Every Agent Needs a Box — Aaron Levie, Box

WEBVTT
1
00:00:00.040 --> 00:00:07.600
Like you don't write code, you talk to an agent, and it goes and does it for you, and you maybe at best review it. That's even probably like- Mm-hmm, mm-hmm... like largely not even what you're doing.
2
00:00:08.180 --> 00:00:17.420
What's happening is we are changing our work to make the agents effective in that model. The agent didn't really adapt to how we work. We basically adapted to how the agent works.
3
00:00:17.740 --> 00:00:20.440
All of the economy has to go through that exact same evolution.
4
00:00:20.680 --> 00:00:31.860
Right now, it's a huge asset and an advantage for the teams that do it early and that are kinda wired into doing this, 'cause you'll see compounding returns, but that's just gonna take a while for most companies to actually go and get this deployed.
5
00:00:31.860 --> 00:00:46.080
[upbeat music] Welcome to The Lane Space Pod. We're back in the Chroma studio with, uh, Chroma CEO Jeff Hoover. Welcome. Returning guest, but now guest host. [laughs] It's a pleasure. Wow.
6
00:00:46.100 --> 00:00:54.840
How'd you get upgraded to, uh, to that? Because he, he's like the perfect guy to be guest host for you. That makes sense, actually. For you. [laughs] We love context. We, we both really love context. We really do.
7
00:00:55.020 --> 00:01:01.400
We really do. Uh, and we're here with, uh, Aaron Levie. Welcome. Thank you. Good to, uh, good to be here. Uh, yeah.
8
00:01:01.540 --> 00:01:08.680
Uh, so we've all met offline and, like, chatted a little bit, but, like, it's always nice to get these things in person and in conversation. Yeah. You just started off with so much energy.
9
00:01:08.700 --> 00:01:17.100
You're, you're ex- super excited about agents. I love agents. Uh, yeah. [laughs] OpenCloud just got bought, got bought by OpenAI. Well, not, not bought, but you know, you know what I mean. Some- Like-...
10
00:01:17.140 --> 00:01:29.040
some, uh, you know, acquihire-y... Executive hire. Executive hire. Okay. Executive hire. Okay. Hey, that's my term. Okay. [laughs] Um, what are you pounding the table on on agents? You have so many insightful tweets.
11
00:01:29.340 --> 00:01:49.980
Well, the thing that, that we get super excited about that I think is probably, you know, should be relatively obvious is we've, we've built a platform to help enterprises manage their files and their, their corporate files, and the permissions of who has access to those files, and the sharing and collaboration of those files, and all of those files contain really, really important information for the enterprise.
12
00:01:50.020 --> 00:01:55.340
It might have your contracts, it might have your research materials, it might have marketing information, it might have your memos.
13
00:01:55.860 --> 00:02:10.000
All that data obviously has pr- you know, predominantly been used by humans, but there's been one really interesting problem, which is that, you know, humans only really work with their files during an active engagement with them, and then they kinda go away, and you don't really see them for a long time.
14
00:02:10.620 --> 00:02:27.840
And all of a sudden, uh, with the power of a- AI and AI agents, all of that data becomes extremely relevant as this ongoing source of, of answers to new questions, of data that will transform into, into something else that, that produces value in your organization.
15
00:02:27.900 --> 00:02:32.500
It h- it contains the answer to the new employee that's onboarding that needs to ramp up on a project.
16
00:02:33.120 --> 00:02:42.500
Um, it contains the answer to the right thing to sell a customer when you're having a conversation to them, with them. Contains the roadmap information that's gonna produce the next feature.
17
00:02:42.560 --> 00:03:01.380
So all that data that previously we've been just sort of storing and, and s- you know, occasionally forgetting about 'cause we're only working on the new active stuff, all of that information becomes valuable to the enterprise, and it's gonna become extremely valuable to end users because now they can have agents go find what they're looking for and produce new, new value and new data on that information.
18
00:03:01.780 --> 00:03:08.560
And it's gonna become incredibly valuable to agents because agents can roam around and do a bunch of work, and they're gonna need access to that data as well.
19
00:03:08.820 --> 00:03:17.000
And, um, and, you know, sometimes that will be an agent that is sort of working on behalf of, uh, of, of you, and, and effectively as you as...
20
00:03:17.040 --> 00:03:23.640
And, and they are kind of accessing all of the same information that you have access to and, and operating as you in the system.
21
00:03:23.680 --> 00:03:32.300
And then sometimes there's gonna be agents that are just effectively autonomous and kind of run on their own, and, and you're gonna collaborate and work with them kind of like you did another person.
22
00:03:32.860 --> 00:03:43.740
OpenCloud being the most recent and maybe first real sort of, you know, kind of, you know, up- updating everybody's, you know, views of this landscape version of, of what that could look like, which is, okay, I have an agent.
23
00:03:43.780 --> 00:03:49.220
It's on its own system. It's on its own computer. It has access to its own tools. I probably don't give it access to my entire life.
24
00:03:49.240 --> 00:04:00.920
[laughs] I probably communicate with it like I would an assistant or a colleague, and then it, it sort of has this sandbox environment. So all of that has massive implications for a platform that manage enterprise data.
25
00:04:01.400 --> 00:04:10.700
We think it's gonna just transform how we work with all of the enterprise content that we work with, and we have just had to make sure we're building the right platform to support that. The sort of shorthand I put it is
26
00:04:11.620 --> 00:04:22.140
as people build agents, every- everybody's just realizing that every agent needs a box. Yes. And it's nice to be called Box and just give everyone a box. [laughs] If I, you know, if we can make that go viral- [laughs]
27
00:04:22.140 --> 00:04:27.800
... uh, like I, I think that, that terminology- I think that's the tagline, "Every agent needs a box"... w- every agent needs a box. If we can make that the headline of this- [laughs]...
28
00:04:27.860 --> 00:04:35.400
I'm fine with this and- That's the billboard I want. Yeah, exactly. [laughs] Every agent needs a box. Um- Yeah... I like it. Can we ship this? Yeah. Like, okay. Let's do it. Uh, my work here is done.
29
00:04:35.400 --> 00:04:38.180
[laughs] And I got the value I needed out of this podcast. We can go have drinks. Yeah.
30
00:04:38.340 --> 00:04:50.040
But, but, um, but, but, you know, so the thing that we, we kind of think about is, um, is, you know, whether you think the number is 10X or 100X or whatever the number is, we're gonna have some order of magnitude more agents than people.
31
00:04:50.080 --> 00:04:57.420
That's inevitable. It has to happen. So then the question is what is the infrastructure that's needed to make all those agents effective in the enterprise?
32
00:04:57.500 --> 00:05:05.420
Make sure that they are well-governed, make sure they're only doing safe things on your information, make sure that they're not getting exposed to data that they shouldn't have access to.
33
00:05:05.840 --> 00:05:17.440
There's gonna be just incredibly, spectacularly crazy security incidents that will happen with agents because you'll prompt inject an agent and sort of find your way through the CRM system and pull out data that you shouldn't have access to.
34
00:05:17.820 --> 00:05:20.340
So we have- Oh, my God. Right. I mean, that's just gonna happen all over the place, right?
35
00:05:20.380 --> 00:05:26.700
So, so then the thing is, is how do you make sure you have the right security, the permissions, the access controls, the data governance?
36
00:05:27.240 --> 00:05:31.880
Um, we actually don't yet exactly know in many cases how we're gonna regulate some of these agents, right?
37
00:05:31.980 --> 00:05:42.440
If you think about an agent in financial services, does it have the exact same financial sort of, uh, requirements that a human did, or is it, is the risk fully on the human that was interacting or created the agent?
38
00:05:42.700 --> 00:05:52.820
All open questions. But no matter what, there's gonna need to be a layer that manages the, the data they have access to, the workflows that they're involved in, pulling up data from multiple systems.
39
00:05:53.200 --> 00:06:03.212
This is the new infrastructure opportunity in the era of agents. You have a piece on agent identities, which I think was today, um, which I think a lot of us- Breaking news. [laughs]...
40
00:06:03.252 --> 00:06:10.452
the security, security people are talking about, right? Like basically, I always think of this as like, well, you need the human you and then there's, you need the agent you. Yes.
41
00:06:10.512 --> 00:06:18.712
And, uh, well, I, it, I don't know if it's that simple, but is Box going to have an opinion on that? Or you're just gonna be like, "Well, we, we're just a sort of this, the source layer- Yeah...
42
00:06:18.792 --> 00:06:27.552
let Okta Auth0 handle that." I think we're gonna have an opinion, and we will work with generally wherever the contours of the market end up.
43
00:06:27.812 --> 00:06:38.772
Um, and the reason that we're gonna have an opinion more than other topics probably is because one of the biggest use cases for why your agent might need it an identity is for file system access.
44
00:06:39.412 --> 00:06:42.412
So, so thus, we have to kinda think about this pretty deeply.
45
00:06:42.932 --> 00:06:49.972
And I think, uh, unless you're like in our world thinking about this particular problem all day long, it might be, you know, like, why is this such a big deal?
46
00:06:50.072 --> 00:06:59.912
And the reason why it's a really big deal is because sometimes sort of say, well, just give the agent a- an account on the system, and it just treats, treat it like every other type of user on the system.
47
00:06:59.932 --> 00:07:10.272
The problem is, is that I, as Aaron, don't really have any responsibility over anybody else's Box account in our organization. I can't see the Box account of any other employee that I work with.
48
00:07:10.792 --> 00:07:19.652
I am not liable for anything that they do, and they have, I have n- I have, you know, strict privacy requirements on everything that they are able to, you know, that they, they work on.
49
00:07:20.152 --> 00:07:29.292
Agents don't have that, you know, don't have tho- those properties. The person who creates the agent probably is gonna, for the foreseeable future, take on a lot of the liability of what that agent does.
50
00:07:29.672 --> 00:07:39.732
That agent doesn't deserve any privacy because, because it's, you know, it, it can't fully be autonomously operated, and it doesn't have any legal, you know, kind of, you know, responsibility.
51
00:07:40.172 --> 00:07:47.252
So thus, you can't just be like, "Oh, I'll just create a bunch of accounts, and then I'll, I'll kinda work with that agent, and I'll talk to it occasionally." Like, you need oversight over that.
52
00:07:47.412 --> 00:08:01.152
And so then the question is, how do you have a world where the agent, sometimes you have oversight of, but what if that agent goes and works with other people and that person, uh, over there is collaborating with the agent on something, you shouldn't have access to what they're doing.
53
00:08:01.472 --> 00:08:08.071
So we have all of these new boundaries that we're gonna have to figure out of, of, you know, it's really, really easy. So far, we've been in, in easy mode.
54
00:08:08.472 --> 00:08:16.052
We've hit the easy button with AI, which is the agent just is you, and when you're in Cloud Code and you're in Cursor and you're in Codex, you're just, the agent is you.
55
00:08:16.312 --> 00:08:25.632
It, you're authing in to your services, it can do everything you can do. That's the easy mode. The hard mode is agents are kinda running on their own. People check in with them occasionally.
56
00:08:25.692 --> 00:08:36.272
They're doing things autonomously. How do you give them access to resources in the enterprise and not dramatically increase the security risk and the risk that you might expose the wrong thing to somebody?
57
00:08:36.792 --> 00:08:42.232
These are all the new problems that we have to get solved. I like the identity layer and, and identity vendors as being a solution to that.
58
00:08:42.572 --> 00:08:59.792
But we'll, we'll need some opinions as well because so many of the use cases are these collaborative file system use cases, which is how do I give an agent a subset of my data and give it its own workspace as well, because it's gonna need to store off its own information that would be relevant for it, and how do I have the right oversight into that?
59
00:09:00.352 --> 00:09:06.392
One thing which, um, I think is kind of interesting about that, you know, how humans work, right? Like, I may not also just, like, give you access to the whole file.
60
00:09:06.712 --> 00:09:13.952
I might, like, sit next to you and, like, scroll to this, like, one part of the file- [laughs]... and just show you that, like, one part and, like- Okay... you know. Partial file access?
61
00:09:14.112 --> 00:09:18.692
Well, I'm just saying, I think, like, RBAC- [laughs]... like, RBAC does seem to be dead, right? Like, if you wanna say something is dead- Uh-huh...
62
00:09:18.772 --> 00:09:24.692
probably RBAC is dead, and, uh, like, the Auth story to me seems, like, incredibly unsolved and unaddressed by, like- Mm...
63
00:09:24.732 --> 00:09:33.032
the existing state of, like, AI vendors, but- Yeah, I think, um, we're, I mean, you're taking obviously really to lev- a level of limit that we probably need to solve for. Yeah.
64
00:09:33.052 --> 00:09:50.552
And we built an access control system that was, was kinda like, you know, its own little world for a l- for a long time, and, um, and the idea was this, it's a many-to-many collaboration system where I can give you any part of the file system, and it's a waterfall model, so if I give you higher up in the, in the, in the system, you get everything below.
65
00:09:51.132 --> 00:09:59.132
And that, that kind of created immense flexibility- Mm... because I can kinda point you to any layer in the, in the tree, but then you're gonna get access to everything kinda below it. Mm.
66
00:09:59.372 --> 00:10:10.472
And that mostly is, is working in this, in this world, but you do have to manage this issue, which is how do I create an agent that has access to some of my stuff and somebody else's stuff as well- Mm...
67
00:10:10.812 --> 00:10:19.652
and which parts do I get to look at as the creator of the agent and, and these are just brand-new problems. Yeah, crazy. And humans, when there was a human there, that was really easy to do.
