WEBVTT

1
00:00:00.111 --> 00:00:02.732
You are the co-founder and CEO of Granola.

2
00:00:03.072 --> 00:00:07.052
One of the first new AI apps where people were like, holy shit, this actually works.

3
00:00:07.492 --> 00:00:11.933
Meeting notes are not the end-all, be-all value that everyone's running after.

4
00:00:12.353 --> 00:00:13.154
There's something much bigger.

5
00:00:13.194 --> 00:00:18.075
And I think we are in the very, very early steps of a computing revolution.

6
00:00:18.375 --> 00:00:22.235
What has come thus far will pale in comparison to what will come soon.

7
00:00:22.676 --> 00:00:24.116
Startups are like knife fights.

8
00:00:24.516 --> 00:00:27.116
And I thought startups were just really hard when they weren't working.

9
00:00:27.457 --> 00:00:29.477
Turns out they're really hard even when they're working as well.

10
00:00:43.877 --> 00:00:46.499
Every is the only subscription you need to stay at the edge of AI.

11
00:00:46.800 --> 00:00:52.124
If you care about being on top of the latest models and using the latest tools, you have to subscribe to Every.

12
00:00:52.504 --> 00:00:54.266
We'll help you separate the signal from the noise.

13
00:00:54.586 --> 00:00:56.808
Go to every.to slash subscribe today.

14
00:00:57.328 --> 00:00:58.669
And now, back to the episode.

15
00:00:59.130 --> 00:01:00.171
Chris, welcome to the show.

16
00:01:01.011 --> 00:01:01.412
Hey, Dan.

17
00:01:01.652 --> 00:01:02.212
Great to see you.

18
00:01:03.265 --> 00:01:03.785
Good to see you too.

19
00:01:03.805 --> 00:01:04.566
It's been a while.

20
00:01:05.246 --> 00:01:11.749
So for people who don't know, you are the co-founder and CEO of Granola, one of my favorite AI apps.

21
00:01:11.909 --> 00:01:24.876
One of the apps that I think really kicked off the wave in like, you know, I guess two or three years ago, it was one of the first like new AI apps where people were like, holy shit, this actually works and it's useful.

22
00:01:24.896 --> 00:01:25.917
Yeah.

23
00:01:26.837 --> 00:01:30.679
And since then, you've gone on to, I think the team is about 55.

24
00:01:30.719 --> 00:01:34.440
You've raised at like a $1.5 billion valuation.

25
00:01:35.021 --> 00:01:36.201
You're just killing it.

26
00:01:36.901 --> 00:01:45.505
And we had a really good conversation in December 2024, so almost two and a half years ago, about putting soul into your product and making products that are delightful.

27
00:01:46.066 --> 00:01:56.070
And I'm just super curious to hear what's on your mind, how things are going, and to sort of share out both of our journeys over the last couple of years and think about the future together.

28
00:01:57.113 --> 00:01:58.275
Couldn't be more excited to chat.

29
00:01:58.555 --> 00:02:03.662
I can, I've been following all the different things you've been doing, you know, from a distance.

30
00:02:03.742 --> 00:02:08.708
I'd love to hear, I'd love to hear the latest from the inside on how that's going.

31
00:02:09.109 --> 00:02:11.212
Just to react on one thing you said, you know,

32
00:02:12.247 --> 00:02:13.428
You said we're killing it, right?

33
00:02:13.508 --> 00:02:15.749
And we're very lucky.

34
00:02:15.789 --> 00:02:17.289
Like things are going well.

35
00:02:17.469 --> 00:02:18.730
Something I was unprepared for.

36
00:02:18.770 --> 00:02:24.112
So I did a previous startup and startups are like they're knife fights, right?

37
00:02:24.152 --> 00:02:28.314
They're like really, you know, like they're really, I don't know, maybe life's easy for you.

38
00:02:28.414 --> 00:02:30.535
But for me, it's like it's really hard.

39
00:02:30.736 --> 00:02:33.297
And I thought startups were just really hard when they weren't working.

40
00:02:33.937 --> 00:02:35.837
Turns out they're really hard when they're working as well.

41
00:02:35.877 --> 00:02:40.258
It's just, it's a knife fight when things are going well, or not going well.

42
00:02:40.618 --> 00:02:41.798
And I was a bit unprepared for that.

43
00:02:41.878 --> 00:02:56.041
So it just, I'm very, I had a professor in journalism school who he was basically like, the last thing the world needs is another profile that puts someone up, you know, it's like, oh, this person's so great.

44
00:02:56.101 --> 00:02:58.641
And it's like, no, no, everything's hard today.

45
00:02:58.741 --> 00:02:59.861
Fight day in, day out.

46
00:02:59.941 --> 00:03:01.362
Like that's my mentality.

47
00:03:01.442 --> 00:03:02.742
So just wanted to share that.

48
00:03:03.245 --> 00:03:04.106
I feel you, man.

49
00:03:05.147 --> 00:03:10.873
I definitely feel the same way where I've been doing companies for a long time.

50
00:03:11.193 --> 00:03:20.282
And every is the first one that's like really growing and just really working in this way that feels like it's a real company.

51
00:03:20.302 --> 00:03:24.246
We have 30 people now, you know, and thank you.

52
00:03:25.447 --> 00:03:25.647
And.

53
00:03:27.012 --> 00:03:31.174
when it is working, like a good example, Claude Tag dropped yesterday.

54
00:03:31.194 --> 00:03:32.355
We've been working on a Slack agent.

55
00:03:32.415 --> 00:03:33.695
Claude Tag dropped yesterday.

56
00:03:34.036 --> 00:03:36.577
And so then everyone's looking at you like, what do we do?

57
00:03:36.597 --> 00:03:39.378
And I'm like, we're prepared for this.

58
00:03:40.379 --> 00:03:42.620
We've been thinking, we knew this was going to happen.

59
00:03:43.601 --> 00:03:48.963
And especially in AI, I feel like the ground shifts so quickly.

60
00:03:49.544 --> 00:03:55.507
And I assume for you, like one of the things you're talking to and I'm talking about it, and I'm actually really curious is,

61
00:03:57.461 --> 00:04:02.763
you were the first one to do really great AI meeting notes and meeting transcriptions.

62
00:04:03.823 --> 00:04:07.585
And then immediately everyone just like put meeting notes into their product.

63
00:04:08.125 --> 00:04:12.346
We actually, we have a, we have this product called monologue that has me a meeting notes button, you know?

64
00:04:12.606 --> 00:04:12.906
Oh, cool.

65
00:04:12.926 --> 00:04:13.427
I didn't know that.

66
00:04:13.807 --> 00:04:15.847
So we're like slightly competitive.

67
00:04:16.007 --> 00:04:16.368
Yeah.

68
00:04:16.408 --> 00:04:20.409
I don't, I don't really think of it as being an exact, an exact one-to-one replacement.

69
00:04:20.429 --> 00:04:23.090
I actually have both on my computer, but, but still,

70
00:04:25.047 --> 00:04:27.329
that's probably what you're referring to when you say knife fight.

71
00:04:27.389 --> 00:04:29.311
So tell me about that.

72
00:04:29.351 --> 00:04:29.991
What is that like?

73
00:04:30.031 --> 00:04:30.512
How is that?

74
00:04:31.112 --> 00:04:32.633
Has that actually affected your business at all?

75
00:04:33.094 --> 00:04:35.816
That's actually not what I was referring to as a knife fight.

76
00:04:35.836 --> 00:04:36.616
I mean, that's part of it.

77
00:04:36.777 --> 00:04:47.085
You know, I think it's just, I guess what I was referring to is, I think by definition, a startup is...

78
00:04:47.888 --> 00:04:59.031
you're either fighting for survival, you're like always fighting for survival and it's like if things are going well, you're like fighting to, it's like this big wave and you're like on the surfboard and you're trying desperately not to fall off that surfboard.

79
00:04:59.251 --> 00:05:02.552
And so you're kind of always just beyond the edge of your abilities.

80
00:05:03.272 --> 00:05:06.253
And it's almost like by definition and therefore,

81
00:05:07.705 --> 00:05:12.108
I'm always doing, you know, you always hear this is like, oh, it's like as a founder, you're always doing stuff you don't know how to do.

82
00:05:12.148 --> 00:05:14.570
It's like, oh yeah, actually that, at least that's been my experience.

83
00:05:14.590 --> 00:05:22.436
It's like, I'm always just in the space where like stuff I've never done before, or it's just more than we've ever like bit off in the past.

84
00:05:24.178 --> 00:05:28.561
Yeah, the competition, like here's my view.

85
00:05:28.601 --> 00:05:33.745
And I feel like I've said this, let me think about it.

86
00:05:40.446 --> 00:05:47.371
So I think when I think about this, this is going to sound like one of those bullshit answers around competition, right?

87
00:05:47.391 --> 00:05:52.135
Cause like you, there's always the interview where it's like, Hey, like Microsoft just copied your product.

88
00:05:52.155 --> 00:05:53.455
How do you, you know, how do you feel about it?

89
00:05:53.476 --> 00:05:55.417
And the founder has to like come up with some, some answer.

90
00:05:55.437 --> 00:06:09.107
Like the, the way I, the way I view it is like, I think we are in the very, very early steps of a computing revolution and that there's a,

91
00:06:10.885 --> 00:06:14.846
What has come thus far will pale in comparison to what will come soon.

92
00:06:15.547 --> 00:06:23.469
And I think that meeting notes are not the end-all, be-all value that everyone's running after.

93
00:06:23.489 --> 00:06:25.590
There's something much bigger, and I think it's that...

94
00:06:26.575 --> 00:06:32.516
what interface we use for work and how we work and like, what, what, what does that look like in the AI native world?

95
00:06:33.216 --> 00:06:37.937
And so when I think about competition, I'm just kind of like, what people are fighting for today doesn't matter.

96
00:06:37.977 --> 00:06:39.898
There's this incredible opportunity ahead.

97
00:06:40.018 --> 00:06:44.258
And I think we are, we have a shot at it and a few other companies have a shot at it.

98
00:06:44.458 --> 00:06:44.719
Right.

99
00:06:45.279 --> 00:06:47.359
But it like, that's really what matters.

100
00:06:47.679 --> 00:06:50.540
So in the same way where people are like, Oh, granola is doing really well.

101
00:06:50.560 --> 00:06:51.580
It's like, well, in my mind, it's like,

102
00:06:52.590 --> 00:06:54.713
easy come, easy go in a way, right?

103
00:06:54.733 --> 00:06:58.758
Like meeting notes are useful, but like a lot of things are going to change in the future.

104
00:06:58.778 --> 00:07:03.925
And just because people use us today doesn't mean they're going to use us for that in the future if we're not the best at that next thing.

105
00:07:05.147 --> 00:07:08.351
So yeah, I know that's my zoomed out view.

106
00:07:08.371 --> 00:07:09.112
Like people...

