Latest / Newcomer Pod / Barry McCardel on Why Everyone Is Copying Palantir’s Playbook (And Getting It Wrong)
Transcript
- 0:00If there is not 10s of millions, hundreds of millions of dollars
- 0:04worth of seats for us to sell because everyone's been laid off
- 0:07the world. Like, like, like I need, I need
- 0:09to own a lot of guns. Like the least of my problems is
- 0:13my fucking SAS business. Like this is what it looks like
- 0:15to live in a Republic. It's messy.
- 0:18You, you, you. Like so it requires on the part
- 0:20of other actors to sort of interpret and what the will of
- 0:24democracy is beyond just like difference.
- 0:26Like every generation has felt broken in some way.
- 0:29I always think it's funny when people we've never been more
- 0:31divided. Bitch, we fought a civil war.
- 0:34Palantir is one of the most mysterious companies in tech.
- 0:37If you're an outsider, it's a company created by Peter Thiel
- 0:39that's gathering intelligence for the deep state.
- 0:42If you're an insider, it's an overvalued enterprise software
- 0:44company that's heavily reliant on human consultants.
- 0:47Today I talked to Barry McArdle, CEO of HECS, which he says
- 0:51aspires to be the cursor for data.
- 0:53He was a long time forward deployed engineer at Palantir
- 0:56and saw the company from the inside.
- 0:58We talk about Palantir subcontractor Anthropics, fight
- 1:01with the government, and argue about how much tech companies
- 1:04should defer to the President of the United States in a
- 1:06democracy. But first we also talk about
- 1:09what it was like working at Palantir, which he called
- 1:11mundane, while he simultaneously talked about projects in unnamed
- 1:16Middle Eastern countries. We dug into the forward engineer
- 1:19craze that Palantir pioneered, and these days every tech
- 1:22company wants to copy. Barry defends the model but
- 1:25doesn't actually use it at his own company.
- 1:28Welcome to the Newcomer Podcast hosted by me, Eric Newcomer.
- 1:30I am the author of the Newcomer Newsletter, which covers the
- 1:33business and people in the startups and venture capital
- 1:36industry. Check us out at newcomer.co.
- 1:39Thanks for listening. Here's my conversation with
- 1:41Barry McArdle. Well, literally I heard today
- 1:51that we didn't get invited to perplexity some perplexity event
- 1:57because they're mad that they were the most shorted company on
- 2:00your fault you're. Just asking the question.
- 2:02Yeah, you put me up to the stunt.
- 2:03Apparently, yeah. I thought, I think it's an
- 2:06interesting question. Who would you short?
- 2:09I think we're all, I think it's good that you're not able to
- 2:12short start-ups 'cause I think that would get into a very
- 2:15pernicious and shitty game. And imagine if venture funds
- 2:19were like hedge funds where they've got long positions and
- 2:21they've got short positions and they're like hedging, right?
- 2:23That'd be fascinating. The All right, I'm here with
- 2:27Barry McArdle. Welcome to the newcomer podcast,
- 2:30CEO of Hex. He's been a long time at
- 2:32Palantir. And as someone who's in my ear
- 2:35with fun ideas and on Twitter, Assassin and Spicy.
- 2:39So I expect you to live up to your Twitter persona on the
- 2:43podcast. Sparse.
- 2:44Sometimes I tweet a lot and then I'll go quiet for like months.
- 2:48You had a what your most viral one was on the subject we want
- 2:52to talk about. I don't want to butcher it.
- 2:53It's the IT was. It was, it's only, it's only for
- 2:58deployed if it's from the Palantir, a region of France.
- 3:00Otherwise it's just sparkling sales engineering I think was
- 3:03the. Exactly.
- 3:04So I, we want to start off with Palantir, I think from 2 levels,
- 3:07one sort of, I don't know the YouTube show level, which is
- 3:11like, I don't know, people hear Palantir and the headlines and
- 3:14like what to make of it. And then we'll do our nerdy tech
- 3:17show question, which is I feel like every startup in the world
- 3:21has learned that you can actually be a consulting company
- 3:24if you call it a forward deployed engineer.
- 3:27And I don't think will you. Be punished on multiples, you'll
- 3:30be rewarded. It's like, oh, all of a sudden
- 3:32it's good. Like, throw humans at it.
- 3:34Yeah. But let's start with like the
- 3:35spooky YouTube. And, you know, I mean, I, I just
- 3:38think a lot of people on the Internet hearing about Palantir
- 3:40just have a like, this is this is the deep state or like,
- 3:44having spent a lot of time, what, like 4 1/2 years, five
- 3:47years, five years. Yeah.
- 3:48Yeah. What Yeah.
- 3:49Is is Palantir the big bad? No, I, I and I'll just, I'll
- 3:54just claim by I haven't, I haven't been there in a few
- 3:57years now, but. You stay close to and you know,
- 3:59the whole network of people starting companies.
- 4:01Yeah, yeah. And I, I'm some, a lot of people
- 4:03there. I think the, the interesting way
- 4:06we understood ourselves then, and I suspect it's still very
- 4:08true now. And we, we said this in almost a
- 4:11slightly diminutive way. It was like we're a data
- 4:13integration company. Like it turns out that like all
- 4:16these companies in the world that these institutions or
- 4:20private companies and public institutions that you could, you
- 4:25know, wanted to be able to operate better.
- 4:28It really the the problems I'll come back to like some form of
- 4:30data integration. Like you actually have a lot of
- 4:32the data like available in some way.
- 4:35What you're saying is. We're, it's actually a really
- 4:37boring company with the CEO that says and sometimes we have to
- 4:39kill people. There is, I mean, I, I'll, I'll
- 4:42be honest, there is a kind of boring core to it.
- 4:44Like there were times where it was kind of like we're kind of
- 4:46just doing a lot of data plumbing and it, it obviously
- 4:49the brand is incredible. And I think the leadership there
- 4:51has done a great job developing and, and nurturing that brand.
- 4:55And look, look, I, I'm not saying that the results aren't
- 4:58there. It's a fabulous company.
- 4:59It's a company I'm very proud to have been at.
