,

1328
00:38:35,140 --> 00:38:37,149
30, 40 pieces

1329
00:38:37,150 --> 00:38:38,709
of tooling that are critical for the

1330
00:38:38,710 --> 00:38:39,909
business, and some of them are big

1331
00:38:39,910 --> 00:38:41,469
and some of them little, and they're

1332
00:38:41,470 --> 00:38:43,569
all whacking some AI on,

1333
00:38:43,570 --> 00:38:45,339
so they all have a chat bot.

1334
00:38:45,340 --> 00:38:46,929
Nobody seems to have a particularly

1335
00:38:46,930 --> 00:38:48,309
compelling vision.

1336
00:38:48,310 --> 00:38:50,139
I know that we are not all that

1337
00:38:50,140 --> 00:38:52,629
impressed by GitHub copilot,

1338
00:38:52,630 --> 00:38:54,469
you know, versus Verses like

1339
00:38:54,470 --> 00:38:56,449
cursor and some of the others.

1340
00:38:56,450 --> 00:38:58,279
It's I don't want to blindly

1341
00:38:58,280 --> 00:39:00,409
buy all of these tools

1342
00:39:00,410 --> 00:39:01,939
that say they have something, that

1343
00:39:01,940 --> 00:39:03,829
maybe have an agent, and

1344
00:39:03,830 --> 00:39:06,379
I still just have to pay shitloads.

1345
00:39:06,380 --> 00:39:08,209
What I would like to understand

1346
00:39:08,210 --> 00:39:11,149
is how to think about

1347
00:39:11,150 --> 00:39:13,189
what the future

1348
00:39:13,190 --> 00:39:15,319
tech stack should be for

1349
00:39:15,320 --> 00:39:17,119
a scale up. So a couple hundred

1350
00:39:17,120 --> 00:39:18,860
people to a thousand people.

1351
00:39:20,410 --> 00:39:22,069
Technology's moving really quickly.

1352
00:39:22,070 --> 00:39:24,289
This is the year of the agent.

1353
00:39:24,290 --> 00:39:27,079
What should I be buying today

1354
00:39:27,080 --> 00:39:28,939
or what should I be evaluating?

1355
00:39:28,940 --> 00:39:30,859
Because my ideal and my vision,

1356
00:39:30,860 --> 00:39:31,939
and I'm guessing that other people

1357
00:39:31,940 --> 00:39:33,529
would have this vision, is that

1358
00:39:33,530 --> 00:39:35,899
everybody has

1359
00:39:35,900 --> 00:39:38,329
their own assistant

1360
00:39:38,330 --> 00:39:40,189
who can do all of this shit work,

1361
00:39:40,190 --> 00:39:41,269
and it's really easy.

1362
00:39:41,270 --> 00:39:43,129
And you just say to your assistant,

1363
00:39:43,130 --> 00:39:44,159
can you look through all of these

1364
00:39:44,160 --> 00:39:45,979
CVS for me and stick

1365
00:39:45,980 --> 00:39:47,989
them and highlight the best

1366
00:39:47,990 --> 00:39:50,179
ones, or contact them or,

1367
00:39:50,180 --> 00:39:52,279
you know, and be able to write The

1368
00:39:52,280 --> 00:39:54,289
foreign intern prompt that's 20

1369
00:39:54,290 --> 00:39:55,099
pages long.

1370
00:39:55,100 --> 00:39:57,019
But once I do that, I never

1371
00:39:57,020 --> 00:39:58,819
have to do that work again.

1372
00:39:58,820 --> 00:40:00,769
And I want everybody in my company

1373
00:40:00,770 --> 00:40:02,359
to be able to do that, not just the

1374
00:40:02,360 --> 00:40:04,159
individual who understands how to

1375
00:40:04,160 --> 00:40:06,169
use VSL or

1376
00:40:06,170 --> 00:40:07,249
API's.

1377
00:40:07,250 --> 00:40:09,649
How far away are we from

1378
00:40:09,650 --> 00:40:11,749
that and

1379
00:40:11,750 --> 00:40:13,579
what are the steps to get there

1380
00:40:13,580 --> 00:40:14,839
this year?

1381
00:40:14,840 --> 00:40:17,059
It is such a

1382
00:40:17,060 --> 00:40:19,699
fast moving space

1383
00:40:19,700 --> 00:40:22,159
that it is just impossible

1384
00:40:22,160 --> 00:40:23,989
to say, you know, use

1385
00:40:23,990 --> 00:40:26,149
this tool or use that tool

1386
00:40:26,150 --> 00:40:27,979
because so much is coming out,

1387
00:40:27,980 --> 00:40:29,989
whether that's models or whether

1388
00:40:29,990 --> 00:40:31,639
it is tools that are set over the

1389
00:40:31,640 --> 00:40:33,679
top of these models that

1390
00:40:33,680 --> 00:40:36,379
help to create these workflows

1391
00:40:36,380 --> 00:40:38,269
is speaking to a CEO.

