1
00:00:06,000 --> 00:00:09,037
Good morning everyone. How's it going?

2
00:00:09,120 --> 00:00:12,557
Everybody excited to be here?

3
00:00:12,640 --> 00:00:16,077
>> Okay, for those of you that were in
the Somebody texted me, they called it the

4
00:00:16,160 --> 00:00:19,277
How I AI sauna last night, the party we

5
00:00:19,360 --> 00:00:21,837
had. I told you all you need to hype me

6
00:00:21,920 --> 00:00:23,437
up this morning because I was out way

7
00:00:23,520 --> 00:00:26,317
past my bedtime, which is 8:15. I've got

8
00:00:26,400 --> 00:00:30,717
a little baby. Um but I am here prepared

9
00:00:30,800 --> 00:00:34,077
to do the thing that I do. So last year,

10
00:00:34,160 --> 00:00:36,237
last year I came here and I said product

11
00:00:36,320 --> 00:00:38,317
management is dead. I said that because

12
00:00:38,400 --> 00:00:43,517
as a product leader I like to make these
big grand statements and then be proved

13
00:00:43,600 --> 00:00:47,677
completely wrong in the market.
Just keeps me really honest. So product

14
00:00:47,760 --> 00:00:51,837
management is obviously not dead.
There are so many amazing product leaders,

15
00:00:51,920 --> 00:00:56,237
product managers, product executives
here in this room. But who's with me

16
00:00:56,320 --> 00:00:57,837
that between the last time we had

17
00:00:57,920 --> 00:00:59,997
Lenny's summit and now product

18
00:01:00,080 --> 00:01:03,277
management is completely different.

19
00:01:03,360 --> 00:01:05,117
Okay, got some hands. It's really

20
00:01:05,200 --> 00:01:07,917
different. And so last year I came for

21
00:01:08,000 --> 00:01:10,397
PMS that didn't work. Let's do it again.

22
00:01:10,480 --> 00:01:14,477
This year I'm coming for the road map.

23
00:01:14,560 --> 00:01:18,957
The Yeah, you're you're welcome. OKR's

24
00:01:19,040 --> 00:01:23,107
next and then my job is over.

25
00:01:23,190 --> 00:01:26,077
>> [laughter] >> So the road
map has been the defining

26
00:01:26,160 --> 00:01:28,637
artifact of our industry, of our

27
00:01:28,720 --> 00:01:31,517
careers. It's supposed to tell us and

28
00:01:31,600 --> 00:01:34,477
our teams where we're going, what's

29
00:01:34,560 --> 00:01:38,557
next, and what matters. And I have been

30
00:01:38,640 --> 00:01:42,077
in product for over two decades. And I

31
00:01:42,160 --> 00:01:45,277
believe, fingers crossed, a lot of us in

32
00:01:45,360 --> 00:01:47,757
our room in this room are about to write

33
00:01:47,840 --> 00:01:50,717
our last road map. Before I go into why

34
00:01:50,800 --> 00:01:53,597
I think that, I want to tell you all I

35
00:01:53,680 --> 00:01:56,157
have a confession.

36
00:01:56,240 --> 00:01:59,277
I am shipping more than ever. I have

37
00:01:59,360 --> 00:02:03,037
access to the most intelligent coding

38
00:02:03,120 --> 00:02:05,517
agents in the world. I have the best

39
00:02:05,600 --> 00:02:09,197
developer tools anybody can ask for. I

40
00:02:09,280 --> 00:02:13,277
have super smart models. I've got

41
00:02:13,360 --> 00:02:17,517
customer context via API, via CLI, via

42
00:02:17,600 --> 00:02:21,997
MCP. all the letters I have. I write so

43
00:02:22,080 --> 00:02:24,077
many skills. And y'all, I counted

44
00:02:24,160 --> 00:02:28,317
yesterday. I have 40 Grock bots. I can

45
00:02:28,400 --> 00:02:32,637
build almost anything. And here is my

46
00:02:32,720 --> 00:02:35,197
true true confession.