107
00:07:10.653 --> 00:07:14.376
And the other thing to say is we were not the first meeting note taker, right?

108
00:07:14.536 --> 00:07:16.757
Meaning AI note takers have been around for ages.

109
00:07:16.797 --> 00:07:21.120
We were, let's say, heavily inspired by a lot of things that came before.

110
00:07:21.220 --> 00:07:24.983
So I can't be like, oh, how dare you be inspired by what we did?

111
00:07:25.543 --> 00:07:28.365
It's just in an ecosystem.

112
00:07:29.145 --> 00:07:31.407
It has affected the company a lot less than I thought it would.

113
00:07:31.687 --> 00:07:32.488
I think within...

114
00:07:33.188 --> 00:07:36.450
like notion copied or whatever, let's say we're in launch something.

115
00:07:36.530 --> 00:07:38.911
I think heavily inspired by granola opening.

116
00:07:38.951 --> 00:07:40.392
I did zoomed it recently.

117
00:07:41.252 --> 00:07:47.535
So it's like, in some ways it's like, I didn't have to sit with the like nightmare submit scenario of like, what would it be like the day that launches?

118
00:07:47.575 --> 00:07:49.877
It's like, that kind of came and came and happened.

119
00:07:49.917 --> 00:07:50.597
And we're, we're here.

120
00:07:50.617 --> 00:07:50.797
Yeah.

121
00:07:51.437 --> 00:07:51.558
Yeah.

122
00:07:51.578 --> 00:08:01.142
We're still like, it hasn't changed anything from our growth rate or whatnot, but again, we're still a drop in the bucket compared to what we'd like to be in terms of how people use us and how many people use us.

123
00:08:01.162 --> 00:08:01.403
So I,

124
00:08:02.343 --> 00:08:04.845
I think it's just very early days is my view.

125
00:08:05.325 --> 00:08:05.786
That makes sense.

126
00:08:06.226 --> 00:08:13.611
I just got to say, I love getting to talk to you because I feel like you're just so honest about what you're dealing with.

127
00:08:14.031 --> 00:08:14.512
It's the best.

128
00:08:14.572 --> 00:08:15.212
It's so fun.

129
00:08:16.233 --> 00:08:17.514
It's very different.

130
00:08:17.914 --> 00:08:20.116
It's very different from a lot of people.

131
00:08:20.216 --> 00:08:22.597
And I think this is going to be a very fun conversation.

132
00:08:22.877 --> 00:08:28.902
But what I want to go back to is, and I'm happy to share too, what I want to go back to is you said...

133
00:08:30.672 --> 00:08:33.173
you're bouncing on the surfboard and you're trying not to fall off.

134
00:08:34.274 --> 00:08:41.158
And, you know, normally in startups, you're like, you're on the surfboard and you're just like paddling and you're like waiting for the wave.

135
00:08:42.038 --> 00:08:48.642
And then I think for both of us, for you more so than me, but for both of us, you're like, I'm on the big one now.

136
00:08:48.902 --> 00:08:53.785
Like, let's try not to get, let's try not to fall off and just get pounded, you know?

137
00:08:54.225 --> 00:08:54.446
Um, and,

138
00:08:55.606 --> 00:09:04.294
And so what was, what are some of those things recently where you're, you've felt like you've been at the edge of your ability and you're like, I gotta figure this, I gotta figure this out.

139
00:09:04.314 --> 00:09:09.258
This is, this is, it's a, maybe it's a good thing, but generally it's like, holy shit, this is very different.

140
00:09:10.119 --> 00:09:16.504
So a lot of it's kind of, um, uh, my limitations or lack of experience as a founder.

141
00:09:16.664 --> 00:09:19.647
So like my last startup, Socratic, we were 12 people.

142
00:09:20.107 --> 00:09:22.109
So just going from like 12 to we're, we're actually over

143
00:09:22.812 --> 00:09:24.974
We're somewhere between, I don't know, 60 and 70 now.

144
00:09:26.815 --> 00:09:29.417
And so, you know, that's very different.

145
00:09:29.618 --> 00:09:31.059
And I've never done that before, right?

146
00:09:31.119 --> 00:09:33.060
So like organizing humans, that's one.

147
00:09:34.582 --> 00:09:40.847
Something that's challenging, and I wish I had like a really succinct, beautiful answer.

148
00:09:40.867 --> 00:09:41.848
Like we have it all figured out.

149
00:09:41.928 --> 00:09:47.893
But it's in an era where you're talking about product having soul and everything.

150
00:09:48.850 --> 00:09:52.213
product feeling consistent and coherent and tasteful.

151
00:09:52.293 --> 00:09:54.155
And how do you scale that?

152
00:09:54.395 --> 00:10:03.324
Like when you have more people working on that, how do you, how can you both get the upside of having lots of people working on these hard problems, but also having this consistent feeling?

153
00:10:03.344 --> 00:10:04.585
That's a really hard problem.

154
00:10:05.005 --> 00:10:11.932
And I think, I actually, there's so many different interesting topics for us to dive into, but one is like the,

155
00:10:13.137 --> 00:10:19.119
The traditional roles of PM and design and engineering and how you group those people together and what are their responsibilities.

156
00:10:19.159 --> 00:10:30.022
Like, it doesn't, it feels like what has worked in the past may not make sense exactly one-to-one in the world of the future, but I don't actually know how to define roles and split work.

157
00:10:30.202 --> 00:10:38.365
You know, there's something really valuable if you're like, oh, you're the designer, I'm the engineer or vice versa of like, okay, now we know where we collaborate, like where the interface line is.

158
00:10:40.879 --> 00:10:43.400
Do you guys have a philosophy on that at every year?

159
00:10:43.840 --> 00:10:44.521
I don't know.

160
00:10:44.581 --> 00:10:48.582
We've done different things, and it depends on the situation and the product.

161
00:10:50.883 --> 00:10:56.864
One mental model that I have that seems to work, especially for early products, is there are two roles that matter.

162
00:10:57.405 --> 00:11:01.406
One is called the pirate, and one is called the architect.

163
00:11:01.446 --> 00:11:05.287
The pirate's job is to just build as fast as possible to find something valuable.

164
00:11:06.901 --> 00:11:08.262
And it can be vibe code.

165
00:11:08.282 --> 00:11:10.302
You don't have to really think about the architecture, all that kind of stuff.

166
00:11:11.303 --> 00:11:23.047
And the architect's job is to pair with the pirate and think about as the pirate finds something that's valuable, how do we make this into a system that is sustainable, that we understand, that can scale?

167
00:11:24.208 --> 00:11:30.370
And what's interesting is I think you need one pirate per product.

168
00:11:31.688 --> 00:11:35.070
but you can have one architect work on multiple products.

169
00:11:36.211 --> 00:11:51.539
You can have them bounce around because the, the tools now are so good that someone who's a really good programmer, really good architect can just drop into a code base and be like, I'm going to learn this code base in an hour.

170
00:11:51.599 --> 00:11:54.621
Cause I just have cloud or codex or whatever, map it for me.

171
00:11:54.641 --> 00:11:56.542
And then I'll identify the,

172
00:11:58.170 --> 00:12:02.733
the structural pillars that the invariants that need to be true in this product for it to work well.

173
00:12:02.893 --> 00:12:08.237
And then I'll just like send off codecs to do like a 24 hour rewrite and then I'll go do something else.

174
00:12:09.057 --> 00:12:10.859
And I think that that's really interesting.

175
00:12:10.899 --> 00:12:12.220
So that's, that's one divide.

176
00:12:12.240 --> 00:12:21.246
And, and in particular, the, the pirate is this, it's this mix of technical product design, like all, all of those all together.

177
00:12:21.286 --> 00:12:24.228
And the architect is like really mostly focused on the technical side of things.

178
00:12:25.537 --> 00:12:28.638
So you don't have standalone designers then, right?

179
00:12:28.698 --> 00:12:30.798
You only have... We do have standalone designers.

180
00:12:30.838 --> 00:12:31.438
Oh, you do.

181
00:12:31.518 --> 00:12:42.601
Once it gets to be a real product that we're going to support, then yeah, we have a standalone design team that parachutes into each product and helps make it better.

182
00:12:43.401 --> 00:12:49.263
But then on the tech side, for example, we have this Slack agent we've been building called Plus One.

183
00:12:49.283 --> 00:12:53.424
And that's just a bunch of engineers that are all working and are...

184
00:12:54.184 --> 00:12:58.029
to some degree, full stack thinkers, but it's very engineering focused.

185
00:12:58.269 --> 00:13:00.692
So there's no one answer.

186
00:13:00.732 --> 00:13:01.794
I haven't quite figured it out yet.

187
00:13:01.874 --> 00:13:07.000
I do think the pirate architect configuration for new things is really helpful.

188
00:13:07.020 --> 00:13:07.922
But what do you guys do?

189
00:13:09.261 --> 00:13:11.762
Yeah, well, we have something similar, but it's not actually on the role.

190
00:13:12.122 --> 00:13:17.565
So for any kind of new feature, we have this idea of stages.

191
00:13:18.065 --> 00:13:19.926
So like the first stage is like shaping.

192
00:13:20.146 --> 00:13:25.008
So it's, and we're influenced by the, what's the?

193
00:13:25.829 --> 00:13:26.209
Shape up.

194
00:13:27.109 --> 00:13:28.310
Yeah, thank you.

195
00:13:28.770 --> 00:13:29.891
Yeah, by Ryan thing.

196
00:13:30.271 --> 00:13:30.491
Yeah.

197
00:13:31.795 --> 00:13:32.255
Is it?

198
00:13:32.375 --> 00:13:32.996
Yeah, I think so.

199
00:13:33.036 --> 00:13:33.676
Yeah.

200
00:13:33.716 --> 00:13:34.476
Like the Basecamp guys?

201
00:13:34.536 --> 00:13:35.457
Yeah.

202
00:13:35.957 --> 00:13:37.037
It's like loosely inspired by that.

203
00:13:37.057 --> 00:13:41.719
I went back and read it, or at least the first half that was about it.

204
00:13:41.839 --> 00:13:43.260
And it's funny.

205
00:13:43.280 --> 00:13:47.262
It felt very dated and a lot more dated from like the...

206
00:13:48.002 --> 00:13:51.304
the features that they use as examples, for example, that they're talking about building.

207
00:13:51.825 --> 00:13:53.706
But some of the ideas I think are really sound.

208
00:13:53.746 --> 00:14:02.693
So we have this idea of like, there's like a shaping phase where what you want there is you want to explore as many potential solutions as possible.

209
00:14:02.713 --> 00:14:04.054
So you start with like a job to be done.

210
00:14:04.214 --> 00:14:05.115
Like here's a user need.

211
00:14:05.155 --> 00:14:09.378
Here's a job we would like the user to hire us to do.

212
00:14:09.838 --> 00:14:10.058
Right.