1392
00:40:38,270 --> 00:40:40,189
This would be my advice

1393
00:40:40,190 --> 00:40:42,229
is that there is never

1394
00:40:42,230 --> 00:40:45,199
been a better opportunity right now

1395
00:40:45,200 --> 00:40:47,359
to internalize

1396
00:40:47,360 --> 00:40:48,799
some of these skills.

1397
00:40:48,800 --> 00:40:50,479
Previously you might have said, you

1398
00:40:50,480 --> 00:40:52,159
know, build versus buy while the

1399
00:40:52,160 --> 00:40:54,169
building is just too complicated.

1400
00:40:54,170 --> 00:40:55,429
We're not going to go and set up

1401
00:40:55,430 --> 00:40:57,499
some massive infrastructure

1402
00:40:57,500 --> 00:40:59,509
and start investing in people to

1403
00:40:59,510 --> 00:41:01,339
run that. So we're going to buy

1404
00:41:01,340 --> 00:41:03,649
we'll buy a CRM, we'll buy

1405
00:41:03,650 --> 00:41:05,599
an air platform, whatever that might

1406
00:41:05,600 --> 00:41:07,909
be. But I think we're seeing now

1407
00:41:07,910 --> 00:41:09,919
that pioneering companies,

1408
00:41:09,920 --> 00:41:10,879
and that doesn't have to be just

1409
00:41:10,880 --> 00:41:12,229
like some funky startup.

1410
00:41:12,230 --> 00:41:13,429
But actually, I mean, you saw this

1411
00:41:13,430 --> 00:41:15,409
with Klarna recently

1412
00:41:15,410 --> 00:41:16,789
and saying, you know what with some

1413
00:41:16,790 --> 00:41:18,799
of these tools, whether that is

1414
00:41:18,800 --> 00:41:20,899
universal, whether that is V0,

1415
00:41:20,900 --> 00:41:23,059
whether it is a

1416
00:41:23,060 --> 00:41:24,889
new development platform

1417
00:41:24,890 --> 00:41:25,819
that people are going to, you know

1418
00:41:25,820 --> 00:41:27,799
what, I can take non

1419
00:41:27,800 --> 00:41:28,999
development people.

1420
00:41:29,000 --> 00:41:31,189
So a product manager or

1421
00:41:31,190 --> 00:41:33,049
someone in an operations team and

1422
00:41:33,050 --> 00:41:35,119
I can start building out proof

1423
00:41:35,120 --> 00:41:37,279
of concepts and building internal

1424
00:41:37,280 --> 00:41:39,259
products without

1425
00:41:39,260 --> 00:41:40,609
having to go and get a SAS

1426
00:41:40,610 --> 00:41:42,949
subscription from someone else.

1427
00:41:42,950 --> 00:41:44,809
And suddenly I'm relying on

1428
00:41:44,810 --> 00:41:46,669
their roadmap rather than

1429
00:41:46,670 --> 00:41:47,449
our own.

1430
00:41:47,450 --> 00:41:49,529
So I think there's going to be,

1431
00:41:49,530 --> 00:41:50,999
you know, some companies that just

1432
00:41:51,000 --> 00:41:52,619
don't get involved in this at all

1433
00:41:52,620 --> 00:41:54,539
and just stay. Let's leave AI

1434
00:41:54,540 --> 00:41:55,439
aside.

1435
00:41:55,440 --> 00:41:56,669
Then there are going to be some that

1436
00:41:56,670 --> 00:41:58,649
go. We want to buy stuff.

1437
00:41:58,650 --> 00:42:00,539
So let's rely on someone

1438
00:42:00,540 --> 00:42:02,549
else's a platform

1439
00:42:02,550 --> 00:42:04,499
for stars

1440
00:42:04,500 --> 00:42:07,169
or for marketing or whatever.

1441
00:42:07,170 --> 00:42:08,159
But I think there's going to be an

1442
00:42:08,160 --> 00:42:10,119
increasing number of companies said,

1443
00:42:10,120 --> 00:42:11,039
you know what?

1444
00:42:11,040 --> 00:42:12,899
We have got insight

1445
00:42:12,900 --> 00:42:15,059
into what are our challenges.

1446
00:42:15,060 --> 00:42:16,709
And we think it's quite unique to

1447
00:42:16,710 --> 00:42:18,569
us. And we're just going to

1448
00:42:18,570 --> 00:42:20,039
have a go at building something

1449
00:42:20,040 --> 00:42:21,989
internally using these

1450
00:42:21,990 --> 00:42:23,819
publicly available tools.

1451
00:42:23,820 --> 00:42:25,649
It's about using as

1452
00:42:25,650 --> 00:42:27,989
many tools as you can come across,

1453
00:42:27,990 --> 00:42:30,059
as they keep getting developed to

1454
00:42:30,060 --> 00:42:31,949
improve your learning and the way

1455
00:42:31,950 --> 00:42:33,119
of working.

1456
00:42:33,120 --> 00:42:35,159
This all feeds into that first

1457
00:42:35,160 --> 00:42:36,719
bit, which is about, you know, AI

1458
00:42:36,720 --> 00:42:38,849
positive or a negative

1459
00:42:38,850 --> 00:42:40,679
AI going to lean into AI

1460
00:42:40,680 --> 00:42:41,789
or lean out of it.