47
00:02:35,280 --> 00:02:37,677
I am out of good ideas, y'all. I'm like

48
00:02:37,760 --> 00:02:42,237
straight up out of good ideas of things

49
00:02:42,320 --> 00:02:45,037
to build. I'm out. I can build. I just

50
00:02:45,120 --> 00:02:47,277
don't think any of it's a good idea. So,

51
00:02:47,360 --> 00:02:50,317
and and I'm not saying I don't have

52
00:02:50,400 --> 00:02:52,397
plausible requests. It's not that I'm

53
00:02:52,480 --> 00:02:54,637
saying I don't have things I could

54
00:02:54,720 --> 00:02:58,557
build. It's just that execution has

55
00:02:58,640 --> 00:03:02,637
outrun my ability to discover

56
00:03:02,720 --> 00:03:06,317
meaningful, meaty, my favorite word,

57
00:03:06,400 --> 00:03:07,997
commercializable

58
00:03:08,080 --> 00:03:11,997
products in the market. And this is a

59
00:03:12,080 --> 00:03:14,477
very different situation that I have

60
00:03:14,560 --> 00:03:17,997
been in than before. You know, it used

61
00:03:18,080 --> 00:03:22,317
to be that engineering capacity was the

62
00:03:22,400 --> 00:03:24,077
scarce resource. That was like I mean

63
00:03:24,160 --> 00:03:26,717
that's kind of why we exist honestly is

64
00:03:26,800 --> 00:03:29,517
like engineering capacity used to be the

65
00:03:29,600 --> 00:03:33,037
really scarce resource. We had way more

66
00:03:33,120 --> 00:03:36,477
ideas than people and we had way more

67
00:03:36,560 --> 00:03:39,437
demand than our ability to deliver

68
00:03:39,520 --> 00:03:42,637
against that demand. And the product

69
00:03:42,720 --> 00:03:46,557
manager, the products leaders job was to

70
00:03:46,640 --> 00:03:49,757
prioritize all those ideas, sequence

71
00:03:49,840 --> 00:03:52,717
them, and then spend like all our time

72
00:03:52,800 --> 00:03:55,357
in meetings and in Slack and in

73
00:03:55,440 --> 00:03:59,357
spreadsheets saying no. Saying no. PMs

74
00:03:59,440 --> 00:04:03,197
should say no more. we should have cut
lines. It was like very aggressive like

75
00:04:03,280 --> 00:04:06,797
here's my cut line for the quarter.
This is my priority list. This is my stack

76
00:04:06,880 --> 00:04:09,677
rank. And while the PM's job was to say

77
00:04:09,760 --> 00:04:12,397
no, it felt like engineering's job was

78
00:04:12,480 --> 00:04:15,037
to say like no, not like that. Not with

79
00:04:15,120 --> 00:04:20,157
all those features, not with perfect
architecture. We can only do this much.

80
00:04:20,240 --> 00:04:23,757
And then it seemed like design's job
was say to like no, wait for us, like

81
00:04:23,840 --> 00:04:26,957
please. And so everybody's like just

82
00:04:27,040 --> 00:04:31,517
constrained on this like precious
engineering capacity. And so things got

83
00:04:31,600 --> 00:04:35,517
narrow narrow narrowed down till you
only shipped a little bit of your road

84
00:04:35,600 --> 00:04:37,637
map. And now I feel like I have more

85
00:04:37,720 --> 00:04:40,717
[clears throat] execution capacity than

86
00:04:40,800 --> 00:04:43,757
like true conviction about what to

87
00:04:43,840 --> 00:04:47,917
build. So my bottleneck has moved from

88
00:04:48,000 --> 00:04:52,477
like building what what can I build to

89
00:04:52,560 --> 00:04:55,277
believing like what do I believe is

90
00:04:55,360 --> 00:04:59,517
actually worth building. And look, I am

91
00:04:59,600 --> 00:05:03,837
like the biggest token maxer on the
planet. I'm like PR's up 3x. Like do it

92
00:05:03,920 --> 00:05:07,677
all. Ship, ship, ship. Agents everywhere.
I don't need a PM. I don't

93
00:05:07,760 --> 00:05:10,317
need any specs. But like I think this

94
00:05:10,400 --> 00:05:12,557
gap between our building capacity and

95
00:05:12,640 --> 00:05:15,997
like true in our heart of hearts

96
00:05:16,080 --> 00:05:18,237
conviction like does any of this code

97
00:05:18,320 --> 00:05:22,077
matter is a much bigger problem than any

98
00:05:22,160 --> 00:05:24,717
of us are willing to admit. I think

99
00:05:24,800 --> 00:05:26,557
we're all getting a lot of pressure from

100
00:05:26,640 --> 00:05:29,277
the board, from the timeline, from each

101
00:05:29,360 --> 00:05:31,917
other, from our teams, from our bosses

102
00:05:32,000 --> 00:05:36,957
to move faster, to get leverage with AI,
to like really lean in and show we're AI

103
00:05:37,040 --> 00:05:38,477
native. We can build in this new way.

104
00:05:38,560 --> 00:05:40,157
Claire said product management is dead.

105
00:05:40,240 --> 00:05:41,917
She showed all these scary circles. She

106
00:05:42,000 --> 00:05:47,197
said like ship to the moon. Do it,
do it, do it. But like I do think none of

107
00:05:47,280 --> 00:05:48,877
us are I know this is a question I asked

108
00:05:48,960 --> 00:05:52,077
one who's who's shipping more than ever

109
00:05:52,160 --> 00:05:56,397
in the last year at their companies
whose revenue is going up proportionate

110
00:05:56,480 --> 00:05:58,637
with the amount of PRs

111
00:05:58,720 --> 00:06:03,517
like that's the fundamental problem
we're facing here and so I'm going to

112
00:06:03,600 --> 00:06:07,117
tell a story and like bring this home
for you this like I have no good ideas

113
00:06:07,200 --> 00:06:10,797
and I don't know if they matter. Um so
I had this idea for this thing called the

114
00:06:10,880 --> 00:06:15,357
product graph for chat PRD. It's like
a It's a real product for product managers

115
00:06:15,440 --> 00:06:17,917
idea and it's not special because
everybody's done it. But it's like we're

116
00:06:18,000 --> 00:06:21,837
going to suck in everything a company
knows about customers, about what you're

117
00:06:21,920 --> 00:06:23,517
working on, blah blah blah blah blah.