213
00:14:10.158 --> 00:14:11.499
And it's like, okay, this is the job.

214
00:14:12.079 --> 00:14:14.179
What are possible shapes of that solution?

215
00:14:14.219 --> 00:14:16.740
And you go and you explore those shapes as quickly as possible.

216
00:14:17.340 --> 00:14:26.443
And then you kind of look at those and you have to be very honest with yourself and say, if we spent time on this, do we think we could actually, would people actually hire us to do this or not?

217
00:14:26.603 --> 00:14:26.823
Right.

218
00:14:26.963 --> 00:14:30.164
And so then, and then we have a next phase, which is like kind of validate.

219
00:14:30.184 --> 00:14:32.164
It's like, okay, we think we can like prove it.

220
00:14:32.545 --> 00:14:37.126
Like for a small, a small number of people, like just prove that you can be useful and that they choose to use you.

221
00:14:37.786 --> 00:14:45.572
And then it's like, if you can prove that, then it's like, okay, let's actually make this good and reliable and scalable and all that.

222
00:14:46.213 --> 00:14:51.397
It still doesn't answer the question of like, what are the roles of the people who work on it?

223
00:14:51.957 --> 00:14:55.600
And yeah, but that's just one of those open questions.

224
00:14:55.940 --> 00:14:58.082
I mean, you have 60, 70 people.

225
00:14:58.102 --> 00:15:00.604
So you have had to answer that at least provisionally.

226
00:15:00.664 --> 00:15:03.346
So what is the structure that you've come to provisionally?

227
00:15:04.403 --> 00:15:24.839
uh we're still we're still messing around with it like we we still have um like we still have titles and i don't know if we should have them like we have you know you're a designer you're a pm you're an engineer right and i always i'm i'm always flirting with the idea of killing those how many pms do you have uh four three

228
00:15:25.327 --> 00:15:26.768
Depends if you count me.

229
00:15:26.848 --> 00:15:27.829
Yeah.

230
00:15:27.869 --> 00:15:28.330
Yeah.

231
00:15:28.410 --> 00:15:28.950
I wouldn't count you.

232
00:15:29.010 --> 00:15:29.371
Yeah.

233
00:15:29.531 --> 00:15:30.152
Okay.

234
00:15:30.392 --> 00:15:30.592
Yeah.

235
00:15:30.632 --> 00:15:31.052
So three.

236
00:15:31.092 --> 00:15:32.033
Yeah.

237
00:15:32.053 --> 00:15:32.754
Interesting.

238
00:15:33.475 --> 00:15:33.875
Yeah.

239
00:15:34.055 --> 00:15:37.659
And like, yeah, I've always thought of the idea of like killing.

240
00:15:37.839 --> 00:15:43.724
Cause it's like, what you really want is on this project right now, someone needs to do certain types of things.

241
00:15:43.744 --> 00:15:45.186
Like someone needs to own a,

242
00:15:46.005 --> 00:15:54.325
like the aesthetic or the user experience and someone needs to own like the building and that could be the same person or those could be two different people, right?

243
00:15:54.385 --> 00:15:55.428
And someone needs to own like,

244
00:15:56.239 --> 00:15:56.979
what are we trying to do here?

245
00:15:56.999 --> 00:15:57.780
And what's the strategy?

246
00:15:57.820 --> 00:16:01.362
And like, what are the next few things that we need to hit in order to hit that?

247
00:16:01.442 --> 00:16:06.024
And, um, but then it's just like a lot of communication overhead if you don't have roles, right.

248
00:16:06.124 --> 00:16:09.005
It's kind of like, okay, well, what exactly are you going to do versus what am I going to do?

249
00:16:10.106 --> 00:16:11.266
Maybe it'll all be agents one day.

250
00:16:11.286 --> 00:16:11.647
Who knows?

251
00:16:11.987 --> 00:16:13.487
We'll, we'll, we'll keep experimenting.

252
00:16:15.348 --> 00:16:15.789
Interesting.

253
00:16:16.069 --> 00:16:24.133
Well, one of the things that you said when we were just about to start recording is you spend most of your time in granola.

254
00:16:24.153 --> 00:16:24.293
Yeah.

255
00:16:25.115 --> 00:16:31.921
And so one of the things you're curious about for me is if I'm playing around and experimenting, like what am I, what am I finding?

256
00:16:32.441 --> 00:16:34.123
So I want to, let's, let's talk about that.

257
00:16:34.143 --> 00:16:35.344
I think there's a lot of meat there.

258
00:16:36.905 --> 00:16:40.308
First of all, tell me about, tell me about that.

259
00:16:40.368 --> 00:16:42.490
Like the, I spend most of my time in granola.

260
00:16:42.510 --> 00:16:51.558
I think that's a, it makes perfect sense, but also as a, as a CEO, as a strategic way of being, tell me about that decision.

261
00:16:51.578 --> 00:16:51.698
Yeah.

262
00:16:52.847 --> 00:17:07.711
So like I mentioned before that I think the big opportunity here is to kind of invent how we're going to work and collaborate with AI to think and do things, right?

263
00:17:07.731 --> 00:17:15.332
Like that's kind of in some shape or form, I think that's what we're all chasing right now in Silicon Valley, right?

264
00:17:16.173 --> 00:17:21.474
And to that end, I think there's a lot of, that's the kind of thing you can like think about.

265
00:17:22.557 --> 00:17:27.880
abstractly, but then you have to be playing with to kind of get a feel for.

266
00:17:28.200 --> 00:17:36.364
And it's hard for me to have the brain space of playing with every new thing that comes up, right?

267
00:17:36.544 --> 00:17:43.707
And in Granola, we have a lot of context and a lot of internal tooling so that I can play around with different ideas.

268
00:17:43.748 --> 00:17:47.109
But they're oftentimes in my little universe as opposed to...

269
00:17:47.569 --> 00:17:50.311
I don't go and set up very complicated workflows.

270
00:17:51.351 --> 00:18:03.621
In Claude, for example, I do it, I play around with it to get a feel, but those aren't load bearing for me because I'm trying to see what makes sense or doesn't make sense to do inside of Granola as we paint that future picture.

271
00:18:04.082 --> 00:18:09.586
It seems like you are at the forefront of experimenting, trying new things, building different workflows for yourself.

272
00:18:09.626 --> 00:18:15.952
So maybe just start off by being like, what's the current state of Dan's tooling stack?

273
00:18:15.972 --> 00:18:16.792
And then we can go from there.

274
00:18:17.333 --> 00:18:18.634
I spend all my time in codecs.

275
00:18:19.500 --> 00:18:19.721
Okay.

276
00:18:20.443 --> 00:18:22.128
Almost as soon as... And how long has that been?

277
00:18:22.549 --> 00:18:23.532
Like, is that a recent thing?

278
00:18:23.572 --> 00:18:25.578
Is that... Like two months, maybe.

279
00:18:25.598 --> 00:18:25.779
Okay.

280
00:18:25.799 --> 00:18:26.260
Two, three months.

281
00:18:26.441 --> 00:18:26.541
Okay.

282
00:18:26.561 --> 00:18:26.742
Yeah.

283
00:18:27.716 --> 00:18:39.380
Um, and then, and I, and I do use the Claude desktop app sometimes, like when Fable was a thing, I, I track all my usage and when Fable was a thing, you can see me like go on the Claude usage.

284
00:18:39.880 --> 00:18:42.041
And then when it, when it left, it went like down.

285
00:18:42.581 --> 00:18:51.905
Uh, so I assume when it's back, cause I think Fable, Fable is just a different beast and I test all the new stuff and I have access to stuff before it comes out and Fable is just different.

286
00:18:52.485 --> 00:18:52.685
Um, yeah.

287
00:18:54.503 --> 00:19:03.727
But I think of, so my overall view is that work is bifurcating into two surfaces.

288
00:19:05.487 --> 00:19:10.429
One is async delegation work that happens in Slack.

289
00:19:12.550 --> 00:19:17.912
That's collaborative, multiplayer, sometimes proactive.

290
00:19:19.465 --> 00:19:20.766
And that happens with Slack bots.

291
00:19:20.786 --> 00:19:22.107
Like we have one called plus one.

292
00:19:22.127 --> 00:19:23.088
There's one called Victor.

293
00:19:23.128 --> 00:19:24.589
There's the Claude's tag.

294
00:19:25.330 --> 00:19:34.257
And that's very much like, okay, every morning it posts in a channel and says, hey, here's all the bugs that got solved.

295
00:19:34.898 --> 00:19:35.979
Here's all the bugs got reported.

296
00:19:36.019 --> 00:19:39.141
Here's all the PRs we pushed.

297
00:19:40.002 --> 00:19:46.808
And then you can add it and say, can you kick off PR for XYZ bug?

298
00:19:46.828 --> 00:19:47.508
And it just does it.

299
00:19:47.689 --> 00:19:47.829
Or...

300
00:19:49.162 --> 00:19:52.023
Can you, we're trying to name this new product.

301
00:19:52.043 --> 00:19:57.025
Can you read all the stuff we've said about what we want the product to be and then propose some names?

302
00:19:57.065 --> 00:20:03.268
And then the team can be like going back and forth with it in the same channel to talk through names.

303
00:20:04.088 --> 00:20:07.270
So you can hand off stuff with it like it's a coworker.

304
00:20:08.170 --> 00:20:11.932
And importantly, because most AI right now is single player.

305
00:20:12.592 --> 00:20:13.192
You can do it.

306
00:20:13.312 --> 00:20:14.273
You do it in public.

307
00:20:15.173 --> 00:20:19.757
And I think that's really interesting in a whole set of things.

308
00:20:22.700 --> 00:20:35.512
However, for serious work, you're always going to get more out of it if you have a better collaboration service between you and the agent.

309
00:20:35.592 --> 00:20:36.633
And Slack is not good for that.

310
00:20:37.892 --> 00:20:50.428
And so the other surface is something like a Codex or a Claude desktop app, where you and the agent are sitting in the same place on the same app together.

311
00:20:51.149 --> 00:20:52.551
And in particular in Codex,

312
00:20:54.280 --> 00:20:58.582
For me, it's like I use almost all of my software in the in-app browser of Codex.

313
00:20:58.782 --> 00:21:07.206
So I'm bringing Codex into my email or into whatever it is in the same way that... What do you mean by that?

314
00:21:07.266 --> 00:21:08.226
Can you describe that?

315
00:21:08.266 --> 00:21:08.926
How do you do that?

316
00:21:09.086 --> 00:21:10.247
I'll explain.

317
00:21:10.267 --> 00:21:17.910
So when you're building something in Codex or Cloud Code and it opens an in-app browser and it lets you see the local host version of the thing and you're iterating on it,

318
00:21:19.311 --> 00:21:32.011
Basically, like all of the patterns, a good mental model for how AI works for me is all the patterns that started with developers or builders eventually make their way into all of knowledge work because we are, what we found is that building a,

319
00:21:35.089 --> 00:21:39.670
building a good enough agent to build any kind of software.