1461
00:42:41,790 --> 00:42:44,339
And again, checkers is important.

1462
00:42:44,340 --> 00:42:46,589
So having done that

1463
00:42:46,590 --> 00:42:48,569
then the primary

1464
00:42:48,570 --> 00:42:51,059
driver of a good change program,

1465
00:42:51,060 --> 00:42:52,559
and everyone will know this

1466
00:42:52,560 --> 00:42:53,849
regardless of what it is you're

1467
00:42:53,850 --> 00:42:55,889
trying to change, is having

1468
00:42:55,890 --> 00:42:57,929
empowered and inspired

1469
00:42:57,930 --> 00:42:59,759
senior leaders that are at

1470
00:42:59,760 --> 00:43:01,349
the top of that program.

1471
00:43:01,350 --> 00:43:03,269
So I would then be looking at

1472
00:43:03,270 --> 00:43:05,489
who's the rest of your executive

1473
00:43:05,490 --> 00:43:06,659
leadership team. So if you are the

1474
00:43:06,660 --> 00:43:08,489
CEO, I'd be looking left at the

1475
00:43:08,490 --> 00:43:10,409
CFO. I'd be looking at that chief

1476
00:43:10,410 --> 00:43:11,219
Revenue Officer.

1477
00:43:11,220 --> 00:43:12,269
I've been looking at the Chief

1478
00:43:12,270 --> 00:43:14,309
People officer and a level

1479
00:43:14,310 --> 00:43:16,439
down into those VP's

1480
00:43:16,440 --> 00:43:18,329
and figuring out quickly,

1481
00:43:18,330 --> 00:43:20,609
how do we get those people

1482
00:43:20,610 --> 00:43:22,829
comfortable with using

1483
00:43:22,830 --> 00:43:25,229
AI for their own tasks?

1484
00:43:25,230 --> 00:43:27,239
It is so important to have that

1485
00:43:27,240 --> 00:43:29,369
team really understand

1486
00:43:29,370 --> 00:43:31,859
how I can help them personally,

1487
00:43:31,860 --> 00:43:33,869
because that immediately filters

1488
00:43:33,870 --> 00:43:35,789
down to the rest of the

1489
00:43:35,790 --> 00:43:37,889
team. So you've got to get that team

1490
00:43:37,890 --> 00:43:39,419
on board so that could be just

1491
00:43:39,420 --> 00:43:40,439
running.

1492
00:43:40,440 --> 00:43:42,299
So inspirational workshops for those

1493
00:43:42,300 --> 00:43:43,109
leaders.

1494
00:43:43,110 --> 00:43:44,639
Maybe have a breakout session at

1495
00:43:44,640 --> 00:43:46,709
your next leadership offsite.

1496
00:43:46,710 --> 00:43:49,169
And it's not about encouraging

1497
00:43:49,170 --> 00:43:50,879
them about how their teams should be

1498
00:43:50,880 --> 00:43:52,319
using AI.

1499
00:43:52,320 --> 00:43:54,539
It's about how do you

1500
00:43:54,540 --> 00:43:55,919
use AI?

1501
00:43:55,920 --> 00:43:57,899
If you're a c o,

1502
00:43:57,900 --> 00:43:59,819
you have a set of

1503
00:43:59,820 --> 00:44:01,979
direct reports that you need to

1504
00:44:01,980 --> 00:44:03,689
manage and inspire.

1505
00:44:03,690 --> 00:44:05,789
You have got colleagues

1506
00:44:05,790 --> 00:44:08,129
who run finance or on

1507
00:44:08,130 --> 00:44:10,229
HR, or who run legal.

1508
00:44:10,230 --> 00:44:11,849
And how do you better understand

1509
00:44:11,850 --> 00:44:13,709
their view of the world as a

1510
00:44:13,710 --> 00:44:15,539
CEO? Or you may be just

1511
00:44:15,540 --> 00:44:16,589
going through a merger or

1512
00:44:16,590 --> 00:44:18,389
acquisition, and you may be having

1513
00:44:18,390 --> 00:44:20,219
to figure out, how do I

1514
00:44:20,220 --> 00:44:22,379
align this whole new set of people

1515
00:44:22,380 --> 00:44:24,239
and processes and data into our

1516
00:44:24,240 --> 00:44:25,139
organization?

1517
00:44:25,140 --> 00:44:26,759
So it's thinking about how does

1518
00:44:26,760 --> 00:44:28,619
ChatGPT, how does Claude, how

1519
00:44:28,620 --> 00:44:30,839
does Gemini, how does that solve

1520
00:44:30,840 --> 00:44:32,819
your personal, daily

1521
00:44:32,820 --> 00:44:35,069
and monthly working processes?

1522
00:44:35,070 --> 00:44:36,449
And if you can figure that out

1523
00:44:36,450 --> 00:44:38,099
immediately, it gives that

1524
00:44:38,100 --> 00:44:40,409
permission to everyone below

1525
00:44:40,410 --> 00:44:42,929
to explore and experiment as well.