118
00:06:23,600 --> 00:06:27,437
We're going to parse it. We're going to
like slap, you know, opus on it and then

119
00:06:27,520 --> 00:06:30,957
it's all going to be available through
agents and it's going to be awesome.

120
00:06:31,040 --> 00:06:33,757
It's going to make you be able to make
these decisions better and build better

121
00:06:33,840 --> 00:06:37,997
products and of course write better PRDs.
And so I built it. I built this

122
00:06:38,080 --> 00:06:39,917
like insights engine. I built this like

123
00:06:40,000 --> 00:06:42,477
very fancy semantic product graph. It

124
00:06:42,560 --> 00:06:45,277
like generates this auto wiki. Agents

125
00:06:45,360 --> 00:06:47,917
can consume it. It was like almost a

126
00:06:48,000 --> 00:06:50,157
oneshot. Not quite, but it was a pretty

127
00:06:50,240 --> 00:06:53,677
close oneshot. It worked. It looks

128
00:06:53,760 --> 00:06:55,597
pretty good. It was like feature

129
00:06:55,680 --> 00:06:58,077
forfeature matching competitors in the

130
00:06:58,160 --> 00:07:01,677
market. And I kind of just like every

131
00:07:01,760 --> 00:07:05,997
time I would build it, I would be like,
"This just belongs in the trash. It

132
00:07:06,080 --> 00:07:09,277
belongs in the trash. All this work,
all this amazing product belongs in the

133
00:07:09,360 --> 00:07:12,397
trash." And why? Because to me it was

134
00:07:12,480 --> 00:07:15,837
competitive, it wasn't differentiated.

135
00:07:15,920 --> 00:07:19,517
And so, you know, even though I had
built this thing that like obviously had

136
00:07:19,600 --> 00:07:21,677
a market, people would probably want it,

137
00:07:21,760 --> 00:07:23,837
customers said it was interesting, like

138
00:07:23,920 --> 00:07:26,637
in my core, I felt like it wasn't worth

139
00:07:26,720 --> 00:07:31,197
my customers attention. I like the ROI
was really shaky to me. I wasn't

140
00:07:31,280 --> 00:07:34,477
convinced on the interface. I was like,
should this it it still had a web

141
00:07:34,560 --> 00:07:37,997
interface. I was like, should it be
agent first? Is this the right thing? I

142
00:07:38,080 --> 00:07:40,317
wanted to build something surprising

143
00:07:40,400 --> 00:07:43,277
that competitors couldn't even imagine.

144
00:07:43,360 --> 00:07:46,797
And my big problem was that these tokens

145
00:07:46,880 --> 00:07:49,277
and this like road map that I had

146
00:07:49,360 --> 00:07:52,477
executed my bet, but it didn't really

147
00:07:52,560 --> 00:07:55,437
test or validate or strengthen my

148
00:07:55,520 --> 00:07:57,517
conviction. Meanwhile, I'm sitting on

149
00:07:57,600 --> 00:08:00,637
all this kind of like code to nowhere. I

150
00:08:00,720 --> 00:08:03,437
kept shipping more than ever. So,

151
00:08:03,520 --> 00:08:06,157
customer fixes were like automated. They

152
00:08:06,240 --> 00:08:10,557
would come in, they go to like one
of my bots, and then they get fixed. Goals

153
00:08:10,640 --> 00:08:14,717
were like obliterating tech debt.
I mean, I re rearchite rearchitected chat

154
00:08:14,800 --> 00:08:19,277
purity 70 times because I could. Um,
we stopped tracking issues. We're like, we

155
00:08:19,360 --> 00:08:20,797
don't need issue tracking anymore. We're

156
00:08:20,880 --> 00:08:23,597
just going to ship PRs. This AI factory

157
00:08:23,680 --> 00:08:27,117
like manifested and started running

158
00:08:27,200 --> 00:08:29,437
itself. It did the thing that we all

159
00:08:29,520 --> 00:08:33,037
want it to do. It's magical. And here's

160
00:08:33,120 --> 00:08:35,677
the secret.

161
00:08:35,760 --> 00:08:38,237
I'm not sure any of it mattered. Like

162
00:08:38,320 --> 00:08:41,277
I'm not yet convinced that any of that

163
00:08:41,360 --> 00:08:46,797
mattered. And that really got me thinking.
It got me thinking about what

164
00:08:46,880 --> 00:08:50,477
is product management? What is the
purpose of building software? How do we

165
00:08:50,560 --> 00:08:52,797
make decisions? How do we prioritize?

166
00:08:52,880 --> 00:08:55,677
How do we allocate all these tokens?