320
00:21:40.630 --> 00:21:44.691
Once it can build any kind of software, it's actually really good for doing any kind of knowledge work that you want.

321
00:21:44.711 --> 00:21:47.311
And that's like the last, that's why Cloud Code went to Cloud Cowork.

322
00:21:47.331 --> 00:21:50.572
That's why, you know, Codex is now all of knowledge work, all that kind of stuff.

323
00:21:51.192 --> 00:22:02.014
So one of those patterns is when you're building something like an app, what you really want is the agent to be in the loop with you.

324
00:22:02.054 --> 00:22:03.074
So as it builds something,

325
00:22:04.326 --> 00:22:07.149
It opens an in-app browser, you can see it, and you can kind of go back and forth.

326
00:22:07.169 --> 00:22:10.992
You can annotate it, you can whatever, and you and the agent are seeing it together.

327
00:22:12.854 --> 00:22:17.038
That pattern also works for any kind of software.

328
00:22:17.218 --> 00:22:18.800
So any kind of website you might visit.

329
00:22:18.860 --> 00:22:20.561
So really simple example.

330
00:22:22.013 --> 00:22:33.861
Um, last night I, uh, I moved departments recently and I needed to change my internet and I just told Codex, go figure that out.

331
00:22:34.361 --> 00:22:37.483
And it knows the address of my old one and my new one, which is actually the same building.

332
00:22:37.503 --> 00:22:38.484
I'm just moving floors.

333
00:22:39.345 --> 00:22:46.049
It just opened up the Verizon website and it's in app browser and it logged in.

334
00:22:46.489 --> 00:22:47.290
I gave it the password.

335
00:22:47.450 --> 00:22:47.930
It logged in.

336
00:22:48.050 --> 00:22:49.952
I gave it the password and then it opened up a chat.

337
00:22:49.972 --> 00:22:50.032
Um,

338
00:22:51.370 --> 00:23:01.016
with a customer service agent who was also probably using AI and it just like chatted with them until it switched my internet.

339
00:23:02.156 --> 00:23:07.579
And it also like, and all these little things come up where it's like, okay, when do you want to move in and out?

340
00:23:07.719 --> 00:23:13.863
And Codex knows because I've been talking to it about like when I moved and when I wanted to switch.

341
00:23:14.703 --> 00:23:15.424
Um, and then it's like.

342
00:23:17.102 --> 00:23:19.245
Here's, here's your old plan doesn't work.

343
00:23:19.305 --> 00:23:20.286
Here's the new plan.

344
00:23:21.767 --> 00:23:25.552
And Codex, Codex knows to be like, I don't want any hidden fees.

345
00:23:25.772 --> 00:23:34.382
It knows how to do the math on, you know, is this a good deal or not and to be able to research it and then to like push them to give me the thing that it thinks I should get.

346
00:23:35.183 --> 00:23:36.485
And I don't have to do any of that.

347
00:23:36.505 --> 00:23:37.366
I'm just like sitting there.

348
00:23:37.746 --> 00:23:43.317
And is that something that you somehow communicated to Codex or is that just like an out of the box?

349
00:23:43.337 --> 00:23:47.544
Like obviously any human would want this when talking to an internet provider.

350
00:23:48.012 --> 00:23:48.792
It's both.

351
00:23:48.912 --> 00:23:50.933
It's easy because it's easy for it.

352
00:23:51.033 --> 00:23:57.055
It knows pretty well when it can make assumptions and when it shouldn't.

353
00:23:58.516 --> 00:24:01.517
And so I can be in the loop on it.

354
00:24:01.797 --> 00:24:04.778
And I also can just go to another thread and be doing something else.

355
00:24:04.818 --> 00:24:06.599
So I'm kind of like flipping back and forth with it.

356
00:24:07.299 --> 00:24:08.800
I think that's something.

357
00:24:09.140 --> 00:24:12.361
But there's a more general thing there where...

358
00:24:13.001 --> 00:24:15.083
I use, this is how I do my email.

359
00:24:15.744 --> 00:24:18.966
It is in the loop with me in Quora, which is our email app.

360
00:24:19.247 --> 00:24:24.091
And also I have this open source thing I experiment, I built called Tend.

361
00:24:24.171 --> 00:24:29.416
But basically it's like my emails are now cards that I read in Codex in the in-app browser.

362
00:24:29.556 --> 00:24:34.941
And then each card, it says, here's what the email said.

363
00:24:35.041 --> 00:24:36.762
Here's what I think the draft should be.

364
00:24:36.782 --> 00:24:38.184
Do you want to send it?

365
00:24:38.404 --> 00:24:39.205
And I just talk to it.

366
00:24:39.857 --> 00:24:40.277
I see.

367
00:24:40.377 --> 00:24:46.802
But this is basically you've vibe-coded your own email client, basically, right?

368
00:24:46.922 --> 00:24:56.670
I vibe-coded an experimental email client that does this, and then we will have a version of Quora, which is our full email client that just has this out of the box.

369
00:24:57.130 --> 00:25:02.674
But you can think of the apps that are structured this way as being...

370
00:25:03.595 --> 00:25:08.899
I've been calling them codex-native apps, but they're basically apps that are responsible for saving the state...

371
00:25:10.241 --> 00:25:13.923
and rendering the UI and you bring your agent to it.

372
00:25:14.424 --> 00:25:26.291
And I think that that is a much, it's a very powerful paradigm because think about, I mean, think about all the work that you have to do for Granola to like make an agent that is good.

373
00:25:27.031 --> 00:25:28.272
And it's really hard.

374
00:25:29.053 --> 00:25:37.618
And you're also competing with the clods and the codexes of the world who are also building the same kind of agent with the same kind of functionality.

375
00:25:39.940 --> 00:25:48.726
And to some degree, I also think that in some cases that's very worthwhile because I do think two agents is better than one.

376
00:25:49.247 --> 00:25:50.208
But ultimately...

377
00:25:51.244 --> 00:25:58.770
I really want to be able to, for example, bring, and I think you're probably seeing this with a lot of Granola users, I really want to bring my agent to Granola.

378
00:25:59.391 --> 00:26:05.816
You already know how AI is changing how everyday work gets done, how much ground you can cover, and how fast a team can scale.

379
00:26:06.056 --> 00:26:10.680
To stay ahead, you need tools that give you a competitive advantage built for this new era.

380
00:26:10.820 --> 00:26:13.162
Adio is the CRM for the agent native world.

381
00:26:13.522 --> 00:26:21.345
It meets you where you work, compounds every customer signal into context, and then acts on it across your pipeline to let you move at unmatched speed and scale.

382
00:26:21.525 --> 00:26:25.647
With agents and automations for every job, Adio orchestrates your work around the clock.

383
00:26:26.007 --> 00:26:27.908
We use it internally at every and we love it.

384
00:26:28.048 --> 00:26:30.209
It's built to handle the scale of your workloads.

385
00:26:30.469 --> 00:26:36.652
It's extensible with an API and MCP access, and is built with infrastructure to keep up with your most ambitious agents.

386
00:26:36.992 --> 00:26:42.153
It's loved by high growth startups like Granola, Modal, Whisperflow, and Every.

387
00:26:42.494 --> 00:26:44.334
Adio runs the work behind EveryWin.

388
00:26:44.674 --> 00:26:46.455
That's Adio, the agentic CRM.

389
00:26:46.695 --> 00:26:50.456
Go to adio.com slash every and get 15% off your first year.

390
00:26:50.816 --> 00:26:52.856
That's adio.com slash every.

391
00:26:53.197 --> 00:26:54.417
And now back to the episode.

392
00:26:55.073 --> 00:26:55.513
What do you think?

393
00:26:55.653 --> 00:26:59.256
Like, yeah, you keep saying that, and I'm really excited about that idea.

394
00:26:59.396 --> 00:27:00.917
I wonder how does that work in practice?

395
00:27:00.937 --> 00:27:05.780
Like if, forget Granola, but like, you know, it's like, let's say you, I don't know, can you do this with Quora right now?

396
00:27:05.800 --> 00:27:11.843
Or like, how, like, how do you build an, how can, how could we build an app that would allow you to bring your own agent?

397
00:27:11.863 --> 00:27:13.504
That's not, you know, MCP.

398
00:27:13.665 --> 00:27:15.065
Cause it feels a little bit different, right?

399
00:27:15.085 --> 00:27:16.486
Cause it's like, you still want to have the UI.

400
00:27:16.726 --> 00:27:16.867
Yeah.

401
00:27:17.756 --> 00:27:18.517
That's the difference.

402
00:27:18.537 --> 00:27:22.019
It's like you and the agent are both looking at the same UI and manipulating the same thing.

403
00:27:22.480 --> 00:27:23.580
MCP is part of it.

404
00:27:24.101 --> 00:27:31.987
But yes, I think MCP is, historically, you're thinking about it as the agent is going to interact with the app and I'm not going to interact with the app.

405
00:27:32.687 --> 00:27:44.056
And I think what this paradigm says is actually you want your agent and the user to be interacting with the app at the same time and be able to trade off.

406
00:27:44.856 --> 00:27:56.001
And so the way that you do that is, one, yeah, you give the agent access to an MCP or a CLI, and then you give the user access to a web interface.

407
00:27:56.721 --> 00:28:03.544
And the user is using the web interface inside of an in-app browser of a codex or of a cloud code.

408
00:28:04.364 --> 00:28:06.965
And then the agent can decide, do I want to...

409
00:28:08.906 --> 00:28:20.892
help the user or collaborate with the user via using the browser use or via using the CLI, which I think CLI is somewhat better, but the trick is the CLI has to then modify the state of the UI.

410
00:28:21.593 --> 00:28:22.353
Right, exactly.

411
00:28:22.433 --> 00:28:23.154
That's the thing.

412
00:28:23.194 --> 00:28:24.695
It's like, that's the thing that needs to be.

413
00:28:25.055 --> 00:28:31.018
So it's almost like if you had, hypothetically, let's say you could have an MCP, right?

414
00:28:31.138 --> 00:28:40.423
That actually had a stateful, like the MCP could modify the state, like real-time state of like a UI that was rendered for a user.

415
00:28:40.783 --> 00:28:46.245
And ideally there was like at least a one-to-one mapping of anything you can do in the UI could also be done through the MCP.

416
00:28:46.285 --> 00:28:48.246
So you have like an agent and a human interface.

417
00:28:48.266 --> 00:28:48.667
Because like...

418
00:28:49.447 --> 00:28:56.612
computer use or like, it's like the web browser use feels like a, it's unnecessarily slow potential, right.

419
00:28:56.652 --> 00:28:59.334
It's like a, it's a hack and not, not, yeah.

420
00:28:59.574 --> 00:28:59.834
Okay.

421
00:29:00.115 --> 00:29:01.035
That's really interesting.