1526
00:44:42,930 --> 00:44:44,639
I tried to just have one

1527
00:44:44,640 --> 00:44:45,899
subscription, but I now end up

1528
00:44:45,900 --> 00:44:47,249
having to have two.

1529
00:44:47,250 --> 00:44:49,199
So I have both Claude

1530
00:44:49,200 --> 00:44:51,029
and Betty because I

1531
00:44:51,030 --> 00:44:52,289
just find Claud writes things

1532
00:44:52,290 --> 00:44:53,999
better. But ChatGPT has a wider

1533
00:44:54,000 --> 00:44:55,289
range of capabilities and

1534
00:44:55,290 --> 00:44:56,969
functionality and also is attached

1535
00:44:56,970 --> 00:44:58,889
to the internet, and I use it when

1536
00:44:58,890 --> 00:45:01,259
I'm stuck thinking,

1537
00:45:01,260 --> 00:45:03,239
but and kind of in those use cases

1538
00:45:03,240 --> 00:45:04,679
you talked about and like the team

1539
00:45:04,680 --> 00:45:06,599
use it for ideas for deal

1540
00:45:06,600 --> 00:45:08,759
reviews or how to write a better

1541
00:45:08,760 --> 00:45:09,779
proposal.

1542
00:45:09,780 --> 00:45:11,879
But what I really want to use it for

1543
00:45:11,880 --> 00:45:13,829
is all of the hard things

1544
00:45:13,830 --> 00:45:14,789
that are boring.

1545
00:45:14,790 --> 00:45:17,309
And I hate like org charts,

1546
00:45:17,310 --> 00:45:19,389
but I ended up spending about 2.5

1547
00:45:19,390 --> 00:45:21,059
hours trying to get it to write me

1548
00:45:21,060 --> 00:45:22,829
an org chart, and I could not.

1549
00:45:22,830 --> 00:45:24,929
And I was asking it to help me

1550
00:45:24,930 --> 00:45:26,819
figure out how to tell it, to write

1551
00:45:26,820 --> 00:45:28,649
me an org chart, and I could not.

1552
00:45:28,650 --> 00:45:30,269
Is this what it's worth, going for

1553
00:45:30,270 --> 00:45:32,219
some sort of AI training course?

1554
00:45:32,220 --> 00:45:33,749
Or is it just really bad at making

1555
00:45:33,750 --> 00:45:35,019
org charts no matter what?

1556
00:45:35,020 --> 00:45:36,539
Because for whatever reason, they're

1557
00:45:36,540 --> 00:45:38,069
incredibly difficult.

1558
00:45:38,070 --> 00:45:39,599
So there's some basic tips that you

1559
00:45:39,600 --> 00:45:41,399
can follow. So whether you're using

1560
00:45:41,400 --> 00:45:43,719
code called ChatGPT.

1561
00:45:43,720 --> 00:45:44,889
Gemini.

1562
00:45:44,890 --> 00:45:46,359
Any of these tools.

1563
00:45:46,360 --> 00:45:48,879
Firstly, in both Claude and ChatGPT,

1564
00:45:48,880 --> 00:45:50,739
you can set up a project and

1565
00:45:50,740 --> 00:45:52,839
set a project as a wrapper

1566
00:45:52,840 --> 00:45:54,699
around a task or a set of

1567
00:45:54,700 --> 00:45:55,599
chats.

1568
00:45:55,600 --> 00:45:57,459
So as a CEO, if

1569
00:45:57,460 --> 00:45:59,679
you're doing anything repeatedly.

1570
00:45:59,680 --> 00:46:01,539
So that might be a set of chats

1571
00:46:01,540 --> 00:46:03,549
about one of your direct reports.

1572
00:46:03,550 --> 00:46:05,469
It might be a set of chats about an

1573
00:46:05,470 --> 00:46:06,909
acquisition you're going through.

1574
00:46:06,910 --> 00:46:08,499
It might be a set of chats about a

1575
00:46:08,500 --> 00:46:10,419
policy or process that you're

1576
00:46:10,420 --> 00:46:11,799
crafting.

1577
00:46:11,800 --> 00:46:12,999
Create a project.

1578
00:46:13,000 --> 00:46:15,399
You then upload background

1579
00:46:15,400 --> 00:46:16,809
information so that could be

1580
00:46:16,810 --> 00:46:18,669
documents. It could be

1581
00:46:18,670 --> 00:46:20,889
just your instructions about how

1582
00:46:20,890 --> 00:46:22,869
you want ChatGPT to

1583
00:46:22,870 --> 00:46:24,339
respond to you.

1584
00:46:24,340 --> 00:46:26,229
And then you get into how do you

1585
00:46:26,230 --> 00:46:27,579
craft a prompt.

1586
00:46:27,580 --> 00:46:29,529
And I'll give you some specifics

1587
00:46:29,530 --> 00:46:31,599
here. So my typical

1588
00:46:31,600 --> 00:46:33,579
prompt, especially something when

1589
00:46:33,580 --> 00:46:35,739
I'm reusing it, is anything from

1590
00:46:35,740 --> 00:46:37,719
6 to 8, nine, ten

1591
00:46:37,720 --> 00:46:39,519
pages long ago.