167
00:08:55,760 --> 00:08:58,397
And I just kept going back to this idea

168
00:08:58,480 --> 00:09:03,357
of the road map. road maps made sense

169
00:09:03,440 --> 00:09:06,157
under engineering scarcity. They like

170
00:09:06,240 --> 00:09:11,037
laid out your strategy. They told you
what features were you going to build.

171
00:09:11,120 --> 00:09:15,117
We had this thing called effort which
was like more than me like voice noting

172
00:09:15,200 --> 00:09:16,797
you know Devon saying can you please

173
00:09:16,880 --> 00:09:19,837
build this and we thought a lot about

174
00:09:19,920 --> 00:09:24,317
sequencing and milestones and MVPs and

175
00:09:24,400 --> 00:09:27,357
scarcity helped us theoretically let's

176
00:09:27,440 --> 00:09:29,357
be honest it helped us theoretically

177
00:09:29,440 --> 00:09:32,317
filter out our bad ideas because the

178
00:09:32,400 --> 00:09:35,277
bottom of the list just never shipped.

179
00:09:35,360 --> 00:09:38,877
It never could. There were not
enough engineers. There was not enough

180
00:09:38,960 --> 00:09:41,997
conviction across the organization. and
said this like below the cut that's what

181
00:09:42,080 --> 00:09:44,157
we called it below the cut line never

182
00:09:44,240 --> 00:09:48,797
shipped and now we can ship all our bad

183
00:09:48,880 --> 00:09:51,837
ideas like congratulations

184
00:09:51,920 --> 00:09:56,077
to us cheap execution I think still

185
00:09:56,160 --> 00:09:59,277
needs judgment and I think we're seeing

186
00:09:59,360 --> 00:10:03,677
this moment where the road map plus
like limitless or feeling like limitless

187
00:10:03,760 --> 00:10:06,237
execution capacity is creating three

188
00:10:06,320 --> 00:10:08,717
traps that I want to warn you about the

189
00:10:08,800 --> 00:10:12,877
first is the backlog track. So this is

190
00:10:12,960 --> 00:10:15,597
if you have a backlog, AI will build it.

191
00:10:15,680 --> 00:10:18,397
They will build every single thing. It

192
00:10:18,480 --> 00:10:22,957
will build every single thing on your
list. And so I hear a lot about like,

193
00:10:23,040 --> 00:10:27,037
you know, I we're burned through our
backlog. It's because all you have to do

194
00:10:27,120 --> 00:10:29,837
is slashgo build the backlog and it gets

195
00:10:29,920 --> 00:10:35,197
done. But the problem is clearing
out requests or ideas does not mean

196
00:10:35,280 --> 00:10:38,717
meaningful progress against your
business or even meaningful progress

197
00:10:38,800 --> 00:10:44,077
against a customer problem. The second
thing is the parity trap. This is

198
00:10:44,160 --> 00:10:45,677
something I really felt in that story

199
00:10:45,760 --> 00:10:48,237
about the product graph. All your

200
00:10:48,320 --> 00:10:51,837
competitors are copying each other and

201
00:10:51,920 --> 00:10:56,397
they all are talking to the same
customers and they're all sucking in the

202
00:10:56,480 --> 00:11:00,637
same insights and they're all arriving
at the same obvious conclusion and

203
00:11:00,720 --> 00:11:03,037
building the same obvious product with

204
00:11:03,120 --> 00:11:07,677
the same obvious like run the taste skill.
Run impeccable. Run whatever.

205
00:11:07,760 --> 00:11:10,637
Make it look good. Get rid of the m
dashes. Get rid of the borders. Like

206
00:11:10,720 --> 00:11:13,597
we're all doing Can you tell I took max?

207
00:11:13,680 --> 00:11:15,997
Yeah. So we're all doing the exact same

208
00:11:16,080 --> 00:11:18,637
thing. And so this like parody trap

209
00:11:18,720 --> 00:11:21,197
where we used to have this like really

210
00:11:21,280 --> 00:11:25,757
interesting dynamic with competitors now
it feels like it's evening out and again

211
00:11:25,840 --> 00:11:29,117
like this question about what Moes are
and what differentiators are is a big

212
00:11:29,200 --> 00:11:31,437
one in my mind. And then the final kind

213
00:11:31,520 --> 00:11:34,877
of trap I see is the churn trap which is

214
00:11:34,960 --> 00:11:38,077
we ship something, we see noise and pick

215
00:11:38,160 --> 00:11:40,877
up or not because we can always ship

216
00:11:40,960 --> 00:11:43,357
something else, we abandon it and then

217
00:11:43,440 --> 00:11:46,477
we never really learn or compound on top

218
00:11:46,560 --> 00:11:49,277
of what we're shipping. And so all of

219
00:11:49,360 --> 00:11:51,997
these things from the inside feel very

220
00:11:52,080 --> 00:11:54,157
productive. Like my backlog's going

221
00:11:54,240 --> 00:11:56,637
down, my competitive features are going

222
00:11:56,720 --> 00:11:59,517
up. like we're shipping more than ever,

223
00:11:59,600 --> 00:12:02,317
but what I think this does is it just

224
00:12:02,400 --> 00:12:05,997
like accelerates our path to mid. And

225
00:12:06,080 --> 00:12:10,557
that's the thing I want to warn
you about. And so this brings me to

226
00:12:10,640 --> 00:12:12,157
something that I'm really observing,

227
00:12:12,240 --> 00:12:15,677
which is this concept of road map zero.