68
00:10:19.992 --> 00:10:27.872
Like, like, if the three of us were all sharing, there'd be a Venn diagram where we'd have an overlapping set of things we've shared, but then we'd have our own ways that we shared with each other.
69
00:10:28.592 --> 00:10:33.892
But in an agent world, somebody needs to take responsibility for what that agent has access to and what they're working on.
70
00:10:34.152 --> 00:10:39.972
These are, like, the, some of the most probably, you know, boring problems for 98% of people on, on the internet.
71
00:10:39.992 --> 00:10:47.252
[laughs] But they will be the problems that are the difference between can you actually have autonomous agents in an enterprise context- Mm. Yeah... that are not leaking your data constantly.
72
00:10:47.492 --> 00:10:56.472
No, like, I mean, you know, I run a very, very small company for my conference, and, like, we already have data sensitivity, -tivity issues. Yes. And some of my team members cannot see- Yes... uh, the others.
73
00:10:56.532 --> 00:11:00.472
And, like, I can't imagine what it's like to run a Fortune 500, and you have to worry about this.
74
00:11:00.752 --> 00:11:10.392
I- I'm just kinda curious, like, you, you talked to a lot, like, like 70, 80% of your cust- uh, of the Fortune 500 are your customers? Yep. 67%, just so we're being very- Okay, yeah... SEC precise. Sorry.
75
00:11:10.432 --> 00:11:17.272
So yeah, no, yeah. Okay. Okay. Something, I'm rounding up. Yes. Yes. Uh, I'm projecting to- I appreciated the round up- I'm projecting to the-... but for the government, uh- I'm projecting to the end of the year. Okay.
76
00:11:17.272 --> 00:11:22.992
Okay, thank you. There we go. There we go. You do make it sound like I, like we, we, well, well we gotta be honest, like we're t- we're taking way too long to get to 80%.
77
00:11:22.992 --> 00:11:28.612
[laughs] Well, no, I mean, so, like, how are they approaching it? Yes. Right? Because you- Yeah... you don't have a, you don't have a final answer yet. Yes.
78
00:11:28.652 --> 00:11:36.352
Well, okay, so, so th- uh, this is actually, this is the stark reality that, like, unfortunately is the kinda like pouring the water on the party a little bit. Yes.
79
00:11:36.832 --> 00:11:50.832
We all in Silicon Valley are like have the absolute best conditions possible for AI ever. [laughs] And I think we all saw the Dwarkesh, you know, kind of Dario podcast and this idea of AI coding. Why is that taking off?
80
00:11:51.192 --> 00:11:53.632
And, and we're not yet fully seeing it everywhere else.
81
00:11:53.672 --> 00:12:01.992
Well, like, if you just like enumerated the list of properties that AI coding has and then compared it to other knowledge work, let's just, let's just go through a few of them.
82
00:12:02.608 --> 00:12:08.548
Generally speaking, you bring on a new engineer, they have access to a large swath of the code base. Like, there's like very... Like, you just...
83
00:12:08.808 --> 00:12:19.108
Like, new engineer comes on, they can just go and find the, the, the stuff that they, they need to work with. It's a fully text in, text out, you know, medium. It's only-- it's just gonna be text at the end of the day.
84
00:12:19.508 --> 00:12:26.108
So it's, like, really great from a, from just a, a, you know, kinda what the agent can work with. Obviously, the models are super trained on that data set.
85
00:12:26.528 --> 00:12:36.808
The labs themselves have a really strong kind of self-reinforcing positive flywheel of why they need to do, you know, agentic coding deeply. So then you get just better tooling, better services.
86
00:12:37.268 --> 00:12:52.368
The actual developers of the AI are daily users of the, of the thing that they're, we're working in versus like the, you know, probably there's only like seven Claude Cowork legal plugin users at Anthropic any given day, but there's like a couple thousand Claude Code, eh, you know, users every single day.
87
00:12:52.428 --> 00:12:57.188
So just, like, think about which one are they getting more feedback on all day long. So you just go through this list.
88
00:12:57.668 --> 00:13:02.628
You have a, you know, everybody who's a developer by definition is technical, so they can go install the latest thing.
89
00:13:03.068 --> 00:13:11.948
We're all generally online, or at least, you know, kind of the weird ones are, and we're all talking to each other, sharing best practices. Like, that's, like, already eight differences versus the rest of the economy.
90
00:13:12.408 --> 00:13:18.108
Every other part of the economy has, like, like, six to seven headwinds relative to that list.
91
00:13:18.388 --> 00:13:43.328
You go into a company, you're a banker in financial services, you have access to like a, a tiny little subset of the total data that's gonna be relevant to do your job, and you're gonna have to start to go and talk to a bunch of people to get the right data to do your job because Sally didn't add you to that deal room, you know, folder and that, the, you know, the information is actually in a completely different organization that you now have to go and, and, and sort of run into.
92
00:13:43.648 --> 00:13:51.248
And it's like you have this endless list of access controls and security as, as you talked about. You have a medium which is not, it's not just text, right?
93
00:13:51.428 --> 00:13:55.208
You have, you have a Zoom call that, that you're getting all of the requirements from the customer.
94
00:13:55.268 --> 00:14:01.268
You have a lot of in-person conversations, and you're doing in-person sales and, like, how do you ever digitize all of that information?
95
00:14:01.748 --> 00:14:08.288
Um, you know, I think a lot of people got upset with this idea that the code base has all the context, um, that I don't know if you follow, you know...
96
00:14:08.448 --> 00:14:14.408
Did you follow some of that conversation that, that went viral is like, you know, it's not that simple that, that the code base doesn't have all the knowledge.
97
00:14:14.808 --> 00:14:22.928
But, like, it's a lot, you're a lot better off than you are with other areas of knowledge work. Like, you, we, like, we, like, have documentation practices. You write specifications.
98
00:14:23.188 --> 00:14:26.528
Those things don't exist for like eighty percent of work that happens in the enterprise.
99
00:14:27.008 --> 00:14:48.328
That's the divide that we have which is, which is AI coding has, has just fully es- uh, you know, we've reached escape velocity of how powerful this stuff is, and then we're gonna have to find a way to bring that same energy and momentum but to all these other areas of knowledge work where the tools aren't there, the data's not set up to be there, the access controls don't make it that easy.
100
00:14:48.648 --> 00:14:54.968
The context engineering is an incredibly hard problem because, again, you have access control challenges. You have different data formats.
101
00:14:55.368 --> 00:15:03.188
You have end users that are gonna need to kinda be kinda trained through this as opposed to they're adopting these tools in their free time. That's where the Fortune 500 is.
102
00:15:03.408 --> 00:15:15.108
And so we, I think, you know, have to be prepared as an industry where we are gonna be on a multiyear march to, to be able to bring agents to the enterprise for these workflows.
103
00:15:15.588 --> 00:15:22.748
And I think probably the, the thing that we've learned most in coding that, that the rest of the world is not yet, I think, ready for, I mean, we're, we're...
104
00:15:22.788 --> 00:15:40.248
They'll, they'll have to be ready for it because it's just gonna inevitably happen is I think in coding what, what's interesting is if you think about the practice of coding today versus two years ago, yeah, it's probably the most changed workflow in maybe the history of time from the amount of time it's changed, right?
105
00:15:40.248 --> 00:15:50.088
[chuckles] Yeah. Like, like, has any, has any workflow in the entire economy changed that quickly i-in terms of the amount of change? I just, uh, you, you know, a-at least in any knowledge worker workflow.
106
00:15:50.428 --> 00:15:58.328
There's, like, very rarely been an event where one piece of technology and work practice has so fundamentally, you know, changed, changed what you do. Like, you don't write code.
107
00:15:58.548 --> 00:16:05.848
You talk to an agent, and it goes and does it for you, and you maybe at best review it, and even i- that's even probably like- Mm-hmm. Mm-hmm... like largely not even what you're doing.
108
00:16:06.348 --> 00:16:15.968
What's happening is we are changing our work to make the agents effective in that model. The agent didn't really adapt to how we work. We basically adapted to how the agent works. Mm-hmm.
109
00:16:16.448 --> 00:16:19.267
All of the economy has to go through that exact same evolution.
110
00:16:19.767 --> 00:16:37.088
The rest of the economy is gonna have to update its workflows to make agents effective and to give agents the context that they need and to actually figure out what kind of prompting works and to figure out how do you ensure that the agent has the right access to information to be able to execute on its work.
111
00:16:37.648 --> 00:16:43.368
I, you know, this is not the panacea that people were hoping for of the agent drops in, just automates your life.
112
00:16:43.768 --> 00:16:54.248
Like, you have to basically re-engineer your workflow to get the most out of agents and, uh, and that, that's just gonna take m- you know, multiple years across the economy.
113
00:16:54.708 --> 00:17:01.248
Right now it's a huge asset and an advantage for the teams that do it early and that are kinda wired into doing this 'cause you'll see compounding returns.
114
00:17:01.568 --> 00:17:07.508
But that's just gonna take a while for most companies to actually go and get this deployed. I love, I love pushing back.
115
00:17:07.968 --> 00:17:16.448
I think that that is what a lot of con- technology consultants love to hear this sort of stuff, right? [laughs] Yeah. First, to, to embrace the AI- Yeah...
116
00:17:16.448 --> 00:17:22.888
to get to the promised land you must pay me so much money- Hundred percent... to adopt the prescribed way of, uh, conforming to the agents. Yes.
117
00:17:22.908 --> 00:17:32.228
And I worry that you will be eclipsed by someone else who says, "No, come as you are"- Yeah... "and we'll meet you where you are" and- And, and, and what was the thing that went viral a week ago?
118
00:17:32.428 --> 00:17:44.376
OpenAI probably, uh, is hiring FDEs- Yeah... uh, to go into the enterprise. Yeah. And then Anthropic is embedded at Goldman Sachs. Yeah. So if the labs are having to do this- [laughs]...
119
00:17:44.396 --> 00:17:53.776
then if, if the labs have decided that they need to hire FDE and professional services, then I think that's a pretty clear indication that this... there's no easy mode- Okay... of workflow transformation. Yeah, yeah.
120
00:17:53.836 --> 00:18:14.676
So, so to your point, I think actually this is a market opportunity for, you know, new professional services and consulting firms that are, like, agent pilled, and they, and they kinda, you know, go into organizations, and they figure out how to re-engineer your workflows to make them more agent-ready and get your data into the right format and r- you know, reconstruct your business process so you're, you're not doing most of the work.
121
00:18:14.716 --> 00:18:22.496
You're telling agents how to do the work, and then you're reviewing it. But I haven't seen the thing that can just drop in and, and kinda let you not go through those changes.
122
00:18:22.776 --> 00:18:30.436
I don't know how that kind of sales pitch goes over. Yeah. Like, you know, you're s- you're saying things like, "Well, in my sort of nice, beautiful walled garden, here's, here's, uh- [laughs]...
123
00:18:30.596 --> 00:18:38.396
here's this, here's this beautiful Box account that has everything." Yes. And I'm like, "Well, most, most real life is extremely messy- Sure... and, like, poorly named, and there's duplicates- But, but, but the, the-...
124
00:18:38.436 --> 00:18:48.316
and there's outdated shit. 100%. [laughs] And so- [laughs] No, no, 100%. And so this is actually... No, so, so this is... I mean, we agree that, that getting to the beautiful garden is gonna be tough. Yeah.
125
00:18:48.376 --> 00:18:52.276
There's also the other end of the spectrum where I c- I just like... It's a technical impossibility to solve.
126
00:18:52.916 --> 00:19:01.036
The agent is, is truly cannot get enough context to make the right decision in, in the, in the incredibly messy land. Like, there's no AGI that will solve that.
127
00:19:01.536 --> 00:19:13.616
So, so we're gonna have to kinda land in somewhere in between, which is, like, we all collectively get better at documentation practices and, and having authoritative, relatively up-to-date information and putting it in the right place.
128
00:19:13.956 --> 00:19:30.856
Like, agents will, will certainly cause us to be much better organized around how we work with our information, simply because the severity of the agent pulling the wrong data will be too high, and the productivity gain of, that you'll miss out on by not doing this will be too high as well- Yeah...
129
00:19:30.896 --> 00:19:37.576
that y- that your competition will just do it, and th- they'll just have higher velocity. So, uh... and, and we, we see this a lot firsthand.
130
00:19:37.636 --> 00:19:46.596
So we, we build a series of agents internally that they can kind of have access to your full Box account and go off, and you give it a task, and it can go find whatever information you're looking for and work with.
131
00:19:47.316 --> 00:19:56.976
And, you know, thank God for the model progress. But, like, if, if you gave that task to an agent nine months ago, you're just gonna get lots of bogus answers- [laughs]...
132
00:19:57.056 --> 00:20:04.716
because it's gonna, it's gonna say, "Hey, here's, here are fi- five, you know, documents that all kinda smell like the right thing, and I'm gonna..." But I...
133
00:20:04.736 --> 00:20:13.936
But you're, you're putting me on the clock 'cause my system prompt says, like, you know, be pretty smart but also try and respond to the user. And it's gonna respond, and it's like, "Ah, it got the wrong document."
134
00:20:14.196 --> 00:20:20.756
And then you do that once or twice as a knowledge worker, and you just- Never again. Never again. Yeah, yeah. You're just, like, done with the system. Yeah. Doesn't work. It doesn't work.
135
00:20:20.796 --> 00:20:29.996
And so, you know, Opus 4.6 and Gemini 3 1 Pro and, you know, uh, the, whatever the latest 5.3 GBT will be- [laughs]...