1592
00:46:39,520 --> 00:46:41,289
That's that's longer the little chat

1593
00:46:41,290 --> 00:46:42,219
window there.

1594
00:46:42,220 --> 00:46:43,929
But because I'm, I'm using that

1595
00:46:43,930 --> 00:46:45,069
prompt again and again, it's

1596
00:46:45,070 --> 00:46:46,449
worthwhile.

1597
00:46:46,450 --> 00:46:48,789
So I break it up with

1598
00:46:48,790 --> 00:46:50,559
XHTML tags.

1599
00:46:50,560 --> 00:46:52,809
Now you don't need to be a developer

1600
00:46:52,810 --> 00:46:54,279
to know this. I'll just explain it

1601
00:46:54,280 --> 00:46:55,509
super simply. If you don't want to

1602
00:46:55,510 --> 00:46:56,589
add XML tags.

1603
00:46:56,590 --> 00:46:58,659
So a left arrow kind of opens

1604
00:46:58,660 --> 00:47:00,009
it. And then you write the word

1605
00:47:00,010 --> 00:47:02,289
objective and then a right arrow

1606
00:47:02,290 --> 00:47:03,939
to close it. So that's your opening

1607
00:47:03,940 --> 00:47:05,589
tag objective.

1608
00:47:05,590 --> 00:47:07,509
And then to close that XML

1609
00:47:07,510 --> 00:47:09,549
tag you do exactly the same, except

1610
00:47:09,550 --> 00:47:11,409
there's a backslash up at

1611
00:47:11,410 --> 00:47:12,999
the start after the first opening

1612
00:47:13,000 --> 00:47:13,779
bracket.

1613
00:47:13,780 --> 00:47:15,879
And I'll put some of the notes about

1614
00:47:15,880 --> 00:47:17,229
this prompt and guidelines so that

1615
00:47:17,230 --> 00:47:19,179
you can have it in the show notes.

1616
00:47:19,180 --> 00:47:21,189
Now when you're writing your prompt

1617
00:47:21,190 --> 00:47:23,019
you're basically structuring.

1618
00:47:23,020 --> 00:47:24,039
So I say objective.

1619
00:47:24,040 --> 00:47:25,119
This is the first part.

1620
00:47:25,120 --> 00:47:27,009
The objective of this prompt is

1621
00:47:27,010 --> 00:47:29,379
to develop a job description

1622
00:47:29,380 --> 00:47:31,269
or a merger plan, whatever it might

1623
00:47:31,270 --> 00:47:33,429
be. Then the next tag might be

1624
00:47:33,430 --> 00:47:35,229
open the tag instructions.

1625
00:47:35,230 --> 00:47:36,729
Right. This is how I want you to do

1626
00:47:36,730 --> 00:47:38,079
it. The user is going to provide you

1627
00:47:38,080 --> 00:47:39,099
with this information.

1628
00:47:39,100 --> 00:47:40,329
I mean, I'm the user, but I'm going

1629
00:47:40,330 --> 00:47:42,009
to provide you with this information.

1630
00:47:42,010 --> 00:47:43,719
I then want you to ask me some

1631
00:47:43,720 --> 00:47:44,919
questions. And here are the

1632
00:47:44,920 --> 00:47:46,569
questions which I put in another

1633
00:47:46,570 --> 00:47:47,919
question tag.

1634
00:47:47,920 --> 00:47:49,899
Then you get to the two most

1635
00:47:49,900 --> 00:47:51,819
important parts of a really

1636
00:47:51,820 --> 00:47:53,289
great prompt.

1637
00:47:53,290 --> 00:47:54,849
And they're often the hardest bits

1638
00:47:54,850 --> 00:47:55,719
to put it in.

1639
00:47:55,720 --> 00:47:57,789
And this is what takes their long

1640
00:47:57,790 --> 00:47:59,229
time and makes it a ten page

1641
00:47:59,230 --> 00:48:00,249
document.

1642
00:48:00,250 --> 00:48:01,989
So examples.

1643
00:48:01,990 --> 00:48:04,569
So think about the intern.

1644
00:48:04,570 --> 00:48:06,489
If I ask you to build

1645
00:48:06,490 --> 00:48:08,559
me an org chart, but I don't

1646
00:48:08,560 --> 00:48:10,419
show you what a good org

1647
00:48:10,420 --> 00:48:11,649
chart looks like.

1648
00:48:11,650 --> 00:48:13,479
2 or 3 of them, you're just

1649
00:48:13,480 --> 00:48:15,129
running around in the darkness.

1650
00:48:15,130 --> 00:48:17,139
If I don't show you a bad

1651
00:48:17,140 --> 00:48:18,849
org chart, you don't know what bad

1652
00:48:18,850 --> 00:48:19,959
looks like.