228
00:12:15,760 --> 00:12:19,997
I don't mean like we've gone through
the road map, although some people truly

229
00:12:20,080 --> 00:12:22,237
have road map zero, nothing on the road

230
00:12:22,320 --> 00:12:26,317
map anymore. What I mean is that every

231
00:12:26,400 --> 00:12:29,597
visible feature becomes both plausible

232
00:12:29,680 --> 00:12:33,677
and buildable. You don't have like
an empty backlog. That's not what I'm

233
00:12:33,760 --> 00:12:37,037
talking about. It's just like everything
on your backlog you can kind of do. And

234
00:12:37,120 --> 00:12:38,797
when everything on your backlog you can

235
00:12:38,880 --> 00:12:41,357
kind of do, what is the point of the

236
00:12:41,440 --> 00:12:44,157
backlog? I think buildability and effort

237
00:12:44,240 --> 00:12:46,237
stop being like this meaningful proxy

238
00:12:46,320 --> 00:12:49,357
for what matters and prioritization

239
00:12:49,440 --> 00:12:52,317
again kind of like why PMs exist as

240
00:12:52,400 --> 00:12:56,717
we've practiced it I think stops being
strategy right like we've been doing

241
00:12:56,800 --> 00:12:58,957
these like rice prioritizations but when

242
00:12:59,040 --> 00:13:01,677
like many of those letters mean nothing

243
00:13:01,760 --> 00:13:04,557
anymore especially effort like why are

244
00:13:04,640 --> 00:13:06,797
we still pretending like prioritization

245
00:13:06,880 --> 00:13:08,957
is the right way to think about our

246
00:13:09,040 --> 00:13:14,397
products and this is why I think you know,
road maps are over. They're dead.

247
00:13:14,480 --> 00:13:17,997
It's because I think they're really
dangerous right now. I think an AI

248
00:13:18,080 --> 00:13:20,877
factory plus an old road map will get

249
00:13:20,960 --> 00:13:25,677
you to those three traps at machine
speed. They will get you there fast

250
00:13:25,760 --> 00:13:27,357
because this factory really can't

251
00:13:27,440 --> 00:13:30,397
distinguish a truly consequential bet

252
00:13:30,480 --> 00:13:33,917
from an idea. And AI will make the

253
00:13:34,000 --> 00:13:38,397
consequences of this weak judgment show
up at your front door faster. And so

254
00:13:38,480 --> 00:13:43,117
your bad ideas will become your problems
quicker than ever. And so I don't think

255
00:13:43,200 --> 00:13:44,877
the road map is the right thing. I think

256
00:13:44,960 --> 00:13:47,517
the right thing is to ask yourself what

257
00:13:47,600 --> 00:13:51,117
do I believe strongly enough to go out

258
00:13:51,200 --> 00:13:55,677
and try and prove because
the remaining constraint is not

259
00:13:55,760 --> 00:13:59,917
code. It's not building. It's not
features. It is truth. It is like on the

260
00:14:00,000 --> 00:14:02,637
ground with our customers truth. AI

261
00:14:02,720 --> 00:14:07,437
cannot take an untested assumption into
fact. No matter how many adversarial

262
00:14:07,520 --> 00:14:09,517
reviews you do, and I know you're doing

263
00:14:09,600 --> 00:14:12,717
them, you need real customers. You need

264
00:14:12,800 --> 00:14:15,757
real data. You need repeated tests. And

265
00:14:15,840 --> 00:14:17,837
then you need a very unique point of

266
00:14:17,920 --> 00:14:21,757
view and a very high unique quality bar.

267
00:14:21,840 --> 00:14:25,117
And what AI allows us to do is make

268
00:14:25,200 --> 00:14:28,077
faster contact with reality, which is

269
00:14:28,160 --> 00:14:29,357
great. We all want to make faster

270
00:14:29,440 --> 00:14:32,237
contact with reality. But that means you

271
00:14:32,320 --> 00:14:35,597
have a higher obligation to encounter

272
00:14:35,680 --> 00:14:39,677
reality in your minds. You have to both
get into the market and accept what it

273
00:14:39,760 --> 00:14:43,677
is telling you. And this is going
to feel really retro to people like, you

274
00:14:43,760 --> 00:14:46,797
know, me and they're a little older
because we've been saying outcomes over

275
00:14:46,880 --> 00:14:49,277
outputs all the time, but yet this like

276
00:14:49,360 --> 00:14:51,997
central artifact, the road map still

277
00:14:52,080 --> 00:14:55,597
lists features and dates. And again,

278
00:14:55,680 --> 00:14:57,597
engineering scarcity once made that

279
00:14:57,680 --> 00:15:02,717
workable. But like I think road map
zero makes those features and dates very

280
00:15:02,800 --> 00:15:06,957
dangerous for the reasons that I
outlined. This feature list I say

281
00:15:07,040 --> 00:15:09,277
becomes ammunition for a very powerful

282
00:15:09,360 --> 00:15:11,757
slob cannon. And so what I think we need

283
00:15:11,840 --> 00:15:14,157
to do is move upstream from what we need

284
00:15:14,240 --> 00:15:16,557
to build to what we actually need to

285
00:15:16,640 --> 00:15:18,797
prove. And so here's how I would think

286
00:15:18,880 --> 00:15:21,757
about what we do instead of a road map.