136
00:20:30.036 --> 00:20:38.996
like, those things are getting better and better, and it's using better judgment, and this sort of, like, the, all of these updates to the agentic tool and search systems are, are...
137
00:20:39.036 --> 00:20:46.016
we're seeing, we're seeing very real progress where the agent kinda can, can almost smell something's a little bit fishy when it's getting...
138
00:20:46.356 --> 00:20:55.836
You know, we, we have this process where we, we have it go fan out to a bunch of searches, pull up a bunch of data, and then it has to sort of do its own ranking of, you know, what are the right documents that, that it should be working with.
139
00:20:56.656 --> 00:21:05.676
And again, like, you know, the intelligence level of a model six months ago, w- it'd be just throwing a dart at, like, I'm just... I'm gonna grab these seven files, and I, I pre- I hope that that's the, the right answer.
140
00:21:06.036 --> 00:21:15.876
And something like an Opus, first 4.5, and now 4.6, is like, oh. It's like, no, that one doesn't seem right relative to this question because I'm seeing some signal that is making that...
141
00:21:15.896 --> 00:21:22.836
you know, that's contradicting the document where it would normally be in the tree and who should have access. Like, it's doing all of that kind of work for you.
142
00:21:23.256 --> 00:21:30.036
But, like, it still doesn't work if you just have a total wasteland of data. Like, it's just not, it's just not possible, partly because a human wouldn't even be able to do it.
143
00:21:30.476 --> 00:21:40.356
So basically, if a, if a really, really smart human could not do that task in 5 or 10 minutes for a search retrieval-type task, like, the, you know, your agent's not gonna be able to do it any better.
144
00:21:40.696 --> 00:21:45.376
Uh- You see this all day long, so [laughs] This touches on a thing that Jeff's passionate about, which is context engineering.
145
00:21:45.576 --> 00:21:50.256
I- I'm just gonna let you ramble or riff on, on context engineering if, if, if there's anything.
146
00:21:50.356 --> 00:21:59.836
Like, y- you did really good work on context rot, which has really taken over as, like, the term that people use and the reference. 100%. Yeah. We, we... all we think about is, is the context rot problem.
147
00:22:00.336 --> 00:22:09.136
Yeah, there's certainly a lot of, like, ranking considerations. Agentic search I think is incredibly promising. Um, yeah, I was trying to generate a question though. I don't have a question right now, Swix. Yeah.
148
00:22:09.216 --> 00:22:23.916
No, but like, like, I think, uh, there was this moment, um, you know, like, I don't know, two years ago before, before we knew, like, where the, the gotchas were gonna be in AI, and I think someone was like, was like, "Well, infinite context windows will just solve all of these problems"- [laughs]...
149
00:22:23.976 --> 00:22:32.916
and because you'll just, you'll just give the context window, like, all the data. And it's just like, okay, I mean, maybe in 2035, like, this is a viable solution.
150
00:22:32.976 --> 00:22:35.396
First of all, it w- it would just, it would just simply cost too much. Yeah.
151
00:22:35.436 --> 00:22:49.356
Like, we just can't give the model, like, the 5,000 documents that might be relevant, and it's gonna read them all, and I've seen enough to, to start believing in crazy stuff, so, like, I'm willing to just say sure, like in, in 10 years from now.
152
00:22:49.356 --> 00:22:54.136
[laughs] Never say never. In, in 10 years from now- Never say never... we'll have infinite context windows at, at a thousandth of the price of today.
153
00:22:54.196 --> 00:22:57.496
Like, let's just, like, believe that that's possible, but right, we're in reality today.
154
00:22:57.996 --> 00:23:08.416
So today, we have a context engineering problem, which is I got, I got, you know, 200,000 tokens that I can work with, or probab- I don't even know what, what the latest graph is before, like, massive degradation.
155
00:23:08.696 --> 00:23:12.756
Like 60. Okay. 60. I have 60,000 tokens that I get to work with where I'm gonna get accurate information.
156
00:23:13.676 --> 00:23:23.996
That's not a lot of tokens for a corpus of 10 million documents that a knowledge worker might have across all of the teams and all of the projects and all the people they work with.
157
00:23:24.096 --> 00:23:36.516
I have, I have 10 million documents which, you know, maybe is times five pages per document or something like that. I'm at 50 million pages of information, and I have 60,000 tokens. Like, holy shit. Yeah.
158
00:23:36.536 --> 00:23:45.624
This is like, how do I bridge the 50 million pages of information with- You know, the couple hundred that I get to work with in that, in that token window. Yeah.
159
00:23:45.844 --> 00:23:56.464
This is like, th- this is like such an interesting problem. And that's why actually so much work is actually like, just like search systems and the databases, and that layer has to just get so locked in.
160
00:23:56.824 --> 00:24:10.024
But models getting better, and importantly, knowing when they've done a search, they found the wrong thing, they go back, they check their work, they, they find a way to balance sort of appeasing the user versus double-checking.
161
00:24:10.384 --> 00:24:20.244
We have this one- Mm-hmm. We have this one test case where we ask the agent to go find ten pieces of information. Is this the complex work eval? Uh, this is actually not an eval.
162
00:24:20.284 --> 00:24:25.844
This is, this is sort of just like we have a bunch of- Different... we have a bunch of internal benchmark kind of scenarios every time we, we update our agent.
163
00:24:26.384 --> 00:24:38.184
We have one which is, I ask it to find all of our office addresses, and I give it the list of ten offices that we have. And there's not one document that has this. Maybe there should be.
164
00:24:38.524 --> 00:24:46.544
That would be a great example of the kind of thing that like maybe over time companies start to, you know, have these sort of like, what are the canonical, you know, kind of key areas of knowledge that we need to have?
165
00:24:46.684 --> 00:24:52.384
We don't seem to have this one document that says, "Here are all of our offices." Mm-hmm. Mm-hmm. We have a bunch of documents that have like, "Here's the New York office," and whatever. Hmm.
166
00:24:52.564 --> 00:24:56.904
So you'd ask this agent, and you, you get... You say, "I need the addresses for these ten offices," okay?
167
00:24:57.444 --> 00:25:05.904
And by the way, if you do this on any ch- you know, public chat model, the same outcome is gonna happen, but for a different kind of query. You give it... You say, "I need these ten addresses."
168
00:25:06.804 --> 00:25:14.664
How many times should the agent go and do its search before it decides whether or not there's just no answer to this question?
169
00:25:15.304 --> 00:25:24.224
Often, and especially the, the, let's say lower tier models, it'll come back and it'll give you six of the ten addresses, and it'll, and it'll just say, "I couldn't find the other four."
170
00:25:24.244 --> 00:25:29.004
'Cause it, it doesn't know what it doesn't know. It doesn't know what it doesn't know. Yeah. So the model is just like, like when should it stop?
171
00:25:29.144 --> 00:25:34.664
When should it stop doing-- Like, should it, should it do that task for literally an hour and just keep cranking through?
172
00:25:35.084 --> 00:25:40.564
Maybe I actually made up an office location, and it doesn't know that I made it up, and I didn't even know that I made it up.
173
00:25:41.004 --> 00:25:51.064
Like, should it just keep re- should it e- read every single file in your entire box account until it, until it [laughs] exhausts every single piece of information? Expensive. These are the new problems that we have.
174
00:25:51.344 --> 00:25:59.684
So, you know, something like let's say a new Opus model is sort of like, "Okay, I'm gonna try these types of queries. I didn't get exactly what I wanted. I'm gonna try again. I'm gonna...
175
00:25:59.744 --> 00:26:09.164
At some point I'm gonna stop searching 'cause I've determined that, that no amount of searching is gonna solve this problem. I'm just not able to do it." And that judgment is like a really new thing- Yeah...
176
00:26:09.204 --> 00:26:14.664
that the model needs to be able to have. It's like, when should it give up on a task? 'Cause, 'cause you just don't... It just can't find the thing.
177
00:26:15.184 --> 00:26:24.264
That's the real world of knowledge work problems, and this is the stuff that the coding agents don't have to deal with. [laughs] 'Cause it, it just doesn't. Like, like you're not usually asking it about...
178
00:26:24.384 --> 00:26:29.244
You're, you're always creating net new information coming right out of the model, for the most part.
179
00:26:29.324 --> 00:26:38.064
Obviously, it has to know about your code base and your specs and your documentation, but, but when you deploy an agent on all of your data, that, now you have all of these new problems that you're dealing with.
180
00:26:39.004 --> 00:26:47.944
Our, uh, follow-up research to context ride is actually on agentic search. Oh. Um, and we've like- Great... sort of stress-tested like frontier models and their ability to search.
181
00:26:48.504 --> 00:26:55.324
Um, and they are not actually that good at searching. Right. Uh, so you're sort of highlighting this like explore exploit. You're just a Debbie Downer. [laughs] Everything doesn't work.
182
00:26:55.364 --> 00:27:01.644
Like [laughs] Well, somebody has to be. Um- Can I just throw out one more thing- Yeah... that is different from coding and, and the rest of the knowledge work that I've, I've failed to mention?
183
00:27:01.724 --> 00:27:08.844
So one other kind of key point is, is that, you know, at the end of the day, whether you believe we're in a slopocalypse or, or whatever.
184
00:27:08.844 --> 00:27:17.084
[laughs] At the end of the day, if you, if you build a working product at the end of... If you, if you've built a working solution, that is ultimately what the customer is paying for.
185
00:27:17.384 --> 00:27:27.444
Like, whether I have a lot of slop, a little slop or whatever, I'm sure there's lots of code bases we can go into in enterprise software companies where it's like just crazy slop that humans did over a twenty-year period.
186
00:27:27.844 --> 00:27:36.284
But the end customer just gets this little interface. They can, they can type into it. It does its thing. Knowledge work, uh, doesn't a- have that property.
187
00:27:36.784 --> 00:27:47.744
If I have an AI model go generate a contract, and I generate a contract twenty times, and, you know, all twenty times it's just three percent different. [laughs] And like that, I...
188
00:27:47.784 --> 00:27:56.264
That, that kind of slop introduces all new kinds of risk for my organization that the code version of that slop didn't, didn't introduce. These are...
189
00:27:56.304 --> 00:28:05.004
And so like so how do you constrain these models to just the part that you want them to work on and just do the thing that you want them to do? And, and, you know, in engineering, we don't...
190
00:28:05.184 --> 00:28:15.404
You can't be disbarred as an engineer. But you can be disbarred as a lawyer. Like, you can do the wrong medical thing in healthcare. You, there's no, there's no equivalent to that of engineering.
191
00:28:15.584 --> 00:28:23.024
Like- Do you want there to be? 'Cause I've considered- [laughs] What was that? Civil engineering there is, right? Yeah. Civil engineering, sure. Oh yeah, for sure. But like in any of our companies- Yeah...
192
00:28:23.064 --> 00:28:31.564
you like, you know, you'll be forgiven if you took down the site. And, and we- [laughs]... we'll do a rollback, and you'll, you'll be in a meeting, but you have not been disbarred as an engineer.
193
00:28:31.584 --> 00:28:36.984
We don't, we don't change your, you know, your computer science, uh, degree. Blame them for postmortem. Yeah, exactly. Exactly.
194
00:28:37.024 --> 00:28:45.864
So, so, uh, now maybe we collectively as an industry need to figure out like what are you liable for, not legally, but like in a, in a management sense, o- of these agents.
195
00:28:46.264 --> 00:28:53.924
All sort- sorts of interesting problems that, that, that, uh, that have to come out. But in knowledge work, that's the real hostile environments that we're operating in.
196
00:28:54.784 --> 00:29:02.944
I do think, like, uh, a lot of the last year's 2025 story was the rise of coding agents, and I think 2026 story is definitely knowledge work agents. Yes, 100%. Right? Like that would...
197
00:29:02.984 --> 00:29:12.464
And I think OpenClonk, CoWork are just the beginnings. Yes. Like it's [laughs] the next one's gonna-- just gonna be absolute craziness. It, it is. And, and, uh, and it's gonna be...
198
00:29:12.504 --> 00:29:23.304
I mean, again, like this is gonna be this, this wave where we, we are gonna try and bring as many of the practices from coding, because that, that will clearly be the forefront, which is tell an agent to go do something.
199
00:29:23.344 --> 00:29:28.384
It has an access to a set of resources. You need to be responsible for re- reviewing it at the end of the process.
200
00:29:29.204 --> 00:29:37.244
That to me is the, is the kind of template that I just think goes across knowledge work, and Odd CoWork is a great example. OpenClonk's a great example.
201
00:29:37.784 --> 00:29:47.352
You know, you can kind of sort of see what Codex could become over time. These are so- some really interesting kind of platforms that are emerging. Okay, um, I wanted to- We touched on evals a little bit.
202
00:29:47.392 --> 00:29:54.312
You had, you had the re- report that you're gonna go bring up, and then I was gonna go into, like, uh, boxes, evals. Yeah. But, uh, go ahead. C- talk, talk about your agentic search thing.
203
00:29:54.572 --> 00:30:02.932
Yeah, mostly I think kind of a, a few of the insights is like everyone, frontier model is not good at search. Humans have this natural explore-exploit trade-off. We kind of understand, like, when to stop doing something.
204
00:30:03.252 --> 00:30:12.432
Also, humans are pretty good at, like, forgetting, actually, like pruning their own context, whereas agents are not. And actually an agent in their kinda context history, if they knew something was bad and even...
205
00:30:12.472 --> 00:30:18.872
You can see in the trace, the reasoning trace, "Hey, that probably wasn't a good idea." If it's still in the trace- Mm... still in the context, they'll still do it again. Uh-huh.