- 5:01I'm proud of almost feeling. Like like $300 billion.
- 5:04I can't look at the stock price now because I sold most of my
- 5:06shares at less this less than the stock price.
- 5:09I have a very discordant relationship with it.
- 5:12But yeah, like I I think at the core there's some really boring
- 5:17stuff that we just got really right.
- 5:19And the initial brilliance of Palantir in the early days was
- 5:22the ontology. It was this kind of insight of,
- 5:24you know, the, the, the slightly hagiographic version of it is
- 5:28like, you know, early team was going, the founders were going
- 5:32to these three letter agencies and sort of realizing that the
- 5:35problems they had. And you know, the mythos was a
- 5:38little bit like the, the reason we missed 911 or one of the
- 5:40reasons you kind of missed events like that is 'cause you
- 5:43actually had all the information somewhere, but you weren't able
- 5:46to get the complete picture. And so by being able to
- 5:49integrate it and build the ontology on top of it, OK, now
- 5:51you can see the whole picture and you can sort of make it
- 5:53something that humans can reason about and understand and engage
- 5:57with. And that shape of dynamic is
- 6:01true in the government, you know, intelligence community,
- 6:04military, also very true in a lot of private companies.
- 6:06And so a lot of what I did when I joined, I joined when we had
- 6:08like a very nascent commercial business.
- 6:11So I started my career in consulting.
- 6:13I'd been like on the really nerdy data tech side of my
- 6:16consulting projects. My friend who worked at Pouncer
- 6:19was like, you just come do this at like for real.
- 6:22And I started, we didn't really have a commercial product at
- 6:24all. And most of what we were doing
- 6:26was just sort of going around and doing effectively these
- 6:28consulting projects, integrating data, kind of stitching things
- 6:31together, giving people a finally, like a complete picture
- 6:34of their organization and the problems they had.
- 6:36And that by itself was really valuable.
- 6:38Because they would buy tech services but then not get the
- 6:40most out of them, stitching them up with everything.
- 6:42Yeah, you'd have, you'd have like the that, that era, you
- 6:46know, 10 years ago, you'd have a lot of people building these
- 6:48like data lakes. I was like, oh, we'll get all
- 6:49our data in the data lake. I was like, yeah, you land all
- 6:51the data in this S3 bucket basically or HDFS cluster.
- 6:55And it's like, now what we have the data, it's like, yeah, you
- 6:59have it. Are you use?
- 7:00Are you able to make better decisions because.
- 7:02Of it and and palander would help them like visualize it and.
- 7:04Yeah, it wasn't even viz. I, you know, it's funny 'cause
- 7:06people picture it like, oh, this analysis with the viz stuff, our
- 7:09viz was always like pretty mid. Is, is really the integration
- 7:13behind the scenes, which I know sounds kind of sexy and unsexy
- 7:16and vague, but like it's literally like, OK, you've got
- 7:18these five different data sets. They're all sort of talking
- 7:21about the same thing. You want to get a complete
- 7:23picture between them? How do you integrate them?
- 7:25What are the relationships between these?
- 7:27How do I see an event or an incident that's happening here
- 7:30and go understand what are the other times something like that
- 7:32has happened? What were the resolutions?
- 7:34And is there a shared link? Like that is a lot of the kind
- 7:37of thing. And it's it's not like the
- 7:38actual like algebra behind the scenes is like, you know, it's
- 7:43just sequel queries. It's like fairly straightforward
- 7:45joins and stuff. And like every once in a while
- 7:47we'd get wild and do some regressions, but it's like.
- 7:50Really the reasons for deployed is you have to understand the
- 7:52customer enough to really understand the.
- 7:53Logic of this model around it then and, and I think the four
- 7:58deployed thing, this, this kind of gets into the, the, the role
- 8:02and what people misunderstand. Like when I joined, you know, we
- 8:06had these 4 deployed engineering teams and, and they still have
- 8:09them. At the time it was really like
- 8:12we didn't have a product. We were hacking a lot of stuff
- 8:15up there was that was the whole. Critique, apology over years it
- 8:17was it's just secretly A consulting like internally we'd
- 8:21we'd like. Take issue with that.
- 8:22Like I remember these emails someone of the guys in
- 8:24leadership would sign and being like, people don't understand
- 8:27us. They think we're just a services
- 8:29company. Like we'd kind of look around
- 8:30and be like, I don't know, it feels like we're doing services.
- 8:32I'd read this e-mail from a windowless conference room and
- 8:35so town where I'm like, basically I'm like hacking
- 8:38together an Excel spreadsheet to dump into a Jupiter notebook so
- 8:41I can export the results into our front end app builder to
- 8:44make it look like we did software.
- 8:45It was like, I don't know, it feels pretty servicey.
- 8:47But what wound up really working was you go and effectively do
- 8:51services. And at the time we didn't have a
- 8:55core platform really to speak of in the commercial side that we
- 8:58were doing these services on top of.
- 9:00But over time we built that platform leverage.
- 9:02But you go and you basically do these services.
- 9:04You go in and say, hey, we're going to solve your hardest
- 9:06problem. You know, Karp, Dr. Karp would
- 9:09meet Aceo or some leader and say, let me send you a team.
- 9:13They'll come in, just tell them your hardest problem.
- 9:15They'll come in and help try to solve it.
- 9:17And you'd kind of have like 3 months, the canonical pilot
- 9:20length to go and work super hard and fly around a lot and try to
- 9:23solve it. And what would wind up happening
- 9:26is you'd, you'd have to build new things in the field.
- 9:29And then like most of those things wouldn't work out, Like
- 9:31the pilot wouldn't convert, it wouldn't be valuable enough,
- 9:33whatever. But some of those things would
- 9:35go on to be multibillion dollar business, you know, like just
- 9:39products and and foundational parts of the platform and the
- 9:43whole model of forward deployed engineering that really, really
- 9:45worked. And the thing that people kind
- 9:47of miss is it's a radical deference to the field on what
- 9:52to build. Like most organizations, they're
- 9:54like, OK, we've got a road map and all these slides.