1653
00:48:19,960 --> 00:48:22,059
So whether it is,

1654
00:48:22,060 --> 00:48:23,979
you know, a job description,

1655
00:48:23,980 --> 00:48:26,289
whether it's a spreadsheet table,

1656
00:48:26,290 --> 00:48:28,719
whether it is a project timeline,

1657
00:48:28,720 --> 00:48:30,849
whatever it is that you want.

1658
00:48:30,850 --> 00:48:32,649
A good prompt should always have at

1659
00:48:32,650 --> 00:48:34,989
least 2 or 3 examples,

1660
00:48:34,990 --> 00:48:36,639
a good one, and then a bad one.

1661
00:48:36,640 --> 00:48:38,829
So that can take up quite a bit.

1662
00:48:38,830 --> 00:48:40,749
The final tag that I always

1663
00:48:40,750 --> 00:48:42,789
have in there is exceptions.

1664
00:48:42,790 --> 00:48:44,709
So what should the

1665
00:48:44,710 --> 00:48:46,869
AI do if

1666
00:48:46,870 --> 00:48:48,729
you don't have certain information

1667
00:48:48,730 --> 00:48:49,659
that you've asked for?

1668
00:48:49,660 --> 00:48:51,609
So for example, if I'm drafting

1669
00:48:51,610 --> 00:48:53,799
a job description and I've asked

1670
00:48:53,800 --> 00:48:55,869
the AI and the instructions to

1671
00:48:55,870 --> 00:48:58,569
ask me for the salary range,

1672
00:48:58,570 --> 00:48:59,709
let's say I haven't got the salary

1673
00:48:59,710 --> 00:49:01,569
range yet. Well, what should the AI

1674
00:49:01,570 --> 00:49:02,919
do in that situation?

1675
00:49:02,920 --> 00:49:04,449
You might say if the user does not

1676
00:49:04,450 --> 00:49:06,999
have a salary range, then

1677
00:49:07,000 --> 00:49:08,739
know that this is to be verified

1678
00:49:08,740 --> 00:49:10,599
later and at the bottom of

1679
00:49:10,600 --> 00:49:12,579
your response, put some next

1680
00:49:12,580 --> 00:49:15,339
actions or follow ups or whatever.

1681
00:49:15,340 --> 00:49:17,199
So this sounds quite

1682
00:49:17,200 --> 00:49:19,239
detailed, but if you're

1683
00:49:19,240 --> 00:49:20,859
doing one to ones with your direct

1684
00:49:20,860 --> 00:49:23,409
reports, you do that regularly

1685
00:49:23,410 --> 00:49:24,549
and you're going to do it a lot.

1686
00:49:24,550 --> 00:49:26,439
So it's worth spending an hour

1687
00:49:26,440 --> 00:49:28,629
writing the prompt

1688
00:49:28,630 --> 00:49:30,339
so that you've got that in there.

1689
00:49:30,340 --> 00:49:32,229
So there's just a bit of guidance.

1690
00:49:32,230 --> 00:49:34,299
If you put the the right

1691
00:49:34,300 --> 00:49:36,139
in front to really

1692
00:49:36,140 --> 00:49:37,939
pass the intern test.

1693
00:49:37,940 --> 00:49:39,199
If you gave that to an intern,

1694
00:49:39,200 --> 00:49:40,669
they'd give you a good response.

1695
00:49:40,670 --> 00:49:42,019
Then I think you'll start to find

1696
00:49:42,020 --> 00:49:42,889
you get better.

1697
00:49:42,890 --> 00:49:45,049
Whether that's an org chart

1698
00:49:45,050 --> 00:49:46,369
has on that, because maybe you're

1699
00:49:46,370 --> 00:49:47,719
trying to get a visual, which it

1700
00:49:47,720 --> 00:49:48,949
probably wouldn't be very good at.

1701
00:49:48,950 --> 00:49:50,359
But in terms of the hierarchical

1702
00:49:50,360 --> 00:49:52,399
structure would probably be

1703
00:49:52,400 --> 00:49:54,349
pretty handy at doing that.

1704
00:49:54,350 --> 00:49:56,299
We in our company

1705
00:49:56,300 --> 00:49:58,459
kick off this year, we had everybody

1706
00:49:58,460 --> 00:50:00,319
working groups to identify use

1707
00:50:00,320 --> 00:50:02,059
cases along the customer journey

1708
00:50:02,060 --> 00:50:03,079
where applying.

1709
00:50:03,080 --> 00:50:05,179
I would be really cool.

1710
00:50:05,180 --> 00:50:07,399
And every single

1711
00:50:07,400 --> 00:50:08,929
team, they came up with loads of

1712
00:50:08,930 --> 00:50:10,219
different ideas, but every single

1713
00:50:10,220 --> 00:50:12,199
team came up with one idea that was

1714
00:50:12,200 --> 00:50:14,299
the same, which is basically

1715
00:50:14,300 --> 00:50:16,189
putting all

1716
00:50:16,190 --> 00:50:18,169
of our internal data

1717
00:50:18,170 --> 00:50:20,659
in a data repository and putting

1718
00:50:20,660 --> 00:50:22,789
ChatGPT on top, or a GPT

1719
00:50:22,790 --> 00:50:24,679
on top. So like all of the

1720
00:50:24,680 --> 00:50:26,599
historic slack

1721
00:50:26,600 --> 00:50:28,699
information, all of our Google

1722
00:50:28,700 --> 00:50:29,899
drives.