206
00:30:18.952 --> 00:30:25.432
Uh, and so, like, I think pruning is also gonna be, like, a really... It's already becoming a thing record, like letting models like self-prune their context windows- Yeah... maybe a big deal.
207
00:30:25.452 --> 00:30:31.912
So, so don't leave the mistake- Um... don't leave the mistake in there. Cut out the mistake, but tell it that you made a mistake in the past and so it doesn't repeat it.
208
00:30:32.032 --> 00:30:38.212
Yeah, like cut it out so it doesn't get, like, distracted by it again- Yeah... 'cause really, you know, what is- So, so it will repeat its mistake just because it's been- It's in the context...
209
00:30:38.432 --> 00:30:45.552
it's in the context so much. That's a few shot example. [laughs] Even if it knows, even if it knows that... Like, it's like, "Oh, this is a great thing to go try" Even if it didn't work. Yeah. Exactly.
210
00:30:45.812 --> 00:30:52.092
So there's like a bunch of stuff there that- Oh my gosh. Groundhog's Day inside these models. [laughs] Yeah. I'm gonna go keep doing the same wrong thing repeatedly. Kind of makes sense, right?
211
00:30:52.112 --> 00:30:58.772
I feel like, you know, some crude analogy, you're trying to like fit a manifold in latent space, which kind of is doing break program synthesis, which is kind of one way to think about like what LLMs are doing, right?
212
00:30:58.912 --> 00:31:06.392
Like- Yeah... you know, certain facts might be like sort of overly pinning it- Yeah... for certain, you know, s- s- sectors of latent space and so like- Yeah, yeah, yeah... plug latent space. Yeah, exactly.
213
00:31:06.392 --> 00:31:13.692
And, uh, and so- We have to bell, uh, our editor adds a bell every time you say that. [laughs] You have to, you have to like remove those like- You should have a gong like, uh, TVPN or something.
214
00:31:13.692 --> 00:31:21.252
[laughs] Yeah, yeah, yeah. You have to remove those links to like kinda give it the freedom to kind of do what you need to do. So- Yes... but yeah. We'll, we'll release more soon. That's awesome. Yeah. Yeah.
215
00:31:21.272 --> 00:31:29.592
That'll, that'll be cool. We're a cerebral podcast, so people listen to us and, and sort of think really deep. So- Yeah... go- uh, we try to keep it subtle. No gong. We try to keep it subtle. Okay, fine.
216
00:31:29.612 --> 00:31:36.332
[laughs] Um, you, you guys, you guys do have evals. You talked about your- Yep... your office thing, but, uh, you've been also promoting Apex agents and complex work. Uh- Yep...
217
00:31:36.452 --> 00:31:43.612
whatever you, wherever you wanna take this, just- Yeah... how you... Apex is, is obviously in our course, uh, uh, kind of, um, a- agent eval.
218
00:31:43.652 --> 00:31:55.812
We, we supported that by sort of opening up some data for them around how we kinda see these, um, data workspaces in, in the, you know, kind of regular economy. So how do lawyers have a workspace?
219
00:31:55.852 --> 00:32:05.432
How do investment bankers have a workspace? What kind of data goes into those? And so we, we partner with them on their, their Apex eval. Our own, um, eval is... So it's actually relatively straightforward.
220
00:32:05.852 --> 00:32:11.032
We have a, a set of, of documents in a, in a range of industries.
221
00:32:11.472 --> 00:32:20.292
We give the agent previously did this as a one-shot test of just purely the model, and then we just realized we, we need to, based on where everything's going, it's just gotta be more agentic.
222
00:32:20.332 --> 00:32:27.232
So now it's a bit more of a test of both our harness and the model, and we have a rubric of a set of things it has to get right and we score it.
223
00:32:27.832 --> 00:32:39.652
Um, and you're just seeing, you know, these incredible jumps in almost every single model in its own family of, you know, Opus four, um, you know, Sonnet four six versus Sonnet four five. Yeah. We have this up on screen.
224
00:32:39.832 --> 00:32:46.392
Okay, cool. So some- you're seeing it somewhere. Like I, I forget the to- it was like 15 point jump, I think on the main, on the overall. Yes.
225
00:32:46.912 --> 00:32:55.632
And it's just like, you know, these incredible leaps that, that are starting to happen. Um, and- And, and Anthropic doesn't know it, like any... It's completely held out from Anthropic. This is not in any- Yeah...
226
00:32:55.632 --> 00:33:04.312
there's no public data, which has, m- you know, be- benefits. And this is just a private eval that we do- Yeah... and then we just happen to show it to, to the world. Mm. So you can't, you can't train against it.
227
00:33:04.532 --> 00:33:16.732
And I think it just is representative of, you know, it's obviously reasoning capabilities, what it's doing it at, you know, kind of test time compute capabilities, thinking levels, all, like the context rot issues.
228
00:33:17.252 --> 00:33:21.832
So many interesting, you know, kind of, uh, uh, capabilities that are, that are now improving.
229
00:33:22.192 --> 00:33:29.832
One sector that you have that's interesting, uh, people are roughly familiar with healthcare and legal, but you have public sector in there. Yeah. Uh, what's that like? What, what is that? Yeah.
230
00:33:29.872 --> 00:33:37.012
And, and we actually test against, I don't know, maybe 10 industries. We- Yeah... we end up usually just cutting a few that we think have interesting gains. So public sector is one.
231
00:33:37.052 --> 00:33:43.792
A lot of like government type documents. Um- What is that? What is a government type document? Is that like- Government filings... like a tax return? Like a- Probably not tax returns.
232
00:33:43.892 --> 00:33:51.552
[laughs] It would be more of what would g- the government be using, uh, as data. So- Okay... um, so think about research, that, that type of, of, of data sets.
233
00:33:51.972 --> 00:33:59.732
And then we have financial services for things like data rooms and what would be in an- Yes... investment prospectus. Uh-huh. That one you can dogfood. [laughs] Yeah, exactly. Exactly. Yes. Yes.
234
00:33:59.792 --> 00:34:09.572
So, uh, so we, we run the models, um, in now, you know, more of an agent mode, but, but still with, with kind of limited capacity and just try and see like on a like for like basis what are the improvements.
235
00:34:09.612 --> 00:34:19.992
And, and again, we just continue to be blown away by how, how good these models are getting. Yeah. I mean, I think every serious AI company needs something like that where like, well, this is the work we do.
236
00:34:20.472 --> 00:34:27.792
Here's our company eval. Yeah. And if you don't have it, well, you're not a serious AI company. There's two dimensions, right? So there's, there's like how are the models improving?
237
00:34:28.072 --> 00:34:38.772
And so which model should you either recommend a customer use, which one should you adopt? But then every single day we're making changes to our agents, and you need to know- If you regress... if you know... Yeah.
238
00:34:38.792 --> 00:34:50.452
You know, I've been fully convinced that the whole agent observability and eval space is gonna be a massive space. Um, super excited for what BrainTrust is doing, excited for, you know, LangSmith, all the things.
239
00:34:50.952 --> 00:34:59.472
And I think what you're gonna... Y- I mean, this is like every enter- like literally every enterprise. W- right now it's like the AI companies are the customers of these tools. Every enterprise will have this. Yeah.
240
00:34:59.512 --> 00:35:02.352
You'll just have to have an eval of all of your work.
241
00:35:02.772 --> 00:35:11.672
And like, we'll h- you'll have an eval of your RFP generation, you'll have an eval of your sales material creation, you'll have an eval of your, uh, invoice processing.
242
00:35:11.872 --> 00:35:22.752
And, and as you, you know, buy or use new agentic systems, you are gonna need to know, like, what's the quality of your, of your pipeline. Yeah. Um, so huge, huge market with agent evals. Yeah.
243
00:35:23.012 --> 00:35:30.132
And, and you know, I'm gonna shout out your, your team a bit. Uh, your CTO, Ben, uh, did a great talk with us last year- Awesome. Yeah... and he's gonna come back again- Oh, cool. Yeah... for, for Worlds Fair. Yep.
244
00:35:30.572 --> 00:35:36.772
Just talk about your team. Yeah. Like, you know, brag a little bit. Yeah. Like, you know, I think, I, I think people take these eval numbers and pretty charts as for granted, but no, there...
245
00:35:36.832 --> 00:35:45.592
I mean, there's, there's lots of really smart people at work doing all this. Biggest shout-out, uh, is we have, we have a couple folks, Aditya, uh, Siddharth, uh, that, that kinda run this.
246
00:35:45.672 --> 00:35:53.812
They're like a, you know, kinda tag, tag team duo on our evals. Ben, our CTO, heavily involved. Yasha, head of AI. Uh, you know, a bunch of folks.
247
00:35:53.912 --> 00:36:03.012
And, um, evals is one part of the story, and then just like the full, you know, kinda AI and agent team is, uh, is a, is a pretty, you know, is core to this whole effort.
248
00:36:03.072 --> 00:36:17.984
So there's probably, I don't know, like- Maybe a few dozen people that are like the epicenter, and then you just have like layers and layers of, of kind of concentric circles of, okay, then there's a search team that supports them and an infrastructure team that supports them, and it's starting to ripple through the entire company.
249
00:36:18.404 --> 00:36:25.024
Uh, but there's that kind of core agent team, um, that's a pretty, pretty close, uh, close-knit group. The search team is separate from the infra team?
250
00:36:25.344 --> 00:36:34.624
I mean, we have like every, every layer of the stack we have to kind of do except for just pure public cloud. Um, but, um, you know, we, we store... I don't even know what our public numbers are and, you know.
251
00:36:34.684 --> 00:36:38.064
But, like, you can just think about it as like a lot of data i- is, is stored in Box.
252
00:36:38.644 --> 00:36:47.644
And so we have, you have every layer of the, of the stack of, you know, how do you manage the data, the file system, the metadata system, the search system, just all of those components.
253
00:36:47.664 --> 00:36:55.644
And then they all are having to understand that now you've got this new customer, which is the agent, and they've been building for two types of customers in the past.
254
00:36:55.704 --> 00:37:04.664
They've been building for users, and they've been building for like applications, and now you've got this new agent user, and it comes in with a different set of properties sometimes.
255
00:37:04.724 --> 00:37:20.244
Like, hey, maybe sometimes we should do embeddings, an embedding base, you know, kind of search versus, you know, your, your typical semantic se- Like, it's just like you have to build the, the capabilities to support all of this, and we're testing stuff, throwing things away, something doesn't work and, and not relevant.
256
00:37:20.284 --> 00:37:27.764
It's like just, you know, total chaos. But, but all of those teams are supporting the agent team that is kind of coming up with its requirements of what, what do we need.
257
00:37:28.204 --> 00:37:36.144
Yeah, the, uh, we just came from, uh, Fireside Chat where you did, and you, you, you talked about how you're doing this. It's, it's kind of like an internal startup- Yeah... within the broader company.
258
00:37:36.244 --> 00:37:43.824
The broader company's like 3,000 people. Yeah. But, you know, at, there's, there's a, this is a core team of like, well, here's the innovation center. Yeah. And like the, every company kind of ru- is run this way. Yeah.
259
00:37:44.224 --> 00:37:48.244
I wanna be sensitive. I don't call it the innovation center- Yeah... only because I think everybody has to do innovation.
260
00:37:48.244 --> 00:38:01.004
[laughs] Um, there's, there's a part of the, the, the company that is, is sort of do or die for the agent wave. Yeah. And it only happens to be more of my focus simply because it's existential that we get it right. Yeah.
261
00:38:01.344 --> 00:38:07.464
All of the supporting systems are necessary. All of the surrounding adjacent capabilities are necessary.
262
00:38:07.504 --> 00:38:20.644
Like, the only reason we get to be a platform where you'd run an agent is because we have a security feature or a compliance feature or a governance feature that, that some team is working on, but that's not gonna be the make or break of, of whether we get agents right.
263
00:38:20.784 --> 00:38:27.744
Like, that already exists, and we need to keep innovating there. I don't know what the right exact precise number is, but it's not 1,000 people, and it's not 10 people.
264
00:38:27.864 --> 00:38:41.904
There's a number of people that are like the, the kind of like, you know, startup within the company that ha- are the make or break on everything related to AI agents, you know, leveraging our platform and letting you work with your data, and that's where I spend a lot of my time.
265
00:38:41.924 --> 00:38:50.444
And Ben and Yash and Diego and Tiri, you know, these are just, you know, people that, that, you know, ha- kind of across the team are working. Yeah. Amazing.
266
00:38:50.604 --> 00:38:54.584
How do you, how do you think about, I mean, you talked a lot about like kind of read workflows over your Box data. Yep. Right?
267
00:38:54.604 --> 00:39:03.604
You know, genetic search, questions, queries, et cetera, but like what about like write or like authoring workflows? Yes. I've already probably revealed too much actually, now that I think about it, so, um- [laughs]...
268
00:39:03.644 --> 00:39:07.144
I've talked about- Wherever, wherever you can, sure. Okay. Yeah, yeah. It's, it's just us. It's just us. Yeah, yeah. Okay, of course. Of course. Of course.
269
00:39:07.164 --> 00:39:17.264
[laughs] So I, I guess I would just, uh, I'll make it a little bit conceptual, uh, because again, I've already, I've already said things that are not even GA, but, but we've, we've kind of like danced around it publicly, so I've- Yeah.
270
00:39:17.304 --> 00:39:26.644
Yeah, okay. Just like hopefully nobody watches this, um, episode. No, it's tidbits for the highly engaged to go figure out like what exactly, um, you know, is, is your sort of line of thinking. Sure.