- 9:57We know the customers essentially.
- 9:59And I think especially a lot of founders like I, I'm like this
- 10:01sometimes, like I know what to build.
- 10:03We gotta go build this thing. We just gotta go find the
- 10:05customers. That's.
- 10:06Part of what you're selling the investors, it's like I have the
- 10:08vision. And the reality is like what
- 10:10really worked, at least in my time there, the way it worked
- 10:13was you just hired a lot of really smart people, kept the
- 10:16bar really high, gave them freakish, irresponsible amounts
- 10:19of ownership and autonomy. We were a bunch of like mid 20
- 10:23somethings being put in charge of these like really high stakes
- 10:25deployments and budgets. And it was just like.
- 10:28In some ways, you know Mckenzie's go to that too, where
- 10:30it's like, you speak for the firm, right?
- 10:31Or you. Kind of.
- 10:32I mean the McKenzie model's much more pyramidal and like command
- 10:36and control. You're improved at this top, but
- 10:38you get to speak you you set up a way that sort of your youngest
- 10:41people can sort of say the firm says blah blah.
- 10:43Blah, I guess, but those people are are parroting the firm's
- 10:46talking points, right? No one's hiring a 25 year old.
- 10:48No one's hiring McKenzie to get 25 year old coming in and
- 10:51telling you what you should do about your business strategy.
- 10:53You're hiring McKinsey because they ostensibly have this
- 10:56centralized expertise and they've got a practice around
- 10:58your thing. You're a telco.
- 10:59You're hiring someone from their telco practice to come in and
- 11:02they've got best practices and there's a lot of different ways
- 11:05to interpret. What you're hiring because the
- 11:06partner has inside information then everybody else is slowing
- 11:09down dresses it up in sort of process and it's like I know the
- 11:13answer we need to get to you like explain why we.
- 11:15There's a lot of different mental models of what consult,
- 11:18what job to be done consulting actually does.
- 11:21There's like the the like one that they'll advertise, which is
- 11:24we're the strategic advisors. We're here to provide our
- 11:27insider business insight and business expertise to help our
- 11:30partners make better decisions. Like, OK, yeah.
- 11:33But like there's different mental, there's different like
- 11:35sort of different versions of this.
- 11:36There's one that's like they work with kind of everyone and
- 11:39yeah, they teams don't talk to each other, but they're all kind
- 11:41of gets distilled into best practices.
- 11:43And, you know, it's kind of like how you can go get make sure if
- 11:46you're an executive, you're doing the right things.
- 11:49The other version of it and the version I saw in my brief stint
- 11:52in management consulting before Palantir was we were hired one
- 11:55time onto a project where we came to learn there was another
- 12:00consulting team from another firm working for another one of
- 12:03the VPS of the company on effectively the same problem.
- 12:07And these two VPS both had, they were like fighting about what to
- 12:10do. And they both hired consulting
- 12:12firms. And we were basically told, this
- 12:17is my theory from the VP, like I need you guys to build a deck,
- 12:20basically build a deck like a slide deck to back that up.
- 12:24And then we came to learn there was another VP who had told
- 12:26their firm that. And it was like we were both
- 12:28like the proxy army for these warring warlords, like these
- 12:33factions. And it was very weird and it was
- 12:36kind of like, what are we doing here?
- 12:37And like it. Says something about like
- 12:38intelligence. It's in the world of models, you
- 12:41know? Yeah, it's just like you're just
- 12:43accumulating your argument when really the the argument is not
- 12:46about the reason. So, so you can kind of, I
- 12:48actually thought about that more subsequently and I'm like, there
- 12:51is something kind of pure about it, which is like it's a war,
- 12:53it's a battle of ideas and you know, but I don't know.
- 12:57Anyway, the at the whole point is that pound here we were going
- 13:01in and you know, we would get attached to these problems and
- 13:05then try to work backwards to like, well, what's the product
- 13:07that could solve this? And sometimes it was kind of a
- 13:09hacky thing, like we just threw together this dashboard that
- 13:11lets you see this joint data set we have.
- 13:13So now you can correlate these two things.
- 13:16But but what actually wound up happening in mid twenty 10s is
- 13:19we had a lot of cases where we were coming into these
- 13:22commercial companies and they wanted some like higher order
- 13:25thing. They wanted like, well, we want
- 13:26to see an analysis about that, Sir.
- 13:28We want a machine learning model that shows us the correlation
- 13:30between that Sir, we want to see whatever and then we would have
- 13:34to go build and and the four deployed engineers would have to
- 13:36go build and kind of hack up a lot of infrastructure behind the
- 13:39scenes to be able to try to pull all that data together.
- 13:41And then you turn that into a product.
- 13:43Then we turn that into a product.
- 13:44So there's a team, the deployment in Europe that
- 13:47famously kind of like started productizing this like well
- 13:49we're writing these same shape of spark job orchestration
- 13:53things over and over again. We should build a thing to help
- 13:56do that. And then that sort of started
- 13:59snowballing and within that deployment and then spread.
- 14:02And one of the really cool things about the forward
- 14:04deployed model at Palantir that I saw was it was almost like a
- 14:06community driven adoption. Like you would see things go
- 14:09viral where you'd have one deployment go and invent
- 14:12something and then another deployment would pick it up
- 14:14because it's on GitHub and it to the once we sort of had a core
- 14:18platform and it got, we got better at that.
- 14:21Like, oh, like your deployment built that thing.
- 14:23Cool. Like can I, you could just
- 14:24install it and it would work on your platform too sick.
- 14:27And then you, you know, it's open for PRS so you can be
- 14:30contributing code back. And then these things would gain
- 14:32momentum. And someone who started a
- 14:35project working as an FDE on a thing and they built this thing
- 14:39just to solve this customer, maybe six months later, they're
- 14:42basically on. They're basically a dev team now
- 14:44that's kind of delaminated from the deployment.
- 14:47And now we're like the dev team. So you got too good at this all
- 14:50right now you're. Yeah, you'd see the cycle.