287
00:15:21,840 --> 00:15:26,157
First we build our convictions. So what
is the direction that we want to go and

288
00:15:26,240 --> 00:15:31,037
what do we believe the future looks like?
Not in 3 months, 6 months, 9

289
00:15:31,120 --> 00:15:32,957
months, in in a year, in two years.

290
00:15:33,040 --> 00:15:35,677
Where do we think all of this is going?

291
00:15:35,760 --> 00:15:40,157
I do think you need to determine
what evidence would show you that your

292
00:15:40,240 --> 00:15:43,757
conviction is true. And you need
to define that upfront. What would I need

293
00:15:43,840 --> 00:15:47,757
to see to prove that I'm doing the right
thing? And what would I need to see to

294
00:15:47,840 --> 00:15:51,437
stop? You do need a factory. So I do
love the factory. I'm saying that's

295
00:15:51,520 --> 00:15:54,877
dangerous, but I love something dangerous.
You need your factory. You

296
00:15:54,960 --> 00:15:57,677
need the ability to build quickly to

297
00:15:57,760 --> 00:16:00,477
intersect reality very very fast. So I'm

298
00:16:00,560 --> 00:16:05,117
not saying get rid of it. I'm saying
put it at the right place in the process.

299
00:16:05,200 --> 00:16:07,437
And then you need to allocate. You need

300
00:16:07,520 --> 00:16:09,597
to go through that cycle. And when you

301
00:16:09,680 --> 00:16:14,557
get into reality, you need to allocate
your investment, your conviction, your

302
00:16:14,640 --> 00:16:18,717
effort, your tokens in the right
direction. And what's interesting I

303
00:16:18,800 --> 00:16:22,157
think most about this new world we're
about to move into is I think you need

304
00:16:22,240 --> 00:16:26,237
to have durable convictions
but disposable features. And I was talking

305
00:16:26,320 --> 00:16:27,517
to Mara who is going to come on the

306
00:16:27,600 --> 00:16:29,917
stage later and she's like but Claire

307
00:16:30,000 --> 00:16:34,397
enterprises love feature road maps and

308
00:16:34,480 --> 00:16:38,637
I'm going to talk to some folks at the end
of the day who have changed a lot of

309
00:16:38,720 --> 00:16:43,437
features around on us a lot of products
around us and we as consumers at least

310
00:16:43,520 --> 00:16:47,277
some of us have accepted that level of
churn. And so I think we're going to get

311
00:16:47,360 --> 00:16:52,717
into this place where customers, where
teams are able to shift to, do I bet on

312
00:16:52,800 --> 00:16:55,517
these convictions? Do I bet on this team?
Do I bet on this space? Do I bet

313
00:16:55,600 --> 00:16:59,917
on this vision? And I understand
features need to come and go. They might

314
00:17:00,000 --> 00:17:04,397
be unique to me. They might be unique to
a market. And so what you want to see is

315
00:17:04,480 --> 00:17:09,757
clear progress against your convictions,
but zero ego about your solutions. And

316
00:17:09,840 --> 00:17:12,317
what this would look like, I think
there's two ways this could go. You

317
00:17:12,400 --> 00:17:15,517
could be you have to be stubborn here
with your durable convictions. But

318
00:17:15,600 --> 00:17:17,917
there's good stubborn and bad stubborn.

319
00:17:18,000 --> 00:17:19,997
Good stubborn is you like stay with the

320
00:17:20,080 --> 00:17:23,277
problem. You revise the solution. You

321
00:17:23,360 --> 00:17:27,197
know, you hold true what's going to happen
even if the flavor of the week is

322
00:17:27,280 --> 00:17:31,597
trending. Bad stubborn is because you
have tokens. You just move the goalpost.

323
00:17:31,680 --> 00:17:34,557
You're like, "Ah, we're close. We'll
just ship again and ship again and ship

324
00:17:34,640 --> 00:17:39,197
again." And so I want you to think about
what are my durable convictions? Am I

325
00:17:39,280 --> 00:17:42,157
being good stubborn? Am I being bad
stubborn? And then how do you actually

326
00:17:42,240 --> 00:17:44,717
build into your system the ability to

327
00:17:44,800 --> 00:17:47,837
absorb disposable features and keeping

328
00:17:47,920 --> 00:17:51,517
rigor around quality and learning? And

329
00:17:51,600 --> 00:17:53,757
this is really scary, especially when it

330
00:17:53,840 --> 00:17:55,757
hits customers. But I think we're moving

331
00:17:55,840 --> 00:17:58,797
to a place where not everything we ship

332
00:17:58,880 --> 00:18:03,277
is a promise. And I I don't has anybody
feeling come on. Like has anybody

333
00:18:03,360 --> 00:18:07,197
shipped some thank you right in the
middle? Like you ship things and you're

334
00:18:07,280 --> 00:18:11,997
like this is a hypo this is like truly
an experiment and I might get it wrong

335
00:18:12,080 --> 00:18:15,117
or the market might move underneath
me or the technology might change, you

336
00:18:15,200 --> 00:18:18,877
know, tremendously. And so I do think
we have to think about this idea of like

337
00:18:18,960 --> 00:18:20,957
not every ship is a process, a promise.