271
00:39:26.664 --> 00:39:27.664
They can connect the dots. Yeah.
272
00:39:28.024 --> 00:39:51.984
So, so I would say that, that, uh, we, you know, as a, as a place where you have your enterprise content, there's a use case where I want to, you know, have an agent read that data and answer questions for me, and then there's a use case where I want the agent to create something and use the file system to create something or store off data that it's working on or be able to have, you know, various files that it's writing to about the work it's doing.
273
00:39:52.364 --> 00:39:58.004
So w- we do see it as a total read write. The harder problem has so far been the read only because- Mm...
274
00:39:58.064 --> 00:40:07.844
because again, you have that kind of like 10 million to one ratio problem, whereas writes are, a lot of that's just gonna come from the model and, and we just like, we'll just put it in the file system and kind of use it.
275
00:40:07.884 --> 00:40:24.603
So it's a little bit of a technically easier problem, but the only part that's like, not gonna say technically hard, it's just like it's not yet perfected in the state of the ecosystem is, you know, building a beautiful PowerPoint presentation is still a hard problem for these models.
276
00:40:24.684 --> 00:40:30.464
Mm. Like, like we still, you know, like, like the, these formats are just n- were not built for- They're working on it. They're wor- they're working on it. Everybody's working on it.
277
00:40:30.504 --> 00:40:34.744
Every launch is like, well, we can do a PowerPoint now. Yeah. We're get... Yeah. Yeah. [laughs] Getting a lot, getting a lot better each time. Yeah.
278
00:40:34.764 --> 00:40:44.084
But then you'll do this thing where you'll ask the update one slide, and all of a sudden like the fonts will be just like a little bit different, you know, on two of the slides- Mm. Mm...
279
00:40:44.144 --> 00:40:47.504
or it moved, you know, some shape over to the left a little bit.
280
00:40:47.704 --> 00:40:57.084
And again, these are the kind of things that like in code, obviously you could really care about if you really care about, you know, how beautiful is the code, but at the end user doesn't notice all those problems.
281
00:40:57.484 --> 00:41:06.164
In file creation, the end user instantly sees it. You're like, "Ah, right, like paragraph three, like you literally just changed the font on me. Like, it's a totally different font, and like midway through the document."
282
00:41:06.364 --> 00:41:13.664
Mm-hmm. Those are the kind of things that you run into a lot of in the, the, in the content creation side. Mm-mm. So we are gonna have native agents that do all of those things.
283
00:41:13.804 --> 00:41:26.304
They'll be powered by the leading kind of models and labs. But the thing that I think is, is probably gonna be a much bigger idea over time is any agent on any system, again, using Box as a file system for its work. Mm.
284
00:41:26.484 --> 00:41:30.844
And in that kind of scenario, we don't necessarily care what it's putting in the file system.
285
00:41:31.024 --> 00:41:41.864
It could put its memory files, it could put its, you know, specification, you know, documents, it could put, you know, whatever its markdown files are, or it could, you know, generate PDFs.
286
00:41:42.024 --> 00:41:48.984
It's just like it's a workspace that is, is sort of sandboxed off for its work. People can collaborate into it. It can share with other people.
287
00:41:49.364 --> 00:41:59.584
And, and so we, we're thinking a lot about what's the right, you know, kind of way to, to deliver that at scale. I wanted to come into sort of the sort of AI transformation or AI sort of, uh, operations things.
288
00:41:59.964 --> 00:42:06.124
Um, one of the tweets that you, that you wanna talk about. This is just me going through your tweets, by the way. Oh, okay. [laughs] I mean, like this is like you're the- We have read your tweets... this is one by one.
289
00:42:06.164 --> 00:42:10.924
Okay. You're the, you're the easiest guest to prep for- Okay... because you, you already have like, "This is the, this is what I'm interested in." Okay.
290
00:42:10.924 --> 00:42:19.844
I'm like, "Okay, well-" Are we gonna get to like J- like February, January or something? Where are we in the, in the timeline? [laughs] How far back are we going? Can you, can you describe Box as a set of skills, right?
291
00:42:19.904 --> 00:42:27.976
Like that, that's like, that's like one of the-... extremes of like, well, if you c- you just turn everything into a markdown file- Yeah... then your agent can run your company.
292
00:42:28.596 --> 00:42:36.036
Uh, but you just have to write, find the right sequence of words to [laughs] - Yes... to do it. Oh, sorry. Is that the question? What, what- So, so, so, so I think the question is like- Okay, I'm sorry [laughs]...
293
00:42:36.056 --> 00:42:42.496
like, w- what if we documented everything- Yes... the way that you exactly said, like- Yes... um, let's get all the Fortune 500s, uh, prepared for agents. Yes.
294
00:42:42.536 --> 00:42:52.116
And, like, you know, everything's in Golden and, and nicely filed away and everything. Yes. What's missing? Like, what's left, right? Like- Yeah... y- you've, you've run your company for a decade, like- Yeah.
295
00:42:52.676 --> 00:43:03.696
I think the challenge is that, that, that information changes a week later. Mm. And because- Mm... something happened in the market for that customer or us as a company that now has to go get updated.
296
00:43:04.076 --> 00:43:15.136
And so these systems are living and breathing, and they have to experience reality and updates to reality, which right now is probably gonna be humans, you know, kind of giving those, gi- giving them the updates.
297
00:43:15.376 --> 00:43:21.196
And, you know, there is this piece about context graphs, uh, as, uh, that kind of went v- Yes, we had, yeah, yeah, yeah... very viral. Yeah. Um, on, on the pod. Yeah.
298
00:43:21.196 --> 00:43:26.456
And I, I, I was like a, I, I, I thought it was super pr- provocative. I agreed with many parts of it.
299
00:43:26.876 --> 00:43:39.996
I disagree with a few parts around, you know, it's not gonna be as easy as, as just if we just had the agent traces, then we can finally do that work because there's just, like, there's so much more other stuff that, that's happening that, that we haven't been able to capture and digitize.
300
00:43:40.016 --> 00:43:45.856
And I think they actually represented that in the piece, to be clear. But, like, there's just a lot of work, you know, that, that has to...
301
00:43:45.936 --> 00:43:53.956
You just can't have only skills files, you know, for your company because there's just gonna be, like, there's gonna be a lot of other stuff that happens. Yeah. Change over time. Yeah.
302
00:43:54.016 --> 00:43:58.416
Um- Because most companies are practically apprenticeships. Most companies are practically apprenticeships, yeah, you're right.
303
00:43:58.416 --> 00:44:05.016
Like, every new employee who joins the team, like, you spend one to three months, like, ramping them up. Yes. All that tacit knowledge- Yes... has been written down. Yes.
304
00:44:05.376 --> 00:44:08.016
But, like, it would have to be if you wanted to, like, give it to an agent, right?
305
00:44:08.056 --> 00:44:14.716
And so, like, that seems to me, like, to be- One is I think you're gonna see, again, a premium on companies that can document this- Mm-hmm... much.
306
00:44:15.116 --> 00:44:23.476
There'll be a huge premium on that because, because, you know, can you shorten that three-month ramp cycle to a two-week w- ramp cycle? That's an instant productivity gain.
307
00:44:24.036 --> 00:44:30.656
Can you re- dramatically reduce rework in the organization because you've documented where all the stuff is and where the answers are?
308
00:44:31.236 --> 00:44:41.896
Can you make your average employee as good as your 90th percentile employee because you've captured the knowledge that's sort of in the heads of, of those top employees and make that available?
309
00:44:42.156 --> 00:44:45.276
So, like, you can see some very clear productivity benefits- Mm-hmm...
310
00:44:45.396 --> 00:44:56.176
if you had a company culture of making sure, you know, your information was captured, digitized, put in a format that was agent ready, and then made available to agents to work with.
311
00:44:56.516 --> 00:45:06.196
And then you just again have this reality of, like, at a 10,000-person company, mapping that to the, you know, access structure of the company is just a hard problem.
312
00:45:06.476 --> 00:45:14.996
Is like, is like, yeah, well, you just, not every piece of information that's digitized can be shared to everybody, and so now you have to organize that in a way that actually works.
313
00:45:15.036 --> 00:45:23.376
There was a pretty good piece, um, this, this, uh, this piece called Your Company as a File Is a File System. Mm-hmm. I don't know, did you see that one? Nope. Uh, yes. You saw it? Yeah.
314
00:45:23.416 --> 00:45:31.716
And, and, uh, I, I actually would be curious your thoughts on it. Um, like, like an interesting kind of like we, we agree with it because, because that's how we see the world. And, uh- Okay.
315
00:45:31.736 --> 00:45:39.776
We have, we have it up on screen. Oh, okay. Yeah. But, but it's all about basically, like, you know, we've already s- we, we, we're already organized in this kind of like, you know, permission structure way.
316
00:45:40.256 --> 00:45:47.556
Uh, and, and these are the kind of, you know, natural ways that, that agents can now work with data, so it's kind of like this, this, you know, kind of interesting metaphor.
317
00:45:47.916 --> 00:45:56.176
But I do think companies will have to start to think about how they start to digitize more, more of that data. What was your take? Yeah, I mean, like, the company's probably like an acid compliant file system. Ah.
318
00:45:56.716 --> 00:46:02.616
Which I can be guessing boxes, right? So- Yeah. Yes. Yeah. Which you have a great piece on, but... Uh, yeah.
319
00:46:02.816 --> 00:46:09.896
Well, uh- [laughs] I, I, my, my, my direction is a little bit, like, I want to rewind a little bit to the graph word. You said that there's, that's the, the tr- magic trigger word for us. Mm.
320
00:46:09.956 --> 00:46:16.736
I always ask, what's your take on knowledge graphs? Yeah. Mm. Uh, because every, especially at every data, database person, I just wanna see what they think.
321
00:46:16.836 --> 00:46:26.156
There's been knowledge graphs, high cycles, and you've seen it all, so. Mm. I actually am not the expert in knowledge graphs. [laughs] So, so that we might need to re- You, you don't need to be an expert. Yeah.
322
00:46:26.196 --> 00:46:31.156
I, I think it's just like, well, how, how seriously do people take it? Yeah. Like, is, is, is there a lot of potential in, in the HOBI?
323
00:46:31.716 --> 00:46:41.596
Um- Um, well, can, can I, uh, understand first if it's, um, is this a loaded question in the sense of are you super pro, super con- No... super anti- I see pro-... medium in the- I see pros and cons. Okay.
324
00:46:41.695 --> 00:46:49.616
Uh, but I, I think your opinion should be independent of mine. Yeah. No, no, totally. Yeah. I just wanna see what I'm stepping into. No. I know it's a, and it's a huge trigger word for a lot of people out- Yeah...
325
00:46:49.676 --> 00:46:54.376
uh, in our audience, and they're, they're trying to figure out- Why is it? [laughs] Because they, they- You're like, "Why is this such a hot, uh- Because-... item for them?"...
326
00:46:54.456 --> 00:47:02.936
because a lot of people get graph religion. Oh. And they're like, "Everything's a graph. Like, of course you have to represent it as a graph." Well, how do you solve your knowledge, um, changing over time?
327
00:47:02.956 --> 00:47:10.796
Well, it's a graph. Yeah. And, and I think there, there's that line of work- Yeah... and then there's, there's a lot of people who are like, "Well, you don't need it." And both are right. [laughs] Yeah.
328
00:47:11.336 --> 00:47:19.336
And what do the people who say you don't need it, what are they arguing for? Markdown files. [laughs] Oh, sure, sure, sure. Simplicity. Yeah. Versus l- it's, it's structure versus less structure, right?
329
00:47:19.416 --> 00:47:29.696
I do- That's, that's all it really is... I think the tricky thing is, um, is, is again, when this gets met with real humans, they're just going to their computer, they're just working with some people on Slack or Teams.
330
00:47:29.736 --> 00:47:35.636
They're just sharing some data through a collaborative file system in Google Docs or Box or whatever.
331
00:47:36.296 --> 00:47:48.276
I certainly like the vision of most, most knowledge graph, you know, kind of futuristic y- kind of ways of thinking about it. Uh, it's just like, you know, it's 2026, we haven't seen it yet kind of play out as, as...
332
00:47:48.376 --> 00:47:51.556
I mean, I remember, do you, you remember Lee, um, in like...
333
00:47:51.836 --> 00:48:02.736
Actually, I don't, I don't even know how old you guys are, but I'll, for, for to show my age, I remember 17 years ago, everybody thought enterprises would just run on wikis. Yeah. And, uh, and that- Confluence...
334
00:48:02.776 --> 00:48:09.656
and, and, and not even, I mean, Con- Confluence actually took off for engineering, for sure. Yeah. Like unquestionably. But, like, this was like everything would be in the wiki.
335
00:48:10.276 --> 00:48:20.936
And I think based on our, uh, our, uh, general style of, of, of what we were building, like we were just like, I don't know, people just like want a workspace. They're gonna collaborate with other people. Exactly. Yeah.
336
00:48:20.936 --> 00:48:26.336
So you were, you were anti-knowledge graph. Not anti, not anti. All right. It is some of- You're not, not- I'm, I'm not, I'm not anti because I think it- [laughs]...
337
00:48:26.356 --> 00:48:36.496
I think your search system, I just think these are two systems that probably n- uh, but like I'm, I'm not in any religious war. I don't wanna be in anybody's YouTube comments on this. [laughs] This is not a fight for me.
338
00:48:36.536 --> 00:48:44.616
We, we love the YouTube comments here. We're, we're- Get in the comments. Okay. Uh- [laughs]... but like, but I, I, it's mostly just a virtue of what we built. Yeah, yeah. And we just continued down that path.