- 14:53So there's a product some very close friends of mine built,
- 14:56they were over in the Middle East working with a customer
- 15:01there. And they built a product to be
- 15:04able to do this sort of drill down data analysis to solve a
- 15:08problem there and a very esoteric problem, signals
- 15:11intelligence problem. And then that product was sweet.
- 15:13And a lot of people, including my deployment, were like, that
- 15:17would be really useful. And so we kind of lobbied and
- 15:19got them to deploy it for us and it was great.
- 15:22And then it spread and spread and now it's like a super core
- 15:24part of the Foundry platform does.
- 15:26It have a name. Or yeah, it's called Contour,
- 15:29like every, like everything there.
- 15:30It went through multiple names. It's been named like 4, it was
- 15:33named 4 things. And there's, there's so many
- 15:35examples like that. And, and there's so many
- 15:37wonderful things about this, like contour is sick.
- 15:40And it was never on anyone's road map.
- 15:42It's just like 3 guys went and invented it because they needed
- 15:45it and it caught on and spread and you know, it was never on
- 15:48anyone. It was never on some central
- 15:50road map like in Q32016. We're going to go build this
- 15:53thing Now. The downside, the sort of
- 15:56invisible downside that's easy to ignore is there was also a
- 15:59ton of shit we built that went nowhere that was a complete
- 16:02waste of time. We had deployments where we were
- 16:04flying super expensive people, first class every week
- 16:08somewhere, you know, just just tons of time and opportunity
- 16:12cost and just pyres of money. Basically like customers were
- 16:15paying for it, right? Customers were paying for it,
- 16:17yeah. And they were kind of paying for
- 16:18the sort of service version of it in the short term.
- 16:20And there were customers. I actually met someone recently
- 16:23who was a customer of ours in that era.
- 16:25And she was, she kind of was like, yeah, we're.
- 16:27We felt pretty burned by you guys and I, I just think it's
- 16:31almost like this kind of necessary downside.
- 16:33I hate to be so glib about it, but like of having this high
- 16:37rate of innovation in the field, like, you know, people say fail
- 16:41fast, fail fast, move fast and break things like these, these
- 16:44ethos things in the valley. It's it's really representative
- 16:48of what we what anyone who's really done this well really
- 16:50knows, which is the reiteration and learning is everything.
- 16:54And one of the big problems in enterprise software is a lot of
- 16:57enterprise software is built by big lumbering teams who are
- 16:59working from 10:00 to 3:45 in air conditioned offices in the
- 17:05valley with ample ping pong breaks and what you know, blah,
- 17:09blah, blah, blah, blah, blah. Working off of these long multi
- 17:12quarter Rd. Maps and they're very sparsely
- 17:14talking to customers and you know you're what you're going to
- 17:17get from that is you're going to get consistency, you know you're
- 17:20going to get predictability, you're going to get all these
- 17:22things that that companies value as they get bigger.
- 17:25You're not going to get emerged. But you're not going to get
- 17:27that. Innovation.
- 17:28Sometimes here it like we didn't really figure it out.
- 17:30Other times it's like it's so valuable, we're just spread it
- 17:33to the rest of government. Yeah, these structures wind up
- 17:35being almost designed to squash like high variance innovation,
- 17:40like what all process does. The whole point of a process is
- 17:43to reduce variance. So you're you're an evangelist
- 17:47and defender for the forward deployed engineer, but your own
- 17:51company doesn't do that. Like what?
- 17:53What is your heuristic for what company should use a forward
- 17:57deployed engineer? And then why did you decide yes,
- 18:00we have a lot of FD not to former?
- 18:01FD ES at hex. So what you see is a lot of
- 18:03start-ups now adopting the FD model where they're like we're,
- 18:07we're, we have 4 deployed engineers.
- 18:09Well, I and I, when I, when I critique this, it's not like to
- 18:13gatekeep the title, whatever. Like it's really like, do you
- 18:17understand what worked? What, why this works for
- 18:20Palantir? Because I think a lot of people
- 18:23are cargo colting. They're saying, oh, Palantir did
- 18:26this and look at their revenue multiple and look at their
- 18:28success. If we do forward deployed
- 18:30engineering, we'll be as successful and they kind of go
- 18:33through the motions and then they're not getting the same
- 18:36results. And they're like, why?
- 18:38Well, you don't under really understand the physics of it.
- 18:41And from us when we were starting the company, you know,
- 18:44we'd obviously lived that, but we were like, we want actually
- 18:46want to kind of go the opposite way.
- 18:47We want to build 1 consistent piece of software that we can
- 18:50deliver in scale and. Super out of fashion right now.
- 18:54Super out of. Fashion right now, I know.
- 18:56But the raising rounds as an investor, but somehow, but not
- 18:59on some of the narratives. But but, but no.
- 19:01But it's, it's, there's a synthesis here, right?
- 19:04Like what we do is we just talk to customers.
- 19:08I'm in New York right now. I'm visiting customers.
- 19:10I'm bouncing all over town the next few days just visiting
- 19:12customers, getting up close. And I'm doing my best not to
- 19:15just meet with them in a conference room, but like, I
- 19:18want to go by someone's desk. I want to be like, great, what
- 19:20are you up to? What are you working on?
- 19:21Can you show me hex? Can you show me what you're
- 19:23working on? Like go get up close.
- 19:25I'm I'm bringing individual, I'm not bringing our sales, my sales
- 19:27people on those visits. I'm bringing the engineers and
- 19:29designers. I'm like, I want to go see this
- 19:31stuff And I try to get our whole team doing that and go get in
- 19:35front of customers, really understand what they're doing,
- 19:37really understand what they're asking for, really understand
- 19:39what they want and what they're using as an alternative.
- 19:41Like go do a great job at that and and talk to customers.
- 19:45And if we actually have a problem at Hex, it's like, I
- 19:47think some days it's like, I worry our EPD team is our
- 19:49engineer product and design teams are talking to customers
- 19:52too much. Like it's like, guys, at some
- 19:54point we got to go write the code.
- 19:54You know, it's like we talk, we talk to customers a lot.
- 19:58And some of it's just you're a smaller company and so you can't
- 20:00just have everybody. Yeah, it's a.