422
00:29:01.476 --> 00:29:01.616
Um,

423
00:29:02.668 --> 00:29:05.029
Yeah, there's something about the presence of the agent.

424
00:29:05.169 --> 00:29:07.210
This is a fascinating topic.

425
00:29:07.290 --> 00:29:08.231
I haven't thought about it very much.

426
00:29:08.291 --> 00:29:12.073
But the embodiment of the agent is also maybe interesting.

427
00:29:12.113 --> 00:29:17.336
It's like when you're in Figma, you can see other people's locations on the canvas.

428
00:29:17.396 --> 00:29:24.139
And there's something there that you maybe would lose if there was just like an MCP and the human – like maybe not.

429
00:29:24.179 --> 00:29:24.760
I don't know how you –

430
00:29:25.620 --> 00:29:37.189
How do you feel like, is it, do you, is the, does the agent just feel like it's just the computer and you know, that's all the same, or does it feel like it's an entity that you're like, you know, you're both looking, you're both looking at the same thing together.

431
00:29:37.670 --> 00:29:43.514
That's that it's more of that it's, it's in the same way that with the Google docs, like we have this Markdown editor called proof that we built.

432
00:29:43.675 --> 00:29:48.779
And like in that one, you can see codex has entered the document and you can see where it is.

433
00:29:49.239 --> 00:29:49.820
Oh, that's cool.

434
00:29:50.100 --> 00:29:50.700
How do you do that?

435
00:29:51.041 --> 00:29:52.822
How do you, how does codex enter the document?

436
00:29:52.962 --> 00:30:06.687
It's just the same kind of CLI MCP type situation, but it has a command that's like, okay, set presence, set location in the document, that kind of thing.

437
00:30:06.707 --> 00:30:08.528
And agents are very good at knowing how to do that.

438
00:30:09.268 --> 00:30:16.431
But there's so many interesting then UI challenges to figure out because an agent can do like a thousand different actions at once.

439
00:30:17.251 --> 00:30:20.713
And so how do you show that to a user

440
00:30:21.637 --> 00:30:21.997
It's interesting.

441
00:30:22.017 --> 00:30:23.478
That's actually not where my brain went.

442
00:30:23.498 --> 00:30:24.338
Because you're right.

443
00:30:24.958 --> 00:30:27.279
If you want to see what the agent's doing, that's one problem.

444
00:30:28.019 --> 00:30:29.299
I was just thinking in an email.

445
00:30:29.339 --> 00:30:30.720
It'd be nice for it to know where your mouse is.

446
00:30:30.860 --> 00:30:33.981
It'd be nice to be able to be like, this part of the email is not great.

447
00:30:34.221 --> 00:30:35.941
And just have that be...

448
00:30:35.981 --> 00:30:37.502
Either you highlight the text or whatnot.

449
00:30:37.682 --> 00:30:40.403
That's not really possible today unless you explicitly...

450
00:30:41.837 --> 00:30:42.577
How would you have to do that?

451
00:30:42.597 --> 00:30:44.798
It would have to be looking at pixels on your screen, right?

452
00:30:44.818 --> 00:30:50.981
There's no native way to do that unless you build it into a UI and you're passing that information to the agent, right?

453
00:30:51.061 --> 00:30:51.581
More or less.

454
00:30:51.701 --> 00:31:00.865
I think the way that I do it is each piece of information that I want to take care of is represented in a card.

455
00:31:01.386 --> 00:31:02.646
And then I can talk to the card.

456
00:31:02.686 --> 00:31:09.109
I can say like, yeah, it doesn't necessarily know where my mouse is, but it knows that I'm talking to this part of the UI.

457
00:31:11.714 --> 00:31:17.419
And it just knows because you say that when you're talking to it or because that is the main thing that's foregrounded in the app?

458
00:31:17.939 --> 00:31:20.261
Whatever I'm looking at is currently focused.

459
00:31:20.522 --> 00:31:22.583
And I know that it's focused.

460
00:31:22.603 --> 00:31:27.207
And then when I'm talking to you, it just knows to put that context next to that thing that I'm looking at.

461
00:31:27.227 --> 00:31:28.669
Okay, that makes sense.

462
00:31:29.229 --> 00:31:31.131
And you're actually speaking, right?

463
00:31:31.151 --> 00:31:33.673
This is like you're using monologue or whisper code or something like that.

464
00:31:35.049 --> 00:31:38.414
That's sort of my view of work and how things are going.

465
00:31:38.555 --> 00:31:42.361
And we've been talking about, okay, the prize is there's a new way of working.

466
00:31:42.862 --> 00:31:48.030
How does that fit into or differ from how you've been thinking about it and your strategy for granola?

467
00:31:48.765 --> 00:31:55.688
I don't know, I took this class once and they were talking about the difference between complex and complicated problems.

468
00:31:55.929 --> 00:31:56.509
Have you heard this?

469
00:31:56.789 --> 00:31:57.249
Yeah, yeah.

470
00:31:57.269 --> 00:32:04.493
And it's like complicated, it doesn't mean it's going to be super hard, but it's kind of knowable versus complex problems.

471
00:32:04.993 --> 00:32:05.573
It's unknowable.

472
00:32:05.593 --> 00:32:08.975
You have to probe the system and kind of see how it reacts and go from there.

473
00:32:09.275 --> 00:32:10.736
And yeah.

474
00:32:11.921 --> 00:32:14.862
So I think we're very much in a complex problem space here.

475
00:32:14.882 --> 00:32:17.862
So I don't, I don't have like an overarching theory of where this is all going.

476
00:32:17.922 --> 00:32:27.404
I think I could make one up that would sound plausible or as plausible as the next person's, but I'm more in the mode of like, what are, what are things that, what are insights or nuggets?

477
00:32:27.444 --> 00:32:28.165
And I'm like, Ooh, okay.

478
00:32:28.305 --> 00:32:31.165
I now, this is like a thing that I've learned.

479
00:32:31.825 --> 00:32:35.606
And now I believe to be true and I'm going to, I'm going to carry that with me.

480
00:32:36.946 --> 00:32:38.447
And so I have, I have a few of those.

481
00:32:38.467 --> 00:32:39.247
Like one is, um,

482
00:32:40.878 --> 00:32:48.760
seems like one fundamental design challenge in like this agentic world is the time traveling problem.

483
00:32:49.140 --> 00:32:52.061
Basically like agents take time to do things.

484
00:32:52.501 --> 00:33:05.084
And therefore like when you, you're talking about Slack as an interface for this, which is like the async delegation problem, I think you said before, which is basically like the moment in which you kick off a task and the moment in which you're reviewing the task are disjoint.

485
00:33:05.244 --> 00:33:06.564
And then you need to get all that

486
00:33:11.237 --> 00:33:16.384
I think you'll see a lot of the interfaces that evolve to be highly optimized for that.

487
00:33:16.664 --> 00:33:19.468
And I haven't seen anything that's great, to be honest.

488
00:33:19.488 --> 00:33:25.676
Like, I don't know, like the conductor, it's like, oh, I've got all these different like chat threads and I'm jumping between them when it's kind of ready.

489
00:33:25.856 --> 00:33:27.879
Like, it feels like there'll be...

490
00:33:29.287 --> 00:33:30.088
I don't know what they look like.

491
00:33:30.108 --> 00:33:32.870
There'll be evolutions beyond that that are coming.

492
00:33:34.492 --> 00:33:41.018
Another one that's interesting, another way to get around this it takes a while for an agent to do thing problem is...

493
00:33:43.400 --> 00:33:44.982
pre-process a bunch of stuff.

494
00:33:45.002 --> 00:34:00.802
Like if like, so in like Granola, like one thing that we figured out is like, if I ask Granola to do something and I have to wait 20 seconds, basically humans will rarely wait for 20 seconds if they're in the like back-to-back meetings, chaotic, crazy workday, which is kind of the user that we're thinking about.

495
00:34:01.143 --> 00:34:01.283
Um,

496
00:34:02.204 --> 00:34:13.892
But if granola kind of thinks about something you might want and then pre-generates it, and then at some point it might be like, oh, here, in case you want this, it's already here and all you have to do is click on it.

497
00:34:13.912 --> 00:34:16.854
That feels completely different.

498
00:34:17.694 --> 00:34:19.876
And so we have some examples there.

499
00:34:20.996 --> 00:34:29.862
We launched this thing recently where granola will try to generate a brief for you before a meeting where it's basically, here's this person you're meeting with, here's the context of like,

500
00:34:31.100 --> 00:34:39.243
like who the person is, if it's a first meeting or, or what you guys talked about last time, if it's a, if it's another meeting and this one, you essentially have to regenerate it.

501
00:34:39.343 --> 00:34:46.326
Cause it's, it's, it's useful when you're running two minutes late to a meeting and you're like, wait, who the heck is this person I'm talking to again?

502
00:34:46.346 --> 00:34:48.867
Like, and, and like, that's the, that's the critical moment.

503
00:34:48.887 --> 00:34:54.409
So you really only have like a 15 second window where it needs to be there or it's kind of useless.

504
00:34:55.709 --> 00:34:57.010
So we, we actually pre-generate,

505
00:34:58.173 --> 00:35:22.035
like millions of like we regenerate a silly amount of these for a small percentage of them actually being like opened in the belief that when you do open it though you really really appreciate it because it's like right it's exactly what you need in the moment of need it's like an interesting it's like a it's a it's an interesting process or trade-off there especially from like a cost perspective because it's like a you know we're

506
00:35:22.736 --> 00:35:29.963
Like hours and I don't know, hundreds of thousands of hours of like agents reasoning about things that may or may not ever get to see the light of day.

507
00:35:31.244 --> 00:35:33.146
And how do you how are you measuring that?

508
00:35:33.386 --> 00:35:34.927
Like the whether that's worth it or not?

509
00:35:37.069 --> 00:35:37.890
It's an open question.

510
00:35:38.711 --> 00:35:39.571
I don't know how to measure it.

511
00:35:40.401 --> 00:35:40.801
Just yet.

512
00:35:41.022 --> 00:35:43.043
What I'm trying to figure out is that we don't have a good metric between that.

513
00:35:43.083 --> 00:35:45.985
Like there's, there's use, like if something's not used at all, it's obviously not good.

514
00:35:46.066 --> 00:35:46.306
Right.

515
00:35:46.486 --> 00:35:52.731
But then there's like, if something is used a little bit or, you know, a decent amount,

516
00:35:54.142 --> 00:35:57.583
but it's really load bearing when it's used and that's still very valuable.

517
00:35:57.623 --> 00:36:00.044
I'm trying to figure out how to, how to measure what that is.

518
00:36:00.264 --> 00:36:00.404
Right.

519
00:36:00.424 --> 00:36:14.449
Like I was at a, I was at a founders conference the other day and like four founders came up to me and talked about this pre-meeting brief feature, which like surprised me because it's not, you know, it, it's not used as much as like proportionally.