339
00:48:44.816 --> 00:48:57.280
Yeah, yeah. And, um, and that, that was what we pursued. But I'm not, this is not a- You know, kind of this is not a- It's not existential for you, great. We're happy to plug into somebody else's graph.
340
00:48:57.620 --> 00:49:09.960
We're happy to feed data into it. We're happy for agents to, to talk to multiple systems. Not, not our fight. Yeah. But I need your answer. Yeah. Graphs or nerd snipes is, is very effective nerd snipe. See? See? Okay.
341
00:49:09.980 --> 00:49:16.200
All right. This is, this is one s- one opinion and then I've- Yeah. And I, I think that the actual graph structure is emergent in the mind of the agent. Ah.
342
00:49:16.240 --> 00:49:22.319
In the same way that it is in the mind of the human, and that's a more powerful graph because it actually can evolve over time. So don't, don't tell me how to graph. I'll, I'll figure it out myself. Exactly. Okay.
343
00:49:22.380 --> 00:49:30.580
All right. And what, what's yours? I like the, the Wiki approach. I'm a- I'm actually like if, uh... You know, obviously I spend some of my time at Cognition, which, uh, you, you know very well. Yep.
344
00:49:30.860 --> 00:49:39.020
And they've had a lot of success with Deep Wiki. Yeah. It powers a lot of Devin Brain. Super powerful. And it's su- it's useful for humans, but it's, oh my God, it's useful for agents. Yes. Mm.
345
00:49:39.060 --> 00:49:47.760
Tell me if you think I'm, I'm wrong on this, but, but not much of an access control structure issue. No. [laughs] It's like the whole... You get the whole code base and everybody gets to it. Well, I...
346
00:49:47.800 --> 00:49:54.219
Before, before I speak too much- You'll love it. Yeah... there, there, there may be some enterprise controls on- Sure... uh, the enterprise Deep Wiki- Okay... offering that I'm not familiar with. Yeah.
347
00:49:54.520 --> 00:50:02.080
But yeah, I don't, I don't have any- Yeah... anything on the public side. But yeah, I, I think like al- almost like every agent should have its own Wiki that it's updating, and that's- Mm...
348
00:50:02.100 --> 00:50:12.400
persistent memory and- Yeah... uh, that is a very weak knowledge graph. Yeah. And you, you could strengthen it if you want more structure, but you may not need it. Yep. Markdown files having links in Wiki style, right?
349
00:50:12.520 --> 00:50:21.420
Yep. Very effective. Right. Lindy. Yep. I like that as a, as a just general pattern. Um, okay. So, uh, the last couple questions. Sure. But feel free to jump on in or, or if you want any rants.
350
00:50:21.700 --> 00:50:29.260
Um, I see you as a very interesting and, and unusual founder, where like you've been in the business and you're, you're both like, you're of like, of two worlds.
351
00:50:29.320 --> 00:50:36.780
Like you're of Silicon Valley, but you're also of the Fortune 500s. And like I feel like your kind of founder mode is very different from the Brian Chesky founder mode.
352
00:50:37.280 --> 00:50:47.360
And I'm just kind of curious if you have like ref- reflections on like how you operate as a founder. What would his founder mode be? Don't delegate. Ah, right. And what, how would you put me? You do delegate. Ah.
353
00:50:47.380 --> 00:50:59.759
[laughs] Okay. I, I, I, I see. The um, I think the... I, I don't know that Brian w- and I would be that far removed from each other when you get to the specifics. Okay. So there's a whole bunch that I delegate.
354
00:51:00.620 --> 00:51:13.900
90% of the work that happens at Box is fully, you know, fully delegated. We've got great leaders running, running all that stuff. It's just too much for my brain to handle, and probably 70% of the work...
355
00:51:13.940 --> 00:51:27.060
I'm gonna make up all the numbers here. Probably 70% of the work at Box, or 70, 80% of the work at Box, I only need to really look at about 5% of that for like some high leverage decisions to be involved in.
356
00:51:27.420 --> 00:51:32.500
You know, what's the marketing message that we think is gonna resonate with, with customers?
357
00:51:32.900 --> 00:51:43.280
So that's a little bit of high leverage thing that, that, that we do in marketing, but most of marketing activities I don't get involved in. What's our sales pitch? Maybe I'll be involved in that a little bit.
358
00:51:43.320 --> 00:51:55.540
Or like what's roughly the investments or push we're gonna do in certain verticals, you know. That's about 5% of like the total bandwidth of, you know, the str- the, the key areas of sales or go to market. Okay.
359
00:51:55.660 --> 00:52:06.100
So like 70, 80% of the company I can just do about 5%, and then, and then just like operationally we've got great leaders and they're gonna execute on that, and we collaborate on the 5% anyway.
360
00:52:06.120 --> 00:52:17.100
It's not like I'm just like making up a decision and saying to go and do it. Then there's this part that is like the existential part of the business, which is if we don't do this right, we're out of business.
361
00:52:17.660 --> 00:52:23.300
And, uh, by virtue of just being a founder you get kind of sucked into that part of the work because you can feel it.
362
00:52:23.320 --> 00:52:32.140
Like, this is like, like you can just see how the AI tsunami could wipe you out if you make just two, three, four, five wrong decisions in this space.
363
00:52:32.400 --> 00:52:43.300
Like couple wrong architecture decisions, couple wrong AI feature decisions, couple wrong API platform decisions and, and you might be out of the game in a year from now. And like you just feel it in your bones.
364
00:52:43.380 --> 00:52:54.280
You, you know this. Uh, like it's just like, like, like we feel this all day long in this space given what's happening. Mm. And so that, in that area it's, you can't kind of delegate in the classic sense.
365
00:52:54.320 --> 00:53:04.560
You still need to make sure you've got great leaders and strong hires and people that, that are, have high agency, 'cause they want to be able to, to own part of the, the strategy and the roadmap, or else you can't hire good people.
366
00:53:04.940 --> 00:53:12.540
But, but, you know, there's gonna be a lot of little micro forks in the road that they will compound to determine whether you succeed or fail.
367
00:53:12.900 --> 00:53:23.680
And so your kind of founder energy just like automatically draws you into, into those because, because they are the determining decisions of, of your company's future. And that's kind of where I spend my time and I...
368
00:53:23.720 --> 00:53:35.220
And you have to kind of, you know, do it in a collaborative way, again, because if you are only dictatorial and just, you know, you just w- won't eventually be able to hire the best people 'cause they won't wanna work on that environment.
369
00:53:35.420 --> 00:53:41.260
But you also just can't like abdicate all the responsibility because the risks are, are just simply too high.
370
00:53:41.300 --> 00:53:54.040
Like, and so you have to somehow obviously add some value, and so the value I add is I've seen 20 years of this business so I, I think I can kind of piece together what I expect the value propositions are gonna be and how customers will react to certain things.
371
00:53:54.500 --> 00:54:01.540
So that's what I can bring to the table. And then you have this kind of existential fear of if I get it wrong, it's all on me anyway. I don't get to blame- [laughs]...
372
00:54:01.580 --> 00:54:08.680
you know, you know, the engineer that was working on that project. Like, it's all, it's, it's, it's my fault, right? Like at the end of the day it'll be my fault if it doesn't work.
373
00:54:09.180 --> 00:54:20.500
So by virtue of, of that liability, uh, responsibility, you just get pulled into needing to make sure like it's all going a- according to, to kind of how you think it needs to end up. It's exciting.
374
00:54:20.540 --> 00:54:28.120
I don't know if, I don't know how Brian would answer that I guess, but like I, I... Yeah. It's a long essay. It's an interesting essay. People should go and compare and contrast your answer- Yeah... versus his.
375
00:54:28.480 --> 00:54:36.820
Uh, I do think that, um, systems have a way of letting entropy get to them. Yep. And you, you, if you step away for too long you need to have a way to like check in and go- Yep...
376
00:54:36.840 --> 00:54:45.400
like, "Well, do I need to come back in or are we good?" And people are gonna tell you things are good, but they're not good. Yes. [laughs] Yes. 100%. Yeah. And that's actually...
377
00:54:45.460 --> 00:54:50.720
I'm, um, I'm a fan of actually process for the, that 70 to 80%. Yeah.
378
00:54:50.780 --> 00:55:11.592
So that 70, 80%, the process is you're gonna do, uh, you- You know, a quarterly business review, and you're gonna have a brand check-in, and you're gonna do those th- like, you're gonna make sure that, that you're seeing all the, the right episodes of, of what's changing and, and h- and how it's kind of, you know, evolving and, and make sure it's kinda going the right direction.
379
00:55:11.612 --> 00:55:21.172
And then there's some areas which is like, no, it's 24/7. Like, like, I guarantee after this podcast, at 11:00 PM, I'll be doing a Zoom with Ben- Uh-...
380
00:55:21.252 --> 00:55:26.392
uh, and probably some other people, 'cause we're gonna be talking about agents and, and new platform features. And like- That's amazing...
381
00:55:26.452 --> 00:55:36.172
that's your just in the cauldron, you know, kind of grinding on, on, on that side. Yeah. Yeah. That's, uh, that's extremely, um, realistic as to- Yeah. [laughs] Yeah, exactly... like what it, what it's like. Yeah.
382
00:55:36.392 --> 00:55:39.852
And I just wanna have people hear your perspective on what it- And this is the- What do you like, dislike...
383
00:55:39.852 --> 00:55:48.672
this is this like, um, you read the post about, you know, everybody having agents running on the weekend, and, um, and it's like, uh, you know, you, you just...
384
00:55:49.472 --> 00:55:57.272
I- I mean, first of all, anybody crazy enough to come to Silicon Valley, like, we don't bring good news about the sort of, like, healthiness of our environment right now.
385
00:55:57.332 --> 00:56:02.772
[laughs] Like, like, like you, you have to- Like sleep and- You have to know what you're signing up for. Yeah. But like- Yeah... like, you know- Yes...
386
00:56:02.772 --> 00:56:11.312
there, there's a real issue which is like, shoot, do I have enough agents running? And, and- Oh, yeah. I, I made a meme that was like semi-viral for me, but- Yes... about this. Like- That was incredible... yeah, exactly.
387
00:56:11.352 --> 00:56:17.552
That's it. [laughs] And, and, and that, that was- You can't even enjoy a party these days- No, because-... 'cause you're working with your tokens and shit... you just compute out there that you're not utilizing. I know.
388
00:56:17.752 --> 00:56:25.972
What the hell? So, so like [laughs] there's a GPU. I paid for the $200. Yeah. I'm gonna spend the $200. Yeah. Uh, I'm gonna spend $6,000 out of the $200. [laughs] Yeah, exactly. Exactly. Exactly.
389
00:56:26.032 --> 00:56:33.572
We need to make Anthropic very unprofitable. [laughs] So. Yeah, yeah. We're not doing a good enough job. [laughs] Cool. I have a closing question if you- unless you have- I, I have a question. Yeah.
390
00:56:33.572 --> 00:56:42.072
I've asked this question in private before, but I'm gonna ask it again, which is, uh... It's a question that Tyler Cowen asks his guests on his podcast, which is, uh, what is the Aaron Levie production function?
391
00:56:42.772 --> 00:56:52.852
And, uh- Oh, I love that... I love this question because there are so few people that I think are good at both executing but also, like, distilling and, like, just putting good ideas into the ether. Mm.
392
00:56:52.892 --> 00:57:03.872
And you put a lot of good ideas into the ether. And so, like, what is the Aaron Levie production function that allows you, you to do that versus others? How do I get that information or- I c- I can give you- Yeah...
393
00:57:03.892 --> 00:57:12.012
an, a, a variant- Yeah, yeah... which is what goes into Aaron Levie- Yeah... and what goes out, and how does it- Yeah... churn inside. I'm just trying to think of, 'cause I mean, you know, there's some very...
394
00:57:12.712 --> 00:57:20.932
Like, I just read a lot of Twitter, uh, as well, and so, like, I just- And you've- Yeah... spent a lot of effort curating your feed. The contrast, you don't see, like, great mini essays from Brian Chesky every day.
395
00:57:21.452 --> 00:57:27.012
Uh- But you do from you. Oh, yeah. Well- And you're kind of weird in that way. Maybe he's health- So why? Maybe he's healthier than me, actually. We should just like- [laughs]...
396
00:57:27.032 --> 00:57:33.672
we should just text him to see if, you know, he's got a more- I think he does work out, right? Yeah, he got bigger muscles. I, well, here's the thing. I- [laughs]... I work out less than him.
397
00:57:33.932 --> 00:57:41.212
[laughs] [laughs] And I tweet more than him. So, so that's the... uh, that's how we're balancing things out. I'm, um, I...
398
00:57:41.352 --> 00:57:47.792
Mostly the way I just think about it is, uh, is just, um, you know, there's, there's lots of work that's happening in the business.
399
00:57:48.172 --> 00:57:57.612
I'm getting to see the, all the problems that we are running into constantly, and I'm trying to, uh, be a little bit of a...
400
00:57:58.052 --> 00:58:10.092
create a flywheel between what we're doing internally, what, what, what then we talk about, uh, getting a feedback loop on that and seeing other people's, you know, experiences of what they're doing, bring that back into the business.
401
00:58:10.852 --> 00:58:23.052
And, and so I just see the, uh, like, my job as, as, you know, hopefully being able to kinda connect the dots of, of what's going on in the world with what's going on in Box, and then I just happen to tweet about that along the way.