- 20:02Cultural, but it's a cultural thing.
- 20:04It's like, I, I think it's really important to have people
- 20:08really build empathy for customers.
- 20:10I mean I I don't I think the worst engineering in the world
- 20:13is done via ticket. Like if you wanted to have some
- 20:15tickets to like manage your work.
- 20:17Yeah, I mean it's a core Y combinator.
- 20:19You know, principle of building companies is like what?
- 20:21Build something people want know your customer.
- 20:23You know that sort of sent. But a lot of most people don't
- 20:26do that. They don't do that.
- 20:27They said they have the road map but.
- 20:28What what start-ups do you think or smaller companies, non
- 20:31palantir companies are doing for deployed engineers in a smart
- 20:34way? The only, I mean, the only one I
- 20:35know who's doing at the Palantir version, which is basically, we
- 20:39don't really know what to build. We're just going to go in bed
- 20:42with customers, get them to pay us some services thing to cover
- 20:45the time maybe and and have some exchange of value.
- 20:48And we'll just build the platform depending on what we
- 20:50learn. Is my friend Arjun at Distill
- 20:53Distill with AY he he's. I think there may be other
- 20:59companies now doing some version of this but he's the one who I
- 21:01know is doing it the most pure. Is he super charismatic?
- 21:05I think he's very charismatic. And not like what's just funny,
- 21:07you refer to Alex Carp earlier as Doctor Carp.
- 21:10He's he's like a philosophy PhD. Yeah, right.
- 21:12Like, like not there uniquely has to sort of like, you know,
- 21:16we we're like Wizard of Oz, like we're gonna do a magic thing.
- 21:19I'm like, can anyone else really sell that besides that company,
- 21:24you know, in the sort of he's just.
- 21:25Called him Doctor Carp. He's just he's doctor.
- 21:27It's like when you it's like when you grow up in your
- 21:28neighbors like Mr. Mr. Smith, you know to you and like later
- 21:33in life you run into them again. You still call him Mr. Smith,
- 21:35this is. Funny you guys how you're like
- 21:37I'm. Jim, I'm Jim now, son.
- 21:38It's all right. He talks about it like, you
- 21:40know, it's a culture of, you know, people will tell him like
- 21:42he's totally wrong or whatever, but that's just a very
- 21:44deferential way to refer to him. I don't know.
- 21:46Yeah, but it was, it is a very decent, you know, they're very
- 21:49open, very transparent, very best idea wins.
- 21:53Not a lot of job titleature. I mean, I had the same job title
- 21:56all five years I was there. I went, I had people reporting
- 21:59to me. I was doing IC stuff, I was
- 22:00doing big lead stuff. It didn't matter.
- 22:03Like in when you're in that environment, you kind of forget
- 22:06about job titles for a while. You forget about there's
- 22:08different, you squeeze the balloon, the problems emerge
- 22:10elsewhere. Like there's going to be shadow
- 22:11hierarchies and stuff. Every cultural thing has
- 22:14trade-offs, but I I thought the culture there really was
- 22:17designed. It's, it's almost like it, it,
- 22:19it, it was very sharply designed around getting a very specific
- 22:24thing, which is maximum pace of iteration and learning.
- 22:30And that meant there's all these negative trade-offs.
- 22:34But culturally I think we always like acknowledge them and
- 22:36accepted them. And there was a time, I remember
- 22:39we had a period when I was there where a lot of people have left
- 22:41because they felt like there wasn't a growth path for them
- 22:44and because we didn't have like a job title ladder to climb like
- 22:47a lot of people are used. To Was it hard to raise money
- 22:49without a clear title where people knew where it's?
- 22:52Hard for me to raise money. Or just I would imagine going
- 22:55off and starting a company you want to be able to say I was
- 22:57important here or somewhere, which is sort of a vague.
- 22:59It was actually interesting because I did another thing in
- 23:02between Palantir and Hex, I went to a relational biotech company
- 23:05here. Yeah, I was just called Trial
- 23:06Spark at the time, not confirmation.
- 23:07I remember interviewing and I was like, what do I, what do I
- 23:10say? I do like I'm a deployment
- 23:12strategist. What's that mean?
- 23:13Like, and I think people who are hiring from Palantir kind of
- 23:17knew they were like, OK, we kind of knew that that was a thing
- 23:22when I was raising money. It wasn't a big I don't that
- 23:24didn't really come up. It was just the.
- 23:25Product just sort of the. It was just that we clearly had
- 23:28a sense of what we were doing and what we wanted.
- 23:29To do but say what Hex does just we don't even think about the
- 23:32company, but just so people know where.
- 23:33You're coming. So we're building AI data
- 23:35analytic software like sometimes people call us cursor for data
- 23:39or things like that. It's basically you can use
- 23:43natural language to go ask any. Other questions just list a
- 23:46bunch of highly valued companies like Cursor for data, Anthropic
- 23:48for. We were opening up for one data.
- 23:50We were called. Bigma, Bigma for data, it's
- 23:52like, yeah, because the first, the first arc of the company
- 23:55really was like, there's all these local fragmented data
- 23:57analysis things. We've spent all this time
- 23:59bringing your data together and you're just putting dashboards
- 24:01on top. You know, we wanted to build
- 24:03this cloud based front end for all of your data work and it was
- 24:07very similar to what Figma had done when we started the
- 24:09company. Figma was, you know, on its way
- 24:12north and was really taking off and sort of an inspiration of
- 24:15like, oh, you can take these local fragmented workflows,
- 24:17bring them to the cloud, bring make them collaborative and in
- 24:21the process of doing that, bring more people in as participants.
- 24:24Those are really the initial thesis of hex and then a couple
- 24:28years end of the journey, like GPT 3 came out and then ChatGPT
- 24:31came out and this was a thing we really got entranced by.
- 24:34So the the second-half of the company or more than half now
- 24:38the company has been figuring out how to really apply AI and
- 24:42now agents to work for a data analysis, which is the Holy
- 24:45Grail. Like the Holy Grail.