338
00:18:21,040 --> 00:18:25,197
And we used to do these road maps and I
would like go hand on heart to go to

339
00:18:25,280 --> 00:18:27,277
market and I'd be like I swear on my

340
00:18:27,360 --> 00:18:29,917
children this feature will ship with

341
00:18:30,000 --> 00:18:32,557
these specific things on this specific

342
00:18:32,640 --> 00:18:37,837
date. You can definitely sell it in the
contract like you know and make these

343
00:18:37,920 --> 00:18:40,877
promises. And now I think we have
to really think how we communicate

344
00:18:40,960 --> 00:18:44,477
internally, how we communicate with
customers. And so I think there's going

345
00:18:44,560 --> 00:18:46,637
to be everything from a probe. This is

346
00:18:46,720 --> 00:18:48,317
like kind of a bet. We're going to

347
00:18:48,400 --> 00:18:52,237
explore a little bit. Um to a more

348
00:18:52,320 --> 00:18:56,717
durable kind of experiment against your
conviction. I think this is something we

349
00:18:56,800 --> 00:19:00,237
really believe in. We're going to stick
with it till we get it right. And then

350
00:19:00,320 --> 00:19:04,477
there are promises. Promises are I have
shipped something. Customers can rely on

351
00:19:04,560 --> 00:19:07,117
it. They can build their customers on it.
We really believe and we'll keep

352
00:19:07,200 --> 00:19:11,277
compounding here. And I think you need
to be honest about what your commitment

353
00:19:11,360 --> 00:19:15,837
is on any feature. I think that's almost
the most um helpful lens on the new road

354
00:19:15,920 --> 00:19:19,997
map is like how strong is our conviction
here and how durable is this promise

355
00:19:20,080 --> 00:19:24,717
versus like how hard is it to build
and what is our estimated impact and while

356
00:19:24,800 --> 00:19:29,037
we do this I you know again you need to
remember code is abundant customer trust

357
00:19:29,120 --> 00:19:32,637
is not. So going back to that product
graph, that feature that I built over

358
00:19:32,720 --> 00:19:34,877
and over and over again until I felt

359
00:19:34,960 --> 00:19:39,677
conviction, it was feature flagged off
because I just knew that as soon as I

360
00:19:39,760 --> 00:19:42,557
put this in front of a customer,
they were going to build on it. They were

361
00:19:42,640 --> 00:19:45,037
going to use it. And if I didn't have
conviction that it was right, I was

362
00:19:45,120 --> 00:19:47,837
really going to burn that customer trust.
And that's something I thought

363
00:19:47,920 --> 00:19:52,957
about really deeply as I was rolling
out this feature. So I do still think you

364
00:19:53,040 --> 00:19:54,717
need a road map, but I think that

365
00:19:54,800 --> 00:19:58,077
roadmap needs to be more about ambition.

366
00:19:58,160 --> 00:20:00,077
I think it needs to be bigger. I don't

367
00:20:00,160 --> 00:20:01,757
want to see onesie twoosies on your road

368
00:20:01,840 --> 00:20:05,277
map. I want to see the future you

369
00:20:05,360 --> 00:20:10,317
believe should exist. I want to see what

370
00:20:10,400 --> 00:20:12,557
evidence shows that we're making

371
00:20:12,640 --> 00:20:17,437
progress. What would prove us wrong?
And then what earns more time tokens and

372
00:20:17,520 --> 00:20:19,437
customer attention. And I want to go

373
00:20:19,520 --> 00:20:21,917
back to ambition because I think the

374
00:20:22,000 --> 00:20:25,997
last 12 to 18 months as you know it was

375
00:20:26,080 --> 00:20:29,757
like you know the year of our claude and
we were shipping everything and PMs were

376
00:20:29,840 --> 00:20:31,757
writing PRs and it was prototypes

377
00:20:31,840 --> 00:20:34,397
everywhere. I really think we're in a

378
00:20:34,480 --> 00:20:37,837
velocity game. We're in a inflect PRs

379
00:20:37,920 --> 00:20:42,237
straight up velocity game. If I can get
prototypes to customers faster they can

380
00:20:42,320 --> 00:20:45,997
give me feedback faster. If I can hand
things to customers or to engineering

381
00:20:46,080 --> 00:20:50,077
faster they can do PRs faster. my agents
will take care of tech debt. My agents

382
00:20:50,160 --> 00:20:53,757
will take care of features. I think
we're really in the last 12 to 18 months

383
00:20:53,840 --> 00:20:58,317
building up our muscle for true
inflected velocity. I don't think that's

384
00:20:58,400 --> 00:21:00,397
the game next year. I don't think that's

385
00:21:00,480 --> 00:21:01,917
the game we should be playing for next

386
00:21:02,000 --> 00:21:05,437
year. I think next year is the ambition

387
00:21:05,520 --> 00:21:07,917
game. I think you should think like what

388
00:21:08,000 --> 00:21:11,197
huge swings can I make? What experiments

389
00:21:11,280 --> 00:21:13,757
can I run in two weeks, three weeks that

390
00:21:13,840 --> 00:21:16,317
would have taken us a year last time?