402
00:58:23.252 --> 00:58:32.132
Yeah. Um, because- It's all you. There's no, like- Yeah... editor. No, it's all me. There's no... Yeah. Yeah. Wow. The, uh, I got, um... There was a funny, uh, uh, my...
403
00:58:32.192 --> 00:58:39.472
I, I tried to get an internship in, um, between freshman and sophomore year of this company, and it was a, it was a film s- uh, kind of production company in New York.
404
00:58:40.312 --> 00:58:52.952
And, uh, I got the internship, and then I emailed my liaison kind of guy who sponsored me for the internship, and I said, "Hey, I'd like to do a blog of my summer internship-" Mm... "where I blog about- Mm...
405
00:58:53.032 --> 00:59:04.492
you know, the, the being an intern at a production company in New York." And about, like, a, I don't know, half a day, a day later, uh, they re- emailed me back saying they've rescinded the internship. [laughs] No.
406
00:59:04.572 --> 00:59:12.212
Um, uh, yeah. Because, because I showed a lack of judgment on, you know, professionalism, you know, or whatever.
407
00:59:12.252 --> 00:59:20.052
Like, like, just even the, the idea that I would ask that question, red flags went up of, like, who the fuck is this guy? So anyway, I, I only say that to say that, like- Mm...
408
00:59:20.072 --> 00:59:30.572
like, to me, just, like, you know, building in public- Mm... is just, like, a natural, is a natural thing. And so I, so I just, you know, go through the day. We, we deal with interesting problems. I tweet about them.
409
00:59:30.752 --> 00:59:38.852
I get information back in the process. I, I see your work. I see your work. You know, I see a bunch of folks and, and try and, you know, kind of incorporate that back into Box.
410
00:59:38.972 --> 00:59:48.272
My job is to try and connect all these things together and, uh, and make, make it useful. That's- And you're, I mean, you're the number one spokesperson, right? So you do have to be out there. Yeah.
411
00:59:48.352 --> 00:59:55.512
I, but I, I kinda would be doing it whether or not. Yeah. Like, it's like I don't really think of it as a job requirement as much as, like, I just like- [laughs]... I like social media. You're just, you're so good at it.
412
00:59:55.692 --> 01:00:03.052
Yeah. It's still hard to believe. So, like- [laughs] Okay. Sorry. Do you get up at 5:00 AM? Oh. [laughs] With coffee? Yeah. Is that your secret? He's like, how do you work?
413
01:00:03.052 --> 01:00:09.052
[laughs] Are you actually just, like, in the back of Waymos? Like, is, do you do it that way? Like, how do you do this? It's, it's, well, no. Right? It's like- It's mo- it's mostly that, though.
414
01:00:09.312 --> 01:00:23.192
It's mostly, uh, there's a, you know, I, I, I have a commute home each night. I try and see, you know, my kids most, most weekdays before I have to hop back online. So there's, like, a 20-minute window there- Yeah...
415
01:00:23.312 --> 01:00:31.812
where I can kinda, like, distill the information that's happened and- Nice... and be like, ah, is there anything I learned today that would be interesting to throw out there or anything that I saw?
416
01:00:31.852 --> 01:00:42.062
And then probably somewhere between, like... 7:30 and 9:00 PM, I finally get a chance to like look through the feed- Mm... and see like did anything crazy happen in AI?
417
01:00:42.541 --> 01:00:51.282
And, um, uh, and then that's, that will also kind of catalyze, you know, something. Mm. Yep. As like that's the best I can kind of, you know? Yeah, yeah. Respect. Yeah. Okay, [laughs] thanks.
418
01:00:52.302 --> 01:00:58.622
Uh, now I know your, you cut off is 8:00 PM. I will try to get AI news out before 8:00 PM so I can help him- Yeah... do, do his thing.
419
01:00:58.702 --> 01:01:06.621
But basically if, if I don't see it before 8:00 to 8:30, I'm not gonna- Yeah, it's, I mean- I'm not gonna be able to like quote tweet or something... maybe that. Yeah, yeah, yeah. Uh, 'cause, uh- Yeah...
420
01:01:06.621 --> 01:01:13.502
'cause then I'm back on Zoom after that. Yeah. So. I wasn't gonna plan, plan on asking this, but you've mentioned, yeah, you mentioned the film stuff. Yeah.
421
01:01:13.562 --> 01:01:22.822
And I know from one of my favorite parts of doing research on you was that, uh, you got the idea for Box from like the, the Paramount lot- Yeah... uh, pushing paper. Uh, are you a film guy? You, you a big- I am.
422
01:01:22.882 --> 01:01:34.462
I, I, I, I, I would say I used to be more of a film guy. Yeah. Um- What, what's your, what, what, what are your favorites if you have, you wanted to list off any? Kind of the classic, uh, wannabe film student classics.
423
01:01:34.642 --> 01:01:47.302
Pul- Are we talking Scorsese? You know- Tarantino? P- Pulp Fiction- Yeah... Magnolia, Requiem for a Dream. Basically like if there was an art house film in the '90s- [laughs]... uh, to early 2000s, that was my genre.
424
01:01:47.502 --> 01:01:58.742
Yeah. That got me into like, wow, wouldn't it be cool to do, you know, y- you know, film? And then I, I thought maybe I could connect digital into it. Like could you, could you do film online?
425
01:01:59.322 --> 01:02:04.002
That just seemed too hard from a licensing standpoint. And then- [laughs]... obviously Netflix, you know, kind of existed.
426
01:02:04.102 --> 01:02:09.242
Um, so I, I never quite was able to fully connect the dots on these things, but the internship at Paramount- Mm...
427
01:02:10.182 --> 01:02:20.802
um, was one kind of catalyst for starting Box, because we were using just traditional enterprise software and I was like, wow, this is like really hard to share data. Th- you know, just like files going back and forth.
428
01:02:21.342 --> 01:02:33.462
Um, but the same thing was happening in school as well, and so that all led to, led to Box basically. Um, well A24 is, uh, you know, kind of giving back this sort of resurgence of the independent film, I guess- 100%...
429
01:02:33.582 --> 01:02:43.862
um, a- in, in, in s- in the face of all the Marvel slop. [laughs] Uh, you know, I was thinking about this the other day, and A24 is, you know, uh, certainly the best, uh, e- example I'm sure of, of this today.
430
01:02:43.962 --> 01:02:55.682
But, um, you know, they just don't, you know, you, it's, it's hard to make a film, uh, like, you know, No Country for Old Men or, um, There Will Be Blood. Like, like w- what is that movie today? Yeah.
431
01:02:55.722 --> 01:03:06.342
Like, what is a brand-new movie that is just like original? You just, you, you just watch it and you're like f- what? What did I just watch? So my, my, you know, sixes movie bench is, uh, Forrest Gump. Okay.
432
01:03:06.522 --> 01:03:17.022
Which iconic in its time. Yeah. Yep. 100%. And then never again. Yeah. [laughs] Yeah. We, we did not make, we don't know how to make Forrest Gump anymore. Um- They will try it with the sequel though at some point. Yeah.
433
01:03:17.042 --> 01:03:24.582
For sure. I, I, honestly- Forrest Gump 2 in 30 years... I would be fine with it. Oh, no. No, like Forrest Gump has a kid. Like- He's still running. Yeah, ex- he's still running. [laughs] Exactly.
434
01:03:24.682 --> 01:03:34.042
Um, I think Forrest Gump has a grandkid would be like a good movie. [laughs] Like, what is the grandkid of Forrest Gump doing in, uh, in 2026? Forrest Gump goes tropical. Yeah.
435
01:03:34.282 --> 01:03:43.282
But, um, yeah, I definitely, let's l- I wanna see good, I wanna see more movies out there. You know, I'm a little bit conflicted on AI and film- Mm... because- Oh, that. Let's do that...
436
01:03:43.602 --> 01:03:58.542
well, because I, um, uh, the world does not need more slop on, on AI entertainment, but I'm kind of like in a mode where- Mm... I think that AI is, is, is gonna be, you know, generally a pure positive.
437
01:03:58.842 --> 01:04:14.782
Because if I'm a, if I was me 25 years ago in high school, for sure I would be making a full production film that had explosions and car chases and, but then there'd be like people that would show up there.
438
01:04:14.842 --> 01:04:24.142
So like I think that ability to- Mm... to just, you get to be Spielberg, you know, is, is, you know, completely amazing and, and democratizing that is incredible.
439
01:04:24.962 --> 01:04:35.522
And I, you know, I'm, I'm concerned about like how do you make sure that we still get P.T. Anderson- Mm-hmm... along the way? And, and can we make sure that those, those guys exist?
440
01:04:35.582 --> 01:04:42.222
And then interestingly, I never, and I never saw it, but Darren Aronofsky, I, I believe, has either put out or is gonna put out a, an AI film.
441
01:04:42.742 --> 01:04:54.362
You know, even some of the best artists are, are, you know, starting to adopt this. But, um, uh, but yeah, I definitely don't want to... What I don't wanna do is just be like in this like TikTok feed of just films. Mm.
442
01:04:54.382 --> 01:04:59.642
And it's just like, oh, there's a film about the car chase that does this thing. And it, it says like we don't need that. Like- Yeah...
443
01:04:59.662 --> 01:05:17.342
like, like this should be a form of entertainment and art, and let's use AI to accelerate the production process, do the really hard CG work that, that you just, you had to spend way too much money on previously to do the, you know, kinda like let's, let's use it to test out all new kind of plot ideas.
444
01:05:17.882 --> 01:05:23.502
Uh- Yeah, previz. Yeah. Exactly. Like backgrounds. And that's incredible. Like whatever. And all those things are super incredible. I still like the...
445
01:05:23.562 --> 01:05:35.102
It's very nostalgic, but I still like the idea of like this is a camera and a person, and a person that says, you know, action, uh- [laughs]... and then, and let's hopefully like surround AI around that. We'll- Yeah...
446
01:05:35.122 --> 01:05:44.682
but we'll, we'll see how that plays out. Yeah. I think, you know, so one of the things that stability AI, uh, made an impression on me was like, well, you know, at least now we can remake Game of Thrones Season 8.
447
01:05:44.842 --> 01:05:49.502
[laughs] And again, you know, uh, like, like it was meant to be, not, uh, not rushed. [laughs] Yeah.
448
01:05:49.542 --> 01:05:55.842
And then you watch, um, I have a six-and-a-half year old, and I, you know, you see a lot of these kid movies, and you're like, "Yeah, that probably will be AI."
449
01:05:56.422 --> 01:06:07.042
I don't totally know the job math 'cause I don't know how many animators there are today, but I actually think weirdly, I think we could be producing more high quality, maybe even slightly educational kids entertainment.
450
01:06:07.522 --> 01:06:10.762
Ah, yeah. And so it's, maybe that's a positive is like we could just have like more...
451
01:06:10.782 --> 01:06:22.602
Like you could just have a Pixar for like, you know, things where kids learn stuff, and it used to be these like very, you know, lo-fi, uh, you know, kinda lesson things. I mean, we had Teletubbies, you know? Yeah.
452
01:06:22.642 --> 01:06:24.182
That, [laughs] that was so slow. Exactly.
453
01:06:24.202 --> 01:06:37.122
So, so we, we, we could have way more of that and, and maybe every animator that today is making a Pixar film is now, you know, we're like we fragment that out and, uh, but now they're responsible for more content and they've got AI agents running.
454
01:06:37.182 --> 01:06:47.262
So like, so, so I think there's some optimistic scenarios on the entertainment side is like there's a lot of great use cases for being able to do, you know, generative media. Yeah, yeah. Um- Edut- edutainment as well.
455
01:06:47.842 --> 01:06:53.942
I guess one question I have is it's kind of like a self-serving one and almost like an advice-... uh, side of the, the, the, the question.
456
01:06:54.282 --> 01:07:03.742
One of the things I just, uh, really enjoyed, uh, researching you was that, uh, Michael Arrington had some influence in The Box journey because you went to his house party. Yes. And then that's how you got funding. Yes.
457
01:07:03.742 --> 01:07:11.062
[laughs] One of Latent Space's... [gentle music] That's a deep cut, right? Yeah, very deep cut. [laughs] That's a '06 deep cut. Yeah. Uh, do you, I mean, do you wanna tell that story?
458
01:07:11.302 --> 01:07:17.022
I don't know if you've told it that much. It's very much of a story. Yeah. Uh, 'cause it probably- It's like a random intro, right? Like [laughs] Um, well, it was just he used to have house parties. Yeah.
459
01:07:17.022 --> 01:07:25.002
Uh, TechCrunch had, had these house parties, and, and it was, um, probably no different than somebody's doing a house party in SF. Uh, you know- Just go. Yeah...
460
01:07:25.002 --> 01:07:30.702
yeah, you just go, and you meet the VCs and founders, and, like, I'm gonna make up examples, so I don't want to...
461
01:07:30.742 --> 01:07:36.902
Like, you know, there'd be, like, Chad Hurley over there pitching his, you know, YouTube to people, and like, uh, like, that's just, like, how it worked.
462
01:07:36.942 --> 01:07:41.942
And it was just like, wow, like, that was this era where all these new companies were, were emerging.
463
01:07:42.362 --> 01:07:51.162
And I met, uh, our first investor, uh, in Silicon Valley at one of these house parties, Emily Melton, who then brought us into DFJ. DFJ. Yeah. That became our Series A.
464
01:07:51.222 --> 01:08:00.782
So that was all because of Arrington's, uh, backyard party. One of my aspirations for Latent Space is to be as [gentle music] helpful, influential, whatever, as TechCrunch was- That's awesome... back in the day. Yeah.