520
00:36:14.549 --> 00:36:16.469
That's not what I would have expected, you know?

521
00:36:17.630 --> 00:36:20.031
So yeah, we have this, we have this,

522
00:36:22.116 --> 00:36:31.110
analogy, metaphor, I always forget which of those who it is, inside of granola, of like how we want the granola product to feel.

523
00:36:31.350 --> 00:36:31.771
And it's...

524
00:36:32.710 --> 00:36:33.490
It's like a handrail.

525
00:36:33.570 --> 00:36:37.151
So if you imagine like stairs, like all stairs have handrails, right?

526
00:36:37.171 --> 00:36:37.851
It's so good.

527
00:36:38.731 --> 00:36:39.051
Yeah.

528
00:36:39.132 --> 00:36:41.232
It's like basically it's like you never notice a handrail.

529
00:36:41.352 --> 00:36:42.232
It's like invisible.

530
00:36:43.052 --> 00:36:52.935
But like that moment you trip, like your hand like shoots out and it needs to be right there and it needs to be load bearing and it makes stairs way safer.

531
00:36:53.935 --> 00:36:56.536
And that's how we think about granola, right?

532
00:36:56.556 --> 00:36:58.376
We kind of want to get out of the way until you really need us.

533
00:37:03.358 --> 00:37:05.441
metrics or frameworks to like measure.

534
00:37:05.661 --> 00:37:14.332
One thing we would thought is like, if we just take it away, like if we take it away from a number of users, like how much do they shout or like what kind of, what's the impact of us doing that?

535
00:37:14.392 --> 00:37:16.535
But even that has, it's not, it's not the best way.

536
00:37:17.240 --> 00:37:22.664
Or maybe making like seeing if they will click to like generate, click to generate as opposed to here it is.

537
00:37:22.984 --> 00:37:24.585
And it's already generated.

538
00:37:24.645 --> 00:37:27.747
But, you know, would you take the proactive action to do it?

539
00:37:27.867 --> 00:37:29.549
Of doing it and then it's and then it's there.

540
00:37:30.109 --> 00:37:30.289
Yeah.

541
00:37:30.549 --> 00:37:30.789
Yeah.

542
00:37:31.250 --> 00:37:34.072
But then once and then it's like, OK, they do.

543
00:37:34.172 --> 00:37:36.073
Then are we always going to make them click?

544
00:37:36.113 --> 00:37:36.493
I don't know.

545
00:37:36.553 --> 00:37:39.095
It's like we could do it as a test, I guess.

546
00:37:39.175 --> 00:37:40.436
It's yeah, it's a tricky one.

547
00:37:40.836 --> 00:37:41.397
That is tricky.

548
00:37:42.323 --> 00:37:55.290
What, well, then how do you think about also what is Granola's role versus the codexes and the co-works of the world?

549
00:37:56.711 --> 00:38:08.577
And what do you see as being places where you want to integrate and make yourself available to what they do versus places where you're like, I want to own this.

550
00:38:08.597 --> 00:38:10.578
This is like something that we can uniquely do well.

551
00:38:11.820 --> 00:38:12.320
Yeah.

552
00:38:12.981 --> 00:38:27.070
So my view here is that it's like a new, again, it's like this new era of computing where for the first time computers can make sense of our context and you can do amazing things with that.

553
00:38:27.150 --> 00:38:27.390
Right.

554
00:38:27.610 --> 00:38:29.751
And to that end, the question,

555
00:38:31.790 --> 00:38:36.737
having the right context is really critical and you being able to get value out of AI.

556
00:38:37.217 --> 00:38:45.829
So my belief is that the context you have in Granola, it's like we should make that as available to people

557
00:38:47.092 --> 00:38:49.012
as like, however you want to use it.

558
00:38:49.292 --> 00:39:00.814
So if you want to use it in codex, if you want to use it in your personal agent, if you want to use like whatever you can do that, if it can make your life better to use Granola Context, you should, and we should make that really easy.

559
00:39:01.154 --> 00:39:11.956
And there's a lot, you know, our API or MCP is gonna get a lot better over the next couple months because like, that's like a first class goal of ours.

560
00:39:13.837 --> 00:39:15.237
I think there are,

561
00:39:16.521 --> 00:39:28.245
a number of jobs to be done, pain points, use cases, that we should just be 5X better than anybody else in the world at.

562
00:39:28.846 --> 00:39:35.008
And the obvious ones are all related to meetings, right?

563
00:39:35.068 --> 00:39:38.029
And it's just one of those things where, and maybe you feel differently.

564
00:39:38.049 --> 00:39:39.509
Maybe you're more AI-pilled than I am.

565
00:39:39.949 --> 00:39:43.331
Like, I think that the models are going to keep getting better, but I think that...

566
00:39:44.477 --> 00:39:58.063
if Granola just cares about a few things way more than anybody else, and we optimize the hell out of those things, they'll just be a better experience, especially if they're like tied to a UI that you have to use in vivo, like during a meeting, for example.

567
00:39:59.704 --> 00:40:01.665
And so that's like our current strategy.

568
00:40:01.685 --> 00:40:06.747
It's basically be best in the world in anything that's like meeting adjacent and then make like the...

569
00:40:07.942 --> 00:40:12.328
be the best way to capture context to power any agents you have.

570
00:40:12.889 --> 00:40:15.834
And that sounds, I can give you an example.

571
00:40:15.854 --> 00:40:20.461
There's like, I was talking to a user and they were doing something really clever.

572
00:40:20.541 --> 00:40:20.922
She was...

573
00:40:22.708 --> 00:40:30.036
She worked in sales and they basically used, I think it was Claude, to generate a microsite before they would talk to a potential client.

574
00:40:30.657 --> 00:40:32.559
And the microsite was, it was really impressive.

575
00:40:32.579 --> 00:40:35.161
They had this template of what that microsite would look like.

576
00:40:35.662 --> 00:40:38.505
And then given a new potential customer...

577
00:40:38.965 --> 00:40:43.128
They would, Claude would like style that template to look like the customer's website.

578
00:40:43.388 --> 00:40:52.475
And then they would connect the Granola MCP and just pull all the context from Granola and like fill out all the data and the fields and the widgets based on that data.

579
00:40:53.116 --> 00:40:58.400
And that's just the kind of thing that we would never try to be best of the world at and wouldn't make sense to, right?

580
00:40:59.400 --> 00:41:01.542
But she would also be...

581
00:41:03.689 --> 00:41:09.871
In her case, she was like writing a lot of emails in Granola right after meetings because like follow-up emails in Granola made a lot of sense for her.

582
00:41:09.931 --> 00:41:13.132
And that's the kind of thing where it's like, maybe we don't want to write every email.

583
00:41:13.152 --> 00:41:15.313
We don't want to be your email inbox.

584
00:41:15.673 --> 00:41:18.894
But there might be a type of email that you want to send right after a meeting that is like...

585
00:41:19.897 --> 00:41:23.519
you know, extremely good on certain dimensions that we could be the best in the world at.

586
00:41:23.619 --> 00:41:25.140
And like, it makes sense to do both.

587
00:41:25.460 --> 00:41:26.961
I don't think it's a one or other.

588
00:41:28.201 --> 00:41:31.263
I was undecided about this two years ago when we started Granola.

589
00:41:31.323 --> 00:41:35.645
Now I have strong conviction, but it kind of took me a while to like figure that out.

590
00:41:36.721 --> 00:41:37.321
That's interesting.

591
00:41:37.742 --> 00:41:38.102
Yeah.

592
00:41:38.843 --> 00:41:48.049
I think from a user perspective, and I may not be your core user, and that's actually really interesting to me to learn that.

593
00:41:48.890 --> 00:41:59.318
The thesis that we're building every round and one of my convictions is what builders do now, people who want to figure out these general purpose tools and build these workflows or whatever...

594
00:42:00.158 --> 00:42:24.390
regular people are going to do in a few years once they're more built in once it's built into the handrails of these products um and so what i do is i just like play around with stuff and try to figure out oh this is what i use codex and slack agents and whatever for and eventually that will make its way into maybe stuff that we build but stuff that the labs build or stuff that other people build and we get to see it first can i ask a question about that because i think um

595
00:42:26.605 --> 00:42:28.748
Is the theory that...

596
00:42:29.388 --> 00:42:36.936
So the stuff that you are trying to automate with AI or the interactions you're trying to build, do you think that other people will also be...

597
00:42:38.287 --> 00:42:44.370
building their own, like, more, like, less AI forward people will be building their version of those automations?

598
00:42:44.390 --> 00:42:46.431
Or do you think those will be productized?

599
00:42:46.711 --> 00:42:47.371
Productized.

600
00:42:47.791 --> 00:42:48.271
Yeah, okay.

601
00:42:48.491 --> 00:42:49.192
I have the same view.

602
00:42:49.492 --> 00:42:52.573
So, yeah, I think you're the guinea pigs, right?

603
00:42:52.593 --> 00:42:56.295
So basically, like, yeah, and we have tons of guinea pigs using granola, right?

604
00:42:56.315 --> 00:42:58.276
So it's like, like, my view is basically...

605
00:42:59.320 --> 00:43:06.865
We should just look at what are the most common and cutting edge ways in which people are using our API and MCP to do crazy stuff.

606
00:43:07.505 --> 00:43:13.569
And then figure out which of those do we think we could be the best in the world at and a lot of users would benefit from.

607
00:43:14.029 --> 00:43:18.252
And then build a really beautiful seamless product experience that...

608
00:43:19.293 --> 00:43:32.932
I ran into one of the founders of Hugging Face the other day at a conference, and he was like, man, I had built this super convoluted, complicated pre-meeting brief workflow in Cloud Code.

609
00:43:34.113 --> 00:43:40.098
And I'd been working on it for months and I just turned it off because yours was better.

610
00:43:40.278 --> 00:43:43.080
And to me, that was like, that's exactly what I want.

611
00:43:43.120 --> 00:43:51.847
And we can't do this for everything, but that's exactly what I want because we're just going to care more than any individual should ever care about any one workflow is kind of how I think about it.

612
00:43:51.887 --> 00:43:52.988
I think that's exactly right.

613
00:43:55.430 --> 00:43:57.311
Okay, so we're definitely aligned there.

614
00:43:58.412 --> 00:44:00.354
My current feeling is...

615
00:44:03.397 --> 00:44:17.561
In general, when I, and this is, again, this is just for me and this may change, but when I see Granola, like the UI come up, it's usually a mistake.

616
00:44:18.702 --> 00:44:19.322
You don't want to see it?

617
00:44:19.722 --> 00:44:20.342
I don't want to see it.

618
00:44:20.482 --> 00:44:21.283
I mean, it's fine.

619
00:44:21.403 --> 00:44:22.543
The recording thing is fine.