391
00:21:16,400 --> 00:21:19,197
and how do I build sort of this roadmap

392
00:21:19,280 --> 00:21:23,357
of very very very large investments in

393
00:21:23,440 --> 00:21:26,397
very very ambitious builds because I

394
00:21:26,480 --> 00:21:29,757
think that is more important than raw

395
00:21:29,840 --> 00:21:31,837
feature velocity and I don't think the

396
00:21:31,920 --> 00:21:35,437
road map as it is serves that goal you

397
00:21:35,520 --> 00:21:39,037
know I was talking to somebody and again
going back to OKRs unfortunately not

398
00:21:39,120 --> 00:21:40,477
dead yet and she was like what should

399
00:21:40,560 --> 00:21:43,197
our OKRs be relative to our AI

400
00:21:43,280 --> 00:21:46,397
transformation like PRs revenue per

401
00:21:46,480 --> 00:21:50,397
headcount like what it should be like
how many [snorts] huge experiments are

402
00:21:50,480 --> 00:21:54,077
you running a month like I don't care
what the experiments are I don't know

403
00:21:54,160 --> 00:21:59,437
what they should be like how many big
big swings are you taking a month with

404
00:21:59,520 --> 00:22:03,277
the presumption that most of them won't
work out and I think that is just a very

405
00:22:03,360 --> 00:22:05,837
new way to think about how you build

406
00:22:05,920 --> 00:22:08,077
your road map and a very new way to

407
00:22:08,160 --> 00:22:10,557
think about how you do your product

408
00:22:10,640 --> 00:22:13,597
again it's big on the ambition and very

409
00:22:13,680 --> 00:22:15,917
fuzzy on the specifics

410
00:22:16,000 --> 00:22:21,037
So here is my ask slashgift to you all.

411
00:22:21,120 --> 00:22:23,677
Build your last road mapap. And I do not

412
00:22:23,760 --> 00:22:26,477
mean build your last plan. Of course

413
00:22:26,560 --> 00:22:30,637
not. I just mean like no more lists of

414
00:22:30,720 --> 00:22:33,357
ideas and features in spreadsheets where

415
00:22:33,440 --> 00:22:36,797
you guess impact and where you like put

416
00:22:36,880 --> 00:22:41,197
a date on them and then lock arms
and say we'll never we'll never change

417
00:22:41,280 --> 00:22:45,357
anything because this is how we work.
I just think that day is over. Instead,

418
00:22:45,440 --> 00:22:48,157
what I want you to do is truly raise

419
00:22:48,240 --> 00:22:51,117
your conviction, raise your ambition,

420
00:22:51,200 --> 00:22:55,997
come up with big ideas, define what
success looks like in like a year or

421
00:22:56,080 --> 00:22:58,877
two. You're going to have to make some
big guesses because I can't honestly

422
00:22:58,960 --> 00:23:02,717
tell what's going to happen in a week or
two. I really think you want to get to a

423
00:23:02,800 --> 00:23:04,637
place where you have a factory that can

424
00:23:04,720 --> 00:23:08,477
surprise you. And so, you really want to

425
00:23:08,560 --> 00:23:13,677
be able to discard software and then
hold your factory to a high bar. And

426
00:23:13,760 --> 00:23:17,037
look, you're going to write a lot
of code and then you're going to trash it.

427
00:23:17,120 --> 00:23:19,357
Good. Your customers do not need trash

428
00:23:19,440 --> 00:23:24,717
products. Your AI is going to have
better ideas than your team. Good. We

429
00:23:24,800 --> 00:23:29,757
need good ideas. And you need to hold
that AI to a very, very high bar. And

430
00:23:29,840 --> 00:23:33,837
you're not going to be able to promise
your team, your customers, what next

431
00:23:33,920 --> 00:23:36,157
year looks like. Good. Because it's

432
00:23:36,240 --> 00:23:37,837
probably going to be better than you can

433
00:23:37,920 --> 00:23:41,037
imagine right now. So, this is my ask to

434
00:23:41,120 --> 00:23:43,677
you. write your last road map I promise

435
00:23:43,760 --> 00:23:47,357
you is extremely fun on this other side.

436
00:23:47,440 --> 00:23:51,277
Enjoy the rest of your summit. I will
see you later this afternoon and please

437
00:23:51,360 --> 00:23:55,830
come say hi. [applause]