- 24:46The reason people buy data tools, ostensibly invest
- 24:48billions and billions of dollars over your data infrastructure is
- 24:51because you want to know things and you want to make better
- 24:53decisions. And we make that really easy and
- 24:55we do a very different set of things.
- 24:57Well, then what Palantir did, we don't do like data integration
- 25:00and infrastructure stuff. We're very much like the app and
- 25:02collaboration layer where. Where do you say relative to
- 25:04business intelligence? Yeah, we are.
- 25:06We think of ourselves as reinventing that like we are.
- 25:09What? People don't like that space or
- 25:10what's the like? You don't use that term.
- 25:12I didn't, yeah. I didn't want to use that term
- 25:14because it had so much baggage. Like when people think for a
- 25:17long time, when people think BI, they're like, OK, well, I think
- 25:20of all the features that Tableau has built for 20 years and I
- 25:24think about dashboards and I want you to do that.
- 25:26And we are actually originally were very insistent on like we
- 25:30actually think that the old way of thinking about BI is like
- 25:32super low value. We think dashboards is going to
- 25:35raise more questions than answers is something I've said a
- 25:38lot. And like the most of the
- 25:40valuable data work is not happening happening in these
- 25:42dashboards. And instead it's people are
- 25:44trying to do it in this melange of other tools.
- 25:47Jupiter notebooks and SQL queries and spreadsheets and
- 25:50screenshots of a chart pasted into APDF of a deck attached to
- 25:56an e-mail you're. More rooted in like the role
- 25:57like you're hiring these smart data analysts and we want to.
- 26:01Originally start, our initial sort of beachhead landing thing
- 26:05was like we're going to come in and we're going to revolutionize
- 26:07your way. Your data team works right and
- 26:09very similar to Figma coming in and we want to revolutionize.
- 26:11Now you're like you don't need to hire them or how do you?
- 26:14Every AI embracing company has its trade off between supporting
- 26:19the role versus saying, oh, we're helping.
- 26:21I think we're. Helping to reinvent and evolve
- 26:22it. And we have a great data team.
- 26:24Hex. It's almost like over
- 26:25provisioned in the sense of we just have really wonderfully and
- 26:30talented people on it, but also hired people that have been
- 26:33thought leaders and writers in our space for a while who are
- 26:37amazing collaborators for us to build the product.
- 26:39But also collaborators just thinking about like what is what
- 26:42are these roles becoming? And it's interesting to see our
- 26:45own employees kind of disrupting their own job a little bit.
- 26:47And it's probably similar to what it feels like to be an
- 26:50engineer or cursor or something like you're trying to reinvent
- 26:52the way your job works. And so you know, where we see
- 26:55ourselves as fellow travellers with that data audience of like
- 26:58how we're helping reinvent the way their job works and what
- 27:01that looks like and what it looks like to fulfill the
- 27:03mission they've always wanted for their organizations, which
- 27:05is you want data to be useful and help solve problems.
- 27:08And now with with AI, you know, the the fastest growing part of
- 27:11our user base, of course, is all these other people in these
- 27:14companies that are now able to use data in a way they weren't
- 27:18able to before. And in that way, we are
- 27:20fulfilling what BI want, always wanted and purported to be and
- 27:24what people thought they were buying with their billions of
- 27:26dollars of BI spend, which was we want to make everyone
- 27:29data-driven. Well, you got a lot of
- 27:31dashboards. Are you feeling data-driven?
- 27:32Are you able, you know, is everyone able to ask their
- 27:35questions? Because if you can't interrogate
- 27:36the dashboard, you don't know it's limited.
- 27:37They're really good. Like dashboards are fine people.
- 27:39There's this big discourse now. Dash dashboards are dead.
- 27:42Everything has to be dead or broken or, you know, whatever.
- 27:46Dashboards are dead. I'm like, dashboards are fine.
- 27:49They're fine, but they've always been fine.
- 27:52Like the? Extent they're dead, I see
- 27:53people want to ask a question directly of data in dashboards,
- 27:57which is. The same as when we started the
- 27:58company. It's like dashboards are great
- 28:00for seeing what happened, this KPI, blah, blah, blah, whatever.
- 28:03It's a collection of these charts.
- 28:05Great, fine. But often what you look at is
- 28:09you look at that chart and you go, OK, well I want to ask a
- 28:11next question. Why, why did that happen?
- 28:13I want to drill down on this and there are other times something
- 28:15like that has happened. That's what we built Hex to
- 28:18really excel at. And now it like the architecture
- 28:22and a lot of the things we built, we kind of feel like we
- 28:23had right place, right time because the way we even
- 28:27structured the product is really conducive to these agent work
- 28:31flows, which is which are about depth.
- 28:34A lot of times, like the magic of these things is you can ask a
- 28:36first question, get an answer, and then you can ask follow-ups
- 28:38and it can go now and drill in and have multiple different tool
- 28:41calls and paths is exploring. And all these things we built,
- 28:44even going back to the very beginning that we didn't know
- 28:47were going to be useful because we have magic robots that can
- 28:50talk to us now. It's all very useful now.
- 28:53Now useful for way more people, which is really really cool.
- 28:57Are you using cloud or what is the intelligence sort of?
- 29:00Behind this, we use cloud and the GPT models and it's very
- 29:03interesting. We have good relationships with
- 29:05both companies. We spend time with their
- 29:08research teams and who's. Doing better right now.
- 29:12They're both doing great. I think the GPT 5.4 models which
- 29:15are the latest ones to come out are really fantastic.
- 29:17And we have a quote, my Co founder Caitlin, who I think
- 29:20you've met has a quote in their launch blog posts summering up
- 29:23our findings. But they really, really good and
- 29:26getting better. I think what we're seeing is
- 29:28these models, the main focus of the labs right now appear to be
- 29:32kind of hill climbing the models on coding tasks, which is you
- 29:37can kind of squint your eyes and look at hacks and be like, it's
- 29:39like coding. It's like you guys write SQL and
- 29:41Python And Dataviz. And the improvements on those
- 29:45models of coding tasks does make our performance for our domain
- 29:49better incrementally. But but surprisingly, actually
- 29:52we even saw with, I won't say which one, but one of the latest
- 29:55new model releases from one of the labs, it was actually a
- 29:57regression for the evals we have.