1723
00:50:29,900 --> 00:50:31,759
Then you have Salesforce

1724
00:50:31,760 --> 00:50:33,679
data, etc., then be able

1725
00:50:33,680 --> 00:50:35,959
to from that repository, ask

1726
00:50:35,960 --> 00:50:37,789
like, give me a summary of

1727
00:50:37,790 --> 00:50:39,019
what the customer relationship has

1728
00:50:39,020 --> 00:50:39,949
been like to date.

1729
00:50:39,950 --> 00:50:42,349
Give me a summary about like

1730
00:50:42,350 --> 00:50:43,849
usage of that customer, all of those

1731
00:50:43,850 --> 00:50:45,109
types of things.

1732
00:50:45,110 --> 00:50:46,969
So that we go think about the

1733
00:50:46,970 --> 00:50:48,469
build versus buy.

1734
00:50:48,470 --> 00:50:50,179
I'd be taking a look at glean.

1735
00:50:50,180 --> 00:50:51,679
I'm not sure if you come across Glen

1736
00:50:51,680 --> 00:50:54,469
GLE and.com.

1737
00:50:54,470 --> 00:50:57,079
And it's exactly that use case.

1738
00:50:57,080 --> 00:50:59,209
So founded by some ex Googlers.

1739
00:50:59,210 --> 00:51:01,609
How do we use AI

1740
00:51:01,610 --> 00:51:03,619
to search across our

1741
00:51:03,620 --> 00:51:06,049
existing data sources.

1742
00:51:06,050 --> 00:51:07,879
And at first glance you say,

1743
00:51:07,880 --> 00:51:09,859
well, you know, the challenge

1744
00:51:09,860 --> 00:51:11,689
is quite easy because we just go

1745
00:51:11,690 --> 00:51:13,519
and, you know, plug it in via APIs

1746
00:51:13,520 --> 00:51:15,529
to all these different things, but

1747
00:51:15,530 --> 00:51:17,749
you need to preserve the

1748
00:51:17,750 --> 00:51:19,639
data privacy rights of

1749
00:51:19,640 --> 00:51:20,899
the source material.

1750
00:51:20,900 --> 00:51:22,999
So if I go search customer

1751
00:51:23,000 --> 00:51:24,979
X, how much did they spend last

1752
00:51:24,980 --> 00:51:26,839
year. Well, I should only be able

1753
00:51:26,840 --> 00:51:28,699
to get that answer if

1754
00:51:28,700 --> 00:51:29,959
I would be able to get that

1755
00:51:29,960 --> 00:51:31,369
information in the actual source

1756
00:51:31,370 --> 00:51:33,139
system, because I've got access to

1757
00:51:33,140 --> 00:51:34,759
that data or whatever.

1758
00:51:34,760 --> 00:51:36,589
Anyway, so I had a chat with

1759
00:51:36,590 --> 00:51:38,089
one of their team just before

1760
00:51:38,090 --> 00:51:39,199
Christmas and it was a super

1761
00:51:39,200 --> 00:51:40,399
interesting use case.

1762
00:51:40,400 --> 00:51:42,739
But that's exactly what Glenn is

1763
00:51:42,740 --> 00:51:44,359
targeted, that.

1764
00:51:44,360 --> 00:51:45,979
When we first kind of reconnected

1765
00:51:45,980 --> 00:51:47,779
and started talking and you talked

1766
00:51:47,780 --> 00:51:49,789
about your Power Hour,

1767
00:51:49,790 --> 00:51:51,529
and I think I have told everybody I

1768
00:51:51,530 --> 00:51:53,629
know about it, and I thought, why

1769
00:51:53,630 --> 00:51:55,819
not share it with the rest of

1770
00:51:55,820 --> 00:51:57,049
our listeners?

1771
00:51:57,050 --> 00:51:58,909
What's your power hour?

1772
00:51:58,910 --> 00:52:01,699
So my Power Hour started

1773
00:52:01,700 --> 00:52:03,859
during Covid,

1774
00:52:03,860 --> 00:52:05,899
and like many of us,

1775
00:52:05,900 --> 00:52:07,789
I was a commuter before

1776
00:52:07,790 --> 00:52:09,559
Covid, jumping on a train.

1777
00:52:09,560 --> 00:52:11,539
God knows what hour into

1778
00:52:11,540 --> 00:52:12,499
London.

1779
00:52:12,500 --> 00:52:14,359
Covid came along and suddenly

1780
00:52:14,360 --> 00:52:16,279
I had all this time in the morning

1781
00:52:16,280 --> 00:52:18,169
and my wife would take the kids to

1782
00:52:18,170 --> 00:52:20,089
school and, you know, I'd

1783
00:52:20,090 --> 00:52:21,889
be up and about, you know, having

1784
00:52:21,890 --> 00:52:22,819
taking the dogs for a walk.