620
00:44:23.083 --> 00:44:31.366
But in general, the way that I want to access it is going to be it pushing something into Codex, which is my work surface.

621
00:44:32.321 --> 00:44:33.902
or a codex pulling something out of it.

622
00:44:35.524 --> 00:44:40.108
And I think this focus on meetings is so right.

623
00:44:40.809 --> 00:44:47.534
And it's actually such a hard and deep problem to just get that, like the context of that right.

624
00:44:47.775 --> 00:44:54.501
So for example, I mean, you know this better than I, but I'm just thinking about when you do a transcript

625
00:44:56.247 --> 00:44:57.470
and I say the name Chris.

626
00:44:58.292 --> 00:44:59.955
Which Chris are we talking about, right?

627
00:45:01.478 --> 00:45:06.649
And so I think of the job of software companies in an AI world

628
00:45:08.070 --> 00:45:15.317
AI is this really super flexible thing, but it's only flexible if it's given the right structure and the right context within which to be flexible.

629
00:45:16.398 --> 00:45:26.929
So if we're using the human body as an analogy, AI is like the ligaments and the muscles, and then software has to be the bones.

630
00:45:29.491 --> 00:45:40.058
And the bones sort of set, like give it form and a structure, but then the AI can kind of like work around that to do anything.

631
00:45:40.899 --> 00:45:48.945
And there's so much, even for example, again, in a meaning transcript, I might say,

632
00:45:51.259 --> 00:45:52.061
That's a good idea.

633
00:45:52.883 --> 00:45:59.838
But the way I say that means, says a lot about what I actually think that is not captured in just a transcript.

634
00:46:01.545 --> 00:46:25.378
And I want Granola to be like, I have a, I have this, this automation that runs that turns all the slacks and all the meetings that I was not in into cards that I just go through and I can see things that happened in the company, but the transcripts are a often wrong and B they, they capture none of the nuance of how and why things were said.

635
00:46:26.418 --> 00:46:27.459
And I think that,

636
00:46:28.829 --> 00:46:31.410
if granola did that for me, it would be the most valuable thing in the world.

637
00:46:31.430 --> 00:46:34.291
And I wouldn't necessarily have to look at the product.

638
00:46:34.751 --> 00:46:41.813
And maybe, maybe like for, for a more normie user who doesn't want to be in codex all the time.

639
00:46:41.873 --> 00:46:44.854
Although I do think that that's, that's going to be more and more of the case.

640
00:46:45.074 --> 00:46:52.696
Like you still want to have the UI, but there's all this background context processing that I would look to you guys to do that.

641
00:46:52.797 --> 00:46:53.937
I don't really want to do.

642
00:46:54.737 --> 00:46:54.957
Yeah.

643
00:46:55.177 --> 00:46:55.757
Yeah.

644
00:46:55.817 --> 00:46:56.598
No, that makes total sense.

645
00:46:56.638 --> 00:46:57.098
I think, um,

646
00:46:58.050 --> 00:47:02.453
I guess the transcript is one way to capture the context from a meeting, right?

647
00:47:02.493 --> 00:47:04.314
It is the simplest way.

648
00:47:04.555 --> 00:47:11.179
And those LMs are weirdly good at making sense of garbled transcripts and doing something useful with that.

649
00:47:11.640 --> 00:47:11.980
I think...

650
00:47:13.245 --> 00:47:15.226
but what's the actual job to be done here?

651
00:47:15.487 --> 00:47:21.431
It's actually like, what are the insights or what are the decisions or what are the actions I need to take on top of this context?

652
00:47:21.591 --> 00:47:32.038
And that there's this layer of intelligence or interpretation, which is like the, the disambiguation, like which Sam did I mean is like a, is like a great example.

653
00:47:32.918 --> 00:47:37.221
We have this, so we have this, I'll give you like, there are challenges.

654
00:47:37.441 --> 00:47:38.442
It's such an interesting space.

655
00:47:38.462 --> 00:47:39.723
So here's a challenge we have, right?

656
00:47:39.763 --> 00:47:40.023
We,

657
00:47:44.665 --> 00:47:47.907
When a user starts using Granola, they use Granola a lot, right?

658
00:47:48.167 --> 00:47:50.988
And therefore, we have a lot of context about that user, right?

659
00:47:51.128 --> 00:48:10.156
And we can, in the background, and we have this, but we don't use it yet, and I'll explain why, generate a pretty high-fidelity picture of who you are, what you're working on, what you're trying to achieve, what are your challenges, who you're working with on different things on.

660
00:48:11.017 --> 00:48:11.177
And...

661
00:48:12.567 --> 00:48:18.511
Let's just for argument state, let's say this is like a two-page summary of who Dan is today, right?

662
00:48:18.672 --> 00:48:20.373
The state of Dan today, and it changes every day.

663
00:48:22.054 --> 00:48:29.840
And we can generate better notes if we pass that context as part of our note generation pipeline.

664
00:48:32.242 --> 00:48:36.164
But now those notes might include information that didn't come from the meeting, right?

665
00:48:36.285 --> 00:48:38.646
So when you said, it's like, oh, that's interesting.

666
00:48:38.666 --> 00:48:39.307
Or what was your example?

667
00:48:39.347 --> 00:48:40.408
It's like, oh, I like that idea.

668
00:48:41.485 --> 00:48:43.448
And we might be like, Dan doesn't actually like that idea.

669
00:48:43.488 --> 00:48:48.517
He thinks that's a terrible idea because we know he hates that general class of idea because he said that five other times in five other meetings.

670
00:48:49.518 --> 00:48:52.523
And so then there's this interesting question of who's the audience, right?

671
00:48:52.544 --> 00:48:54.006
And today, a lot of people still...

672
00:48:55.035 --> 00:48:59.378
they'll use the notes for themselves, but they'll also, there's this mixed use with also share it with other people.

673
00:48:59.658 --> 00:49:00.278
Right.

674
00:49:00.678 --> 00:49:06.042
Whereas like the perfect notes for you are actually probably a little bit different or the perfect, like notes are just one.

675
00:49:07.543 --> 00:49:09.464
It's just one way to represent that information.

676
00:49:09.484 --> 00:49:15.768
You know, there's like other ways and it's like, what level of intelligence or interpretation would you, do you want granola to do?

677
00:49:15.788 --> 00:49:17.969
But yeah, I, I totally hear you.

678
00:49:18.149 --> 00:49:19.090
This is why I think it's,

679
00:49:20.070 --> 00:49:22.311
It's infancy days for us.

680
00:49:23.012 --> 00:49:26.133
Or honestly, you seemed really stressed in that meeting.

681
00:49:26.153 --> 00:49:27.074
Do you want to talk about it?

682
00:49:27.514 --> 00:49:31.896
Would be really interesting.

683
00:49:32.016 --> 00:49:41.061
Maybe people would be freaked out by that, but also one of the things I use Granola for a lot is management stuff and being like, how did I do there?

684
00:49:41.562 --> 00:49:45.604
Or someone on my team had a difficult conversation with someone on their team,

685
00:49:46.004 --> 00:49:46.985
And they're talking to me about it.

686
00:49:47.005 --> 00:49:50.049
And I'm like, let's talk to Claude about the meeting transcript so we can decide what to do.

687
00:49:50.610 --> 00:50:00.902
All that kind of stuff is in order to do it well, I need not just the transcript, but there was a long pause after XYZ said this thing.

688
00:50:02.089 --> 00:50:05.270
No one said anything for three minutes that you don't see that in the transcript.

689
00:50:05.290 --> 00:50:06.250
No.

690
00:50:07.090 --> 00:50:07.571
Yeah.

691
00:50:07.651 --> 00:50:16.653
Stuff like that is just, there's so much like richness there to be had that granola to me, I want granola to do for me.

692
00:50:17.614 --> 00:50:17.814
Yeah.

693
00:50:17.914 --> 00:50:19.034
That makes, that makes total sense.

694
00:50:20.034 --> 00:50:20.214
Yeah.

695
00:50:20.274 --> 00:50:23.275
I also think I'm curious.

696
00:50:23.315 --> 00:50:25.116
What are all the, the main way?

697
00:50:25.136 --> 00:50:26.936
Cause he was like, I don't want to see the granola UI.

698
00:50:26.956 --> 00:50:29.077
I want to like pull that in codex or whatnot.

699
00:50:29.177 --> 00:50:29.357
And yeah,

700
00:50:31.041 --> 00:50:40.305
I also wonder, okay, what are the main workflows or reasons why people are pulling transcripts or meeting notes into their agents?

701
00:50:40.886 --> 00:50:46.168
And which of those, if you were to optimize around, you would do a better job at?

702
00:50:46.188 --> 00:50:49.790
And which of those are like the long tail of like, you know, give people power and they'll do a better job?

703
00:50:50.610 --> 00:50:53.372
a better job is also an interesting question for us.

704
00:50:53.553 --> 00:50:53.713
Yeah.

705
00:50:54.373 --> 00:51:00.838
What is on your mind for the next like six to 12 months as, as you've got like, okay, you saw Fable come out.

706
00:51:02.059 --> 00:51:03.280
Does that affect your roadmap?

707
00:51:03.620 --> 00:51:06.522
Does that affect how you're hiring or how you're building?

708
00:51:08.183 --> 00:51:08.323
Yeah.

709
00:51:08.343 --> 00:51:08.824
Tell me about that.

710
00:51:09.377 --> 00:51:17.404
The order of operations, the questions that go through my mind is first, like, what does, how does this change how we build internally and how do we structure our teams?

711
00:51:17.424 --> 00:51:18.565
Like, that's the first question.

712
00:51:19.146 --> 00:51:22.989
I think our roadmap is, and this is the kind of thing where it's like.

713
00:51:23.397 --> 00:51:25.678
you can have one plan in the abstract and then you actually play with something.

714
00:51:25.878 --> 00:51:26.578
You should adjust it.

715
00:51:26.678 --> 00:51:31.880
And I unfortunately didn't play with Fable enough before it got yanked.

716
00:51:32.420 --> 00:51:37.322
But like the roadmap is more, it's kind of baked into the roadmap a little bit more, which is like, okay, models will get better.

717
00:51:37.762 --> 00:51:39.683
Here's the general strategic direction.

718
00:51:39.723 --> 00:51:49.026
And that's more like, I think our strategy is more dependent on what people use us for and what are the jobs to be done or use cases around that that we can tackle.

719
00:51:49.046 --> 00:51:50.567
And model intelligence doesn't.

720
00:51:51.367 --> 00:51:54.009
affect that directly too much, if that makes sense.

721
00:51:54.149 --> 00:52:00.434
Whereas model capabilities do affect how many people do we staff on a pod?

722
00:52:01.795 --> 00:52:03.497
How do we define goals there?

723
00:52:04.017 --> 00:52:07.160
Should we be approaching things completely differently?

724
00:52:08.000 --> 00:52:11.703
So yeah, that's the thing that's on my mind for sure.