- 30:00We have a really extensive set of really rigorous data
- 30:03analytics evals that are not like the naive ones, like how
- 30:06many widgets did we saw last quarter, but like really brutal,
- 30:09like the real world shit that people ask of data.
- 30:13And it's interesting to see the models sometimes even regress
- 30:17cuz clearly they've been trying to be like hill climbed on
- 30:18something else. And one of the big problems is
- 30:20in data analysis, it's non verifiable and there's actually
- 30:23different ways to answer a question and it's it's.
- 30:26Right coding, has this sort of code worked it built this thing
- 30:29you can sort of right whereas the sort of analysis.
- 30:34Yeah, it's a little squishier. And so this is something we
- 30:36work, we try to This is why we spend time at the labs is like
- 30:39trying to like help figure out like how, how can we make the
- 30:42bottles better for this? Should we be RL in our own
- 30:44models? I know there's a lot of like
- 30:45even conflicting opinions internally.
- 30:47Are you spending money on your own models?
- 30:49We do we run some of our own inference for some specialized
- 30:52tasks. The main we are using the like
- 30:54GPT and clawed sort of state-of-the-art models for like
- 30:57our main agent backbone. But we, we see some interesting
- 31:00opportunities and it's not impossible that the lab, those
- 31:03anthropic and open AI even allow they, they used to sort of in
- 31:08the older era had a lot more fine tuning tools.
- 31:10And I think they may come back in some places too, where they
- 31:13sort of allow you to do these RL regimes on top of their base.
- 31:16And where are you on sort of agent mania, you know, are you
- 31:19building? Are are you going to be like
- 31:21sell, you know, are agents as representative as humans in your
- 31:24product? You're going to be selling seats
- 31:26to agents or you think that's sort of out of control?
- 31:28Or I, I think that's, I think it's probably overshot a little
- 31:31bit. I mean, we've talked I, you
- 31:34know, with the pricing thing is interesting.
- 31:37You're selling seats. We sell seats.
- 31:40Another not cruel thing. We sell seats usage.
- 31:43Pricing right now, what's going on?
- 31:45We are, we're adding, you know, usage on top because one thing
- 31:47that's crazy is our weekly actives chart and our messages
- 31:50per week chart is like straight vertical now.
- 31:53It's like gone fully parabolic. Like the product is really
- 31:55working. The agents are really good.
- 31:57People are loving them and using them.
- 31:59Yeah. They're costing you more.
- 32:00Fortunate part is our cost chart looks the same and if you're if
- 32:04you're a if you're a company today that has working agents,
- 32:06these things are expensive to run and you need to charge for
- 32:08them. So we're we're going to do it
- 32:09with our seats have like a base amount of tokens and then or
- 32:13credits and then you can get above.
- 32:15It's pretty standard now for a lot of tools.
- 32:17It's like lovable replay, etcetera, but.
- 32:19Why seats at all? It's a good question.
- 32:22I think in some ways it allows a base of predictability for
- 32:25customers. We were, we're a product that
- 32:26wants to get really wide in your organization and if we can say,
- 32:29Yep, it's, you know, however many $50 or whatever for a seat
- 32:32for a month, you, you know, and that comes with this number of
- 32:34credits and the average user is going to their usage is going to
- 32:37be covered by that. That's really nice for
- 32:39predictability. I mean, that is a big thing.
- 32:40People who have never sold or bought at scale enterprise
- 32:44software know like having it be fully variable is like that is
- 32:48tough As for a lot of companies. And, and generally, like I, I do
- 32:52have these funny conversations sometimes about seats where even
- 32:55with smart V ZS, which was funny, that are like, well, you
- 32:58know, there's going to be pressure on the number of seats
- 33:01you're going to be able to sell because customers are hiring
- 33:03less people or whatever. And I'm like, we've grown a lot
- 33:08and we're doing really well, but we're still, the amount of
- 33:10revenue we make today is still very, very small to the
- 33:13incumbents and where we want to go and all the time we think we
- 33:16can capture. And I'm like, if there is not
- 33:2010s of millions, hundreds of millions of dollars worth of
- 33:22seats for us to sell because everyone's been laid off.
- 33:25The. World's like, like I need, I
- 33:27need to own a lot of guns. The least of my problems is my
- 33:31fucking SAS business. Like I'm there's gonna be riots
- 33:34in the street. And so I think it's kind of
- 33:37funny this conversation about this like, yes, on the margin,
- 33:40is it possible that some of our customers will just be like
- 33:42small or in absolute terms, the number of people?
- 33:45Yes. Are there still a lot of seats
- 33:47to sell for our products like ours?
- 33:49Yes. Do a lot of customers still want
- 33:51to buy seats cuz it makes sense for what they're doing?
- 33:54Yes. Can I layer a consumption model
- 33:56on top that helps me both cover costs and also capture upside
- 33:59from real power users who are getting excess value?
- 34:02Yes. And over time we may tweak those
- 34:04dials. I don't know, a year from now,
- 34:05maybe we turn the seats down to 0.
- 34:07Maybe we maybe it goes back. Elena Verna at Lovable has this
- 34:10blog post that we talked about internally where she's like, I
- 34:13think the token cost is gonna come down significantly and
- 34:16we're all gonna be back to selling seats, Who knows?
- 34:19So we like having the options and we wanna, ultimately our job
- 34:23is to actually meet our customers where they're at.
- 34:25And people always joke. Customers don't wanna pay
- 34:27anything. Well, I actually disagree.
- 34:29I think customers understand the value and want to pay for
- 34:32something that's going to provide good value for them.
- 34:35They want to support you. They want to be supported.
- 34:37I'm a buyer of a lot of software at Hex now and I'm, I'm very
- 34:40happy to pay for stuff that works well.
- 34:41And if I feel like there's price to value and if I see the bill
- 34:43for what we're paying for something and I'm like, yeah,
- 34:45but we love it. I'm like great, cool.