1785
00:52:22,820 --> 00:52:24,169
7:00.

1786
00:52:24,170 --> 00:52:25,069
And then the kids are off.

1787
00:52:25,070 --> 00:52:26,419
And then I had to sort of hour from

1788
00:52:26,420 --> 00:52:28,399
eight till nine when nothing

1789
00:52:28,400 --> 00:52:30,139
was really happening.

1790
00:52:30,140 --> 00:52:32,089
And so I decided that I was going

1791
00:52:32,090 --> 00:52:33,979
to start to write a book

1792
00:52:33,980 --> 00:52:35,689
during that hour.

1793
00:52:35,690 --> 00:52:37,519
And so I would just write, you

1794
00:52:37,520 --> 00:52:39,529
know, a chapter in an hour, having

1795
00:52:39,530 --> 00:52:40,309
thought about what was going to

1796
00:52:40,310 --> 00:52:41,599
write in the shower.

1797
00:52:41,600 --> 00:52:42,799
And suddenly I found it was like

1798
00:52:42,800 --> 00:52:43,729
super productive.

1799
00:52:43,730 --> 00:52:45,679
Like by the time I got to 9:00, I

1800
00:52:45,680 --> 00:52:47,719
had done a ton of work and probably

1801
00:52:47,720 --> 00:52:49,579
my most creative work of the

1802
00:52:49,580 --> 00:52:51,349
day. I was lucky enough to finish

1803
00:52:51,350 --> 00:52:53,149
published that book, and I just kept

1804
00:52:53,150 --> 00:52:54,979
that power hour going.

1805
00:52:54,980 --> 00:52:56,749
And it's the hour that I work for

1806
00:52:56,750 --> 00:52:58,699
myself, whether that's on

1807
00:52:58,700 --> 00:53:00,169
sort of personal writing, whether

1808
00:53:00,170 --> 00:53:02,389
it's recording a YouTube video.

1809
00:53:02,390 --> 00:53:04,369
And I'm very humbled

1810
00:53:04,370 --> 00:53:05,929
that I get lots of lovely comments

1811
00:53:05,930 --> 00:53:07,669
from people like, I've got no idea

1812
00:53:07,670 --> 00:53:09,799
how you create so much stuff.

1813
00:53:09,800 --> 00:53:11,509
You know, there's always a new

1814
00:53:11,510 --> 00:53:12,859
e-book that you've published, or

1815
00:53:12,860 --> 00:53:14,569
you've recorded a video for YouTube,

1816
00:53:14,570 --> 00:53:16,399
or you've done a podcast, or you've

1817
00:53:16,400 --> 00:53:17,929
published a book and you know, how

1818
00:53:17,930 --> 00:53:18,979
are you doing all this stuff?

1819
00:53:18,980 --> 00:53:20,389
And you've got four children and two

1820
00:53:20,390 --> 00:53:22,159
dogs and you've got, you know, your

1821
00:53:22,160 --> 00:53:24,199
work and everything, and it's that

1822
00:53:24,200 --> 00:53:26,389
Power hour has been

1823
00:53:26,390 --> 00:53:27,529
super helpful.

1824
00:53:27,530 --> 00:53:29,729
And once you get into the

1825
00:53:29,730 --> 00:53:31,589
habit of it, then

1826
00:53:31,590 --> 00:53:33,029
it's very difficult to break.

1827
00:53:33,030 --> 00:53:34,919
It doesn't actually seem like work

1828
00:53:34,920 --> 00:53:36,809
anymore. It's just a routine I

1829
00:53:36,810 --> 00:53:37,859
find myself.

1830
00:53:37,860 --> 00:53:39,689
Weekends included at

1831
00:53:39,690 --> 00:53:41,699
my desk from eight till nine.

1832
00:53:41,700 --> 00:53:44,099
Just crank through some great work.

1833
00:53:44,100 --> 00:53:46,109
The book that inspired that was

1834
00:53:46,110 --> 00:53:48,089
Atomic Habits, which came out

1835
00:53:48,090 --> 00:53:50,609
pre-COVID. I'm sure many people have

1836
00:53:50,610 --> 00:53:52,169
read it was an Amazon bestseller,

1837
00:53:52,170 --> 00:53:53,939
but it was all about, you know, if

1838
00:53:53,940 --> 00:53:55,859
you create the framework

1839
00:53:55,860 --> 00:53:57,899
at 8:00, I will sit down

1840
00:53:57,900 --> 00:53:59,999
at my desk, create the input,

1841
00:54:00,000 --> 00:54:02,009
and then the output arrives

1842
00:54:02,010 --> 00:54:03,479
magically.

1843
00:54:03,480 --> 00:54:05,009
Lovely. So if you like what you

1844
00:54:05,010 --> 00:54:06,509
hear, please leave us a comment or

1845
00:54:06,510 --> 00:54:08,429
subscribe and we will wrap on

1846
00:54:08,430 --> 00:54:10,139
this episode of The Operations Room.

1847
00:54:10,140 --> 00:54:11,159
Thank you for joining us, Charlie.