465
01:08:01.262 --> 01:08:09.662
What would a new TechCrunch today look like? You know, what, what, what- Yeah... what should I, what should I do? [laughs] I think- There, there used to be TechCrunch Disrupt. Yeah. You know, I could do that- Yeah...
466
01:08:09.662 --> 01:08:13.502
with my conference, but I haven't done it yet. Well, I mean, I think, um- Is that useful? I don't know.
467
01:08:13.902 --> 01:08:22.102
Uh- [laughs] Well, you know, actually, interestingly, I would, I would argue Disrupt came after the period that was the- Yeah... was that deep cut period. Okay.
468
01:08:22.202 --> 01:08:32.102
So, so, I mean, D- D- Disrupt, you know, ended up being, you know, s- you know, catalyzing, I don't even... I think Cloudflare launched at Disrupt. Yes. Is that the story, right? They were runners up. Okay, okay.
469
01:08:32.202 --> 01:08:43.442
So, like, so, like, I think anytime, anytime you can be in a, a launchpad is, is just great because it draws in people that are- That's what I'm trying to do... in that, in that creative moment.
470
01:08:43.902 --> 01:08:49.482
And whether it needs to be a contest or, or just, like, everybody gets, like, five minutes and you're fundraising. I mean, I... Who knows?
471
01:08:49.802 --> 01:08:54.962
But, but I mean, for what it's worth, like, I don't really have that much advice 'cause I think you, you're, you're already doing it effectively.
472
01:08:55.002 --> 01:09:05.122
Like, I, I just, like, watch the YouTube videos late at night, um, uh, from the events. I haven't been to one of your events, but, like, from the, from the camera angles, it looks like everybody's there. Yeah.
473
01:09:05.162 --> 01:09:06.342
So, like- Trying. Trying...
474
01:09:06.382 --> 01:09:16.502
what's great is that people are gonna be in the audience as, like, two random people, and they'll be like, you know, the next, the next big AI company will come from, you know, people coming to a meetup- Yes...
475
01:09:16.522 --> 01:09:25.142
'cause they were like, "Oh, I came in from Chicago," and, "I'm, uh, from, you know, Poland, and let's go do a startup." Yeah. Like, that's the magic of- Thanks, uh-... of the Valley. Thanks, Parthi.
476
01:09:25.222 --> 01:09:31.962
Founder and co-founder at AIE. Oh. And, and I know of at least one marriage that's, that's, that's come- Wow, you have marriages already? Yeah. [laughs] I don't... I never heard that about TechCrunch.
477
01:09:32.002 --> 01:09:39.362
That's my- There we go... that's my favorite KPI. Wow, we have AI marriages- [laughs]... at the, at the AI engineer conferences. Yeah. These are like humans, to be clear. Okay. [laughs] Exactly. Yes. Yes.
478
01:09:39.382 --> 01:09:44.882
That's a very good clarification. I like that you have to check. Yes. Yeah. That's a very good clarification. Uh, no, but I, I think you have... You're, you're insightful.
479
01:09:44.922 --> 01:09:56.502
You're a business leader with, like, a lot of thoughts on media, so I just figured I would- I mean, media is such an interesting space right now because, because, uh, you know, with the go direct model, every company is gonna have to be a media company.
480
01:09:56.582 --> 01:10:05.062
Like- You are going to... You are the OG go direct. Yeah, but, but, but, you know, we- we're, we're still, like, like, I, I think what, what you guys are doing...
481
01:10:05.262 --> 01:10:14.482
And I don't even know all the overlapping relationships, but, like, I watch your guys' videos of your events, watch your event videos, but, like, it's clearly, like, this is the new format, right?
482
01:10:14.622 --> 01:10:25.382
Companies have to become channels to communicate with audiences. Yeah. I think the resurgence... Uh, resurgence maybe is a bad word 'cause it implies a decline, but, like, DevRel is hot. Yeah.
483
01:10:25.422 --> 01:10:34.502
Like, the hottest thing of all time right now. I... Like, if you could produce a fricking factory of DevRel people, like, there's just, like, unlimited jobs right now on the other end of that. Yeah.
484
01:10:34.622 --> 01:10:43.822
Um, 'cause we're gonna... Everybody needs their services and APIs to be used by agents, and so we have to all find a way to, like, like, "Hey, look at me." Like- Yeah...
485
01:10:43.842 --> 01:10:52.542
like, "Agent, over, uh, please come over here, agent," and that's gonna... That's a content game. Like, how do you get the agents- Yeah... to see your stuff- Yeah. There you go... and know your APIs? Yeah.
486
01:10:52.562 --> 01:11:02.922
And, like, this is, like, a new world that- Yeah... that we are in, and, uh, it's gonna be a... It's, it's gonna completely be a digital marketing, you know, kinda world that we're in. Yeah.
487
01:11:03.022 --> 01:11:13.842
Uh, for what it's worth, I'm trying to help by doing little writing boot camps and basically turn into a DevRel boot camp, um, where, you know... Well, it's a demand and supply problem. There's this huge demand. Yeah.
488
01:11:13.862 --> 01:11:22.262
There's no supply- Wow... until this increases supply. Why is there no supply? The one- The really good ones work for themselves. Uh-huh. Creator economy screwed- Yeah... screwed you over. So I see.
489
01:11:22.282 --> 01:11:29.182
[laughs] So, so Substack and- Yes... YouTube payouts and... That's... Is that really it? They're making- Patrion. Yeah. Like, the, the most- Wow...
490
01:11:29.222 --> 01:11:37.562
talented guys are making, you know, millions and just working for themselves. Why do the work for you? That's not good. We don't want them to make that much money. [laughs] [laughs] We need to be able to hire people.
491
01:11:37.642 --> 01:11:43.962
What... I mean, I think, I think, like, you know, do, do what some companies are doing. You know, I'm not saying it's my situation exactly, but, like, give them equity, and, like- Uh-huh...
492
01:11:44.022 --> 01:11:51.562
you know, it should probably would be worth more, uh, just, like, sort of helping them out. Well, they are getting... Oh, uh, sorry. As full-time employees or not? Uh, part-time. You need full-time. I'm part-time.
493
01:11:51.922 --> 01:11:58.802
Yeah, but, but you're, you're- [laughs]... you're N of one. Like, we, like, also people that are full-time. Yeah, yeah. My classic joke or- Yeah...
494
01:11:58.822 --> 01:12:07.582
or, like, observation was, like, this was when HubSpot bought, like, their... They bought, like, a newsletter business. Uh, and then they bought the My First Million, like, the, the, the sort of podcast. Oh. Dharmesh.
495
01:12:07.642 --> 01:12:15.922
You must know Dharmesh, yeah. Yeah. Um, so he's, like, obsessed with this guy. Okay. So, so my conclusion was, like, every company must either build or buy a media company. Yes. Right? And until you...
496
01:12:16.242 --> 01:12:24.642
unless you realize that, that you have to take it that seriously, that you are running a media business in your company- Yes... you will never be, uh, good at it. Yes. Yeah. 100%. Yeah. Yeah.
497
01:12:24.682 --> 01:12:30.942
No, we're, we're very much taking that seriously, but, but still, and yet DevRel, I mean, I gotta do one plug. I don't... Me plugging in all this- Go ahead. Please, please, please, please, please...
498
01:12:30.962 --> 01:12:37.402
I was like, "We're hiring in DevRel." Yeah. Like [laughs] like, please. No, we need all engineers here. Yeah. Okay. Like, yeah. And- And, like, you've made it, like... And I just said ev- every agent needs a box.
499
01:12:37.462 --> 01:12:41.702
Like, let's go. Let's go. Thank you. No, that, that's the headline- Nice. [laughs]... and we are hiring DevRel to make that happen.
500
01:12:42.062 --> 01:12:51.782
Uh, but yeah, I think DevRel is, like, the future job, so we're all just gonna be doing DevRel of, in some form. Okay. Yeah. I mean, what is FD- Developers are ruling the earth, yeah. What is FD? I don't know.
501
01:12:51.902 --> 01:12:58.294
Um- No, it's, it's DevRel. Yeah? Okay. No, you just, you're going to a company- It's, isn't it just like glorified consulting? That's, that's the downside. Sure. Yeah.
502
01:12:58.314 --> 01:13:07.854
I mean, I guess nobody can, like, actually d- you know, d- fully define this. [laughs] But, um, uh, but I think it's, it's m- it's micro DevRel. Got it. Like, you're in the company, you're helping them with the services.
503
01:13:07.894 --> 01:13:14.214
Yeah. You're doing a little bit of extra implementation. Yeah, yeah. Um, but, uh, but yeah. So it's, uh... I, I think we're all...
504
01:13:14.614 --> 01:13:24.174
You know, the thing that's gonna happen on the ledger of software is we're gonna produce far more output of code, and thus features per dollar.
505
01:13:25.174 --> 01:13:37.134
But on the other end of this, we're gonna actually end up spending probably just as much on how do you get all of that stuff to the customer, and it's gonna create a new set of roles that we are all doing, partly because I...
506
01:13:37.234 --> 01:13:47.914
either because there's so much choice, now you have to kind of fight for attention there, or because the stuff is, is just changing so quickly that you have to technically help your customers along the journey. Yeah.
507
01:13:47.974 --> 01:13:54.934
So, so I just think, like... I, this is why I, I f- I always laugh when, you know, people say, "You don't need to be an engineer, don't do computer science."
508
01:13:55.034 --> 01:14:06.674
I actually think, like, that is, like, still one of the most protected job categories because things are only getting more technical, things are only gonna get harder, and anybody in a technical position is in the best position- Yeah...
509
01:14:06.714 --> 01:14:17.134
to get agents deployed, get them built, get them adopted, build the, the, the custom code software to the, for the IT system, all of that. Yeah. So, yeah. Yeah.
510
01:14:17.174 --> 01:14:29.374
My, my classic founding story of, like, why I picked AI engineer as a title and as, as a, as a theme for this podcast, as a theme for my conference was, um, back in tw- like early 2023, uh, someone non-tech came to me and said, like, "I'm all in on AI.
511
01:14:29.434 --> 01:14:38.274
What should I do?" And I was like... I just looked at her, I was like, "Goddammit, there's nothing you can do." Like, "Engineers are about to get so much more powerful than you." Uh. "You don't even understand."
512
01:14:38.394 --> 01:14:44.794
[laughs] Tell me this. Should she go and then learn now? No, I didn't, I didn't say any of that to her, just- Oh, oh, okay, okay, okay. Yeah, yeah. Okay. I, I'm not, I'm not that honest.
513
01:14:44.814 --> 01:14:50.334
Well, I ho- I hope somewhere out there- [laughs]... she, she did, went to some online academy- Exactly, learn to code... and became a... Yeah, yeah.
514
01:14:50.334 --> 01:14:56.794
[laughs] But there, there's a lot of people, like, there's a lot of people who believe AI too much and then they're like, "Well, you don't need to learn to code, so I won't learn to code." Yeah.
515
01:14:56.834 --> 01:15:04.594
And then there's, there's, like, there's a bunch of us who are, like, just in that sweet spot of, like, we can code and we can wield AI a thousand times more effectively than you can. Yeah.
516
01:15:04.654 --> 01:15:07.534
And, like, well, who's gonna win here? Like... [laughs] I think, I think I...
517
01:15:07.554 --> 01:15:17.314
This is another, uh, a tweet, but it was like the observation that, like, really software engineering for the past 30 years was the primary career track for, like, technical high agency people that wanted to have a large outsize impact on the world.
518
01:15:17.574 --> 01:15:20.694
Yeah. And, like, software was a means to, you know, do that, right- Yeah... effectively.
519
01:15:21.114 --> 01:15:28.374
Um, and so, yeah, with AI, is it like that, uh- And, and for a- AI is gonna eat software engineering or is software engineering gonna eat all other kind of domains and disciplines? You're...
520
01:15:28.414 --> 01:15:36.194
Those pr- same principles then get applied to every other- And the same people, right? Yeah, exactly. Yeah, yeah. I mean, G Team engineering is that- 100%... and everything else, yeah. Well, this is the...
521
01:15:36.974 --> 01:15:43.954
Y- you know, uh, anybody who believes that an enterprise, and I'm, I'm, I'm mixed on the, I'm mixed on this, is...
522
01:15:44.174 --> 01:15:55.754
Uh, but if you believe that an enterprise is going to build its own software for all of its problems, then you must be the most long on computer science, you know, as a discipline of all time, because guess what? Hmm.
523
01:15:56.074 --> 01:15:59.334
Most of the economy does not have enough engineers- Yeah...
524
01:15:59.374 --> 01:16:10.614
to then maintain all those systems, to update the, all those systems, to figure out the, the relationship between the business problem and what the code needs to do to go and actually manage that.
525
01:16:10.934 --> 01:16:18.934
And so, so, like, that's a, that's a very pro-engineering job argument of what the future's gonna look like. I'm still...
526
01:16:18.994 --> 01:16:27.134
Again, I go back and forth on like are you gonna really build all these things versus n- no prepackaged software? But no matter what, there's gonna be 10 to 100 times more code. Yeah.
527
01:16:27.174 --> 01:16:37.734
So I think you can be very long engineering right now as just a, you know, purely on the dimension of, of software's gonna become increasingly more important once agents are, are, you know, turning everything into software.
528
01:16:38.194 --> 01:16:47.554
Yeah. All right, three software guys say software engineering. Okay. [laughs] Not biased at all. [laughs] Okay. But, uh, Aaron, you're an inspiration. All right, thank you. It's such a pleasure. All right, get a beer.
529
01:16:47.934 --> 01:16:57.674
[outro music]