725
00:52:12.844 --> 00:52:17.768
The other thing is I think...

726
00:52:20.616 --> 00:52:26.058
So maybe this touches on a little bit, like it's the stuff that you're using codecs for.

727
00:52:26.959 --> 00:52:38.284
I think there's one or multiple canonical UIs that haven't been invented yet that are, like right now, so Granola, you have meeting notes, right?

728
00:52:38.424 --> 00:52:42.946
And you can use meeting notes or transcripts or context for lots of things.

729
00:52:43.985 --> 00:52:48.706
usually when you're trying to make use of that context, it's not per meeting.

730
00:52:48.946 --> 00:52:51.147
It's usually you have a set of contexts.

731
00:52:51.647 --> 00:52:54.928
In our case, let's just say it's like all the meetings you had today or all the meetings you had this week.

732
00:52:54.988 --> 00:52:59.649
And maybe it's actually all the meetings and all the Slack messages and all the emails, but let's just keep it to meetings for a second.

733
00:53:01.230 --> 00:53:06.551
At that level of altitude, what is the right UI and interaction design

734
00:53:10.417 --> 00:53:12.878
to manipulate or work with that context.

735
00:53:12.958 --> 00:53:15.460
I don't think anyone has figured that out.

736
00:53:15.560 --> 00:53:18.481
And maybe you're cooking on something, Dan, or maybe you'll figure it out.

737
00:53:18.541 --> 00:53:23.864
But like, when I look around, I'm like, I've seen very little actually.

738
00:53:24.224 --> 00:53:25.645
And, you know,

739
00:53:26.682 --> 00:53:33.986
coming up with new UI paradigms is, it's like very hard and it's very rare for new ones to stick around.

740
00:53:34.006 --> 00:53:36.787
I think they're more discovered than invented, if that makes sense.

741
00:53:37.267 --> 00:53:39.949
But there's a very clear gap that I see there.

742
00:53:39.969 --> 00:53:48.553
Like we have like a chat UI and chats, chat threads are like super valuable, like actually surprisingly versatile and powerful when you're doing one thing.

743
00:53:48.813 --> 00:53:55.597
But when you're actually trying to do multiple things across varied contexts, like I don't think we have the right,

744
00:53:56.297 --> 00:53:57.898
the right metaphor there.

745
00:53:58.658 --> 00:54:00.859
So that's something that's very much on my mind.

746
00:54:02.000 --> 00:54:09.563
Because when you talk about the normie user, like I think you're going to need a simple metaphor that the normie user can just grok and interface with.

747
00:54:09.723 --> 00:54:15.726
It's not going to be the kind of hoops that you currently jump through with Codex.

748
00:54:16.066 --> 00:54:17.727
Yeah.

749
00:54:19.407 --> 00:54:23.549
How do you think about token spend both on your team and in the product?

750
00:54:24.189 --> 00:54:24.790
I think...

751
00:54:25.810 --> 00:54:26.571
Okay, I'll do team first.

752
00:54:26.591 --> 00:54:27.712
So token spend on team.

753
00:54:28.373 --> 00:54:34.660
I think folks on the team should be using AI to augment their abilities as much as possible.

754
00:54:35.081 --> 00:54:35.701
I think it's...

755
00:54:36.742 --> 00:54:42.464
silly to say, if you aren't spending this many tokens, then you're doing it wrong.

756
00:54:42.484 --> 00:54:44.445
That's like, I feel like a weird way to do it.

757
00:54:45.345 --> 00:54:53.048
I kind of trust the team to take the spirit of the goal, which is like, be the most productive you can be and be thoughtful about how you use it.

758
00:54:53.988 --> 00:55:01.111
We didn't even track token spend until like last week internally, just because it wasn't a priority for me.

759
00:55:01.791 --> 00:55:04.632
In terms of token spend in the product.

760
00:55:05.327 --> 00:55:06.007
Wait, wait, wait.

761
00:55:06.027 --> 00:55:09.069
Before we get there, but you started last week.

762
00:55:09.449 --> 00:55:12.330
What spurred it and then what did you come to?

763
00:55:12.850 --> 00:55:14.491
Fable.

764
00:55:14.511 --> 00:55:18.872
Basically, what spurred it was like, do we even know how much we're spending?

765
00:55:18.892 --> 00:55:22.834
We're like, we should probably know how much we're spending.

766
00:55:23.534 --> 00:55:28.696
It was basically because folks are using Fable and someone's like, I think I spent a couple of thousand dollars today.

767
00:55:28.716 --> 00:55:32.258
We're like, okay, maybe this is something we should at least be aware of.

768
00:55:32.758 --> 00:55:34.438
We did like literally the same thing.

769
00:55:34.738 --> 00:55:41.180
Fable came out and Kieran, who runs Quora, he was like on track to spend like a million dollars in credits.

770
00:55:41.740 --> 00:55:43.401
In a year.

771
00:55:43.481 --> 00:55:44.361
In a year.

772
00:55:44.421 --> 00:55:44.741
In a year.

773
00:55:44.801 --> 00:55:47.281
And we were like, okay, we need to figure it out.

774
00:55:47.301 --> 00:55:48.802
Maybe we need to think about this a little bit.

775
00:55:48.902 --> 00:55:49.082
Yeah.

776
00:55:49.202 --> 00:55:49.742
Yeah.

777
00:55:49.822 --> 00:55:51.082
It actually could be worth it.

778
00:55:51.162 --> 00:55:53.023
And also we need to think about it, you know?

779
00:55:53.243 --> 00:55:53.643
Exactly.

780
00:55:53.783 --> 00:55:54.003
Yeah.

781
00:55:54.223 --> 00:55:55.585
Yeah.

782
00:55:55.605 --> 00:55:58.049
This goes back to like, you know, things always outside of our ability.

783
00:55:58.069 --> 00:56:02.536
It's just one of those like, it's like, yeah, we should probably, we should probably at least know how much we're spending.

784
00:56:02.576 --> 00:56:06.221
That sounds like a really obvious thing, but it's just like, we're just so busy with so many other things.

785
00:56:06.281 --> 00:56:07.383
It's like, now's the time.

786
00:56:08.525 --> 00:56:09.726
In product, I think, um,

787
00:56:11.072 --> 00:56:27.719
Like I'm still a believer that it's early days and we have so much to gain by figuring out what are the killer kind of AI native experiences that Granola can build.

788
00:56:27.799 --> 00:56:28.379
And therefore...

789
00:56:30.717 --> 00:56:34.160
we're not very cost conscious on the token spend.

790
00:56:34.180 --> 00:56:46.010
So like I said before, like we pre-generate, I don't know, millions of these briefs, even if only like 10% of them get opened because we want to like figure out what's that right experience for the user.

791
00:56:46.530 --> 00:56:54.136
And our belief is both costs will go down over time and we can optimize costs once we know what the best experiences are.

792
00:56:54.216 --> 00:56:55.477
So like that's been...

793
00:56:56.826 --> 00:56:57.787
That's our philosophy.

794
00:56:58.087 --> 00:57:01.350
Argentic features are really expensive.

795
00:57:02.791 --> 00:57:08.655
And so as we build more and more of them, at some point that'll break the math for us.

796
00:57:09.716 --> 00:57:11.097
But for now, that's been the priority.

797
00:57:12.018 --> 00:57:16.341
And how much are people using true Argentic workflows in Granola?

798
00:57:16.822 --> 00:57:25.849
I asked that as like a, I'm really curious about normie users learning how to use like power features like those.

799
00:57:27.113 --> 00:57:29.434
It's the word agentic that's tough, right?

800
00:57:29.454 --> 00:57:34.998
Like I think that like, um, roughly half is the, is the answer, right?

801
00:57:35.018 --> 00:57:40.381
Like weekly, like, like every week, about half the granola users use us in an agentic way.

802
00:57:41.221 --> 00:57:46.364
Um, and, but I don't know if they would describe it that way, if that makes sense.

803
00:57:46.644 --> 00:57:46.825
Right.

804
00:57:46.845 --> 00:57:54.389
Like they have, they, they'll be like, they'll ask a query that is like a kind of a complex query around, uh, um.

805
00:57:54.836 --> 00:58:03.664
you know, like let's say your coaching or improvement question that will look at a series of meetings over time and then do several steps of analysis over that, right?

806
00:58:03.964 --> 00:58:07.587
Like that kind of thing is roughly half a week.

807
00:58:07.627 --> 00:58:07.848
Yeah.

808
00:58:09.089 --> 00:58:11.050
Really interesting.

809
00:58:11.791 --> 00:58:15.895
Chris, if people are looking to find you online to use Granola, where should they find you?

810
00:58:17.664 --> 00:58:19.004
Oh, I don't know.

811
00:58:19.084 --> 00:58:23.166
So Granola's Twitter handle or X handle is Meat Granola.

812
00:58:24.047 --> 00:58:25.508
Mine is CJ Pedregal.

813
00:58:25.828 --> 00:58:27.028
So CJ, my last name.

814
00:58:29.470 --> 00:58:31.571
And yeah, I guess those are the best places.

815
00:58:32.211 --> 00:58:32.611
Amazing.

816
00:58:32.811 --> 00:58:34.432
I always love talking to you.

817
00:58:34.452 --> 00:58:35.793
Love what you're building.

818
00:58:35.813 --> 00:58:36.393
Thanks for coming on.

819
00:58:36.413 --> 00:58:37.314
We got to do this more often.

820
00:58:38.185 --> 00:58:40.048
Yeah, absolutely.

821
00:58:40.108 --> 00:58:45.938
Whether it's on the pod or just us chatting, I really enjoyed these conversations.

822
00:58:46.359 --> 00:58:46.780
Sounds great.

823
00:58:55.046 --> 00:59:00.749
Oh my gosh, folks, you absolutely positively have to smash that like button and subscribe to AI and I.

824
00:59:00.969 --> 00:59:01.229
Why?

825
00:59:01.689 --> 00:59:03.871
Because this show is the epitome of awesomeness.

826
00:59:04.131 --> 00:59:06.032
It's like finding a treasure chest in your backyard.

827
00:59:06.332 --> 00:59:11.835
But instead of gold, it's filled with pure unadulterated knowledge bombs about chat GPT.

828
00:59:12.635 --> 00:59:20.057
Every episode is a roller coaster of emotions, insights, and laughter that will leave you on the edge of your seat, craving for more.

829
00:59:20.337 --> 00:59:21.158
It's not just a show.

830
00:59:21.678 --> 00:59:25.299
It's a journey into the future with Dan Shipper as the captain of the spaceship.

831
00:59:26.519 --> 00:59:28.060
So do yourself a favor.

832
00:59:28.320 --> 00:59:31.601
Hit like, smash subscribe, and strap in for the ride of your life.

833
00:59:32.041 --> 00:59:37.383
And now, without any further ado, let me just say, Dan, I'm absolutely hopelessly in love with you.