- 34:47Where are you on this? Yeah, White collar works about
- 34:50to get decimated and our country's going to be, I don't
- 34:52know, going for the guns. Like, I, I literally was talking
- 34:55to a, you know, a very respected person, you know, who was like,
- 35:01oh man, we're going to enter like an insane political moment.
- 35:03Like every, you know, democratic cities.
- 35:06You know, if you know, all the Goldman analysts or something
- 35:08are about to get fired, you know, then you're going to have
- 35:11this sort of weird class, you know, just like you're going to
- 35:14have pockets of people who are very resentful about the new
- 35:16economy. Like I don't know how I.
- 35:18Already see it? I mean downward mobility,
- 35:20downwardly mobile elites. That is right.
- 35:23The most dangerous force in America is the leading.
- 35:25Indicator well, I mean, forget about America, just study
- 35:27history, right? Like that is a bad thing like
- 35:31that, that is that will lead to problems and I, I don't know for
- 35:35sure what the future looks like. I would say I applaud Dr. Karp.
- 35:39I'll still say Doctor Karp, who I think has been a very bold
- 35:41voice in saying like we should all be thinking about how we're
- 35:44using AI to like increase employment and productivity.
- 35:47And I, I agree with that. Now I'm building a tool that you
- 35:52can kind of look at like, do I need as many data scientists or
- 35:55data analysis I have? Sure.
- 35:58I also see a lot of the data scientists and data analysis we
- 36:00have using the shit out of it, doing a ton, being way more
- 36:03productive, being more valuable for their companies and often
- 36:06now starting to work at these higher levels of strategic
- 36:08insight and maybe their roles are starting to evolve.
- 36:10So it's not at all clear to me. We build an AI product.
- 36:14We use a lot of AI tools internally.
- 36:15I'm trying to hire as fast as I can.
- 36:17So we'll see. The other thing I think about is
- 36:20like where is the productivity? I mean, I, I can look within our
- 36:22own customers and see them doing more and all of this.
- 36:24I think a really interesting question is we're a year and a
- 36:28half now into coding agents being mainstream and we're
- 36:33billions of dollars of revenue into tokens paid to Claude and
- 36:38Codex and Cursor and other things that start with C and
- 36:44where where's the where's the explosion in software?
- 36:47We're getting a lot of vibe coded stuff.
- 36:49We're getting a lot of sort of slop wear.
- 36:50But if you look at the the the software companies you buy from,
- 36:54it's like, where's the where's the like cornucopia?
- 36:56Like why am I not getting new features every day?
- 36:58Right. It's so we.
- 37:00We've built a maybe. The bottleneck looks elsewhere.
- 37:02I don't know. My my my business lead has built
- 37:04us, you know, internal like ticketing tool that he vibe
- 37:07coded. We're we're about to see, you
- 37:09know, when the metal meets the pavement or whatever the whether
- 37:13you know, we can accept payments and everything safely through
- 37:16it, but. Like why like you that?
- 37:19But I get. That's right, you can.
- 37:20Doesn't need to be perfect, is. That Riley, yeah, yeah, that
- 37:23Riley can do that is cool. It's huge.
- 37:25I mean, it's pretty. It's pretty, but.
- 37:27But like, should he be doing that?
- 37:29Yeah, there's a lot of things to do, but the I think it'll be
- 37:32good, you know, we'll keep track of everyone.
- 37:35But to your point, I mean, do you feel pressure as the at at a
- 37:38start up, your investor was want to hear this story of like,
- 37:41yeah, we're moving so much faster because of, you know,
- 37:45cursor or whatever. And and you're saying no, but
- 37:48it's not like I do. Feel like we are.
- 37:49It is also interesting. We are living, you know, I've
- 37:52got, I've just came from our office.
- 37:54I've walked by any engineer's desk and they've got cursor or
- 37:57cloud code or codecs. Now they've gotten very popular
- 38:01up on their screen. They're using it.
- 38:02I think we're still figuring out how to use these things, like
- 38:05we're setting up cloud dev environments to make it easier
- 38:07to have multiple different agents that can work in their
- 38:09own dev boxes. And it's like, OK, so now our
- 38:11engineers are just bouncing between 5 different agents
- 38:13working on different tasks at a time.
- 38:15Who's reviewing all this code? Who's initiating the thought on
- 38:19what we should build? I mean, I'm almost like really
- 38:21excited about the idea of just getting this torrent of PRS from
- 38:24all of our engineers. I'm also slightly worried
- 38:26because we're a product that's prided itself on sign and
- 38:29consistency and fit and finish and feel.
- 38:31And our designers are busy now like trying to do Polish commits
- 38:35on to all these PRS. And it's how do we even get all
- 38:40this out into the product in a way that our customers
- 38:42understand? It's gonna be changing fast and
- 38:45that's exciting, but also stressful.
- 38:46It's like, I think we're all still figuring out how to
- 38:48metabolize all of this something.
- 38:49Tells me there's just like a pace of that humans can absorb
- 38:53the products. Change like I, I think it's
- 38:56literally like if you eat too much food, like your body can
- 38:58only metabolize so much of it. And like, I think we're still
- 39:05just figuring this out. And so I, I don't know that
- 39:08these things are going to happen overnight.
- 39:10And like, if your model of a Goldman analyst is that they're
- 39:12just doing this really rote, repeatable work that an agent
- 39:16can replace. Sure.
- 39:18Is that all that a Goldman analyst does?
- 39:20It's the same, I think about for data people, if you're like,
- 39:24what, how do we understand those jobs?
- 39:26What are actually the bottlenecks and companies doing
- 39:28certain things we're supposed to early in figuring this out?
- 39:33Particularly, it's interesting to me, I've observed almost like
- 39:36coding agents have disrupted product management faster than
- 39:38engineering because the engineers are almost becoming
- 39:41PMS in a funny way where it's like, it's kind of unintuitive,
- 39:44but like the, the a lot of our engineers, you know, are now
- 39:49their, their, their job is less like, OK, I take the, the, the
- 39:54product spec that was written in English and I translate it to
- 39:57code syntax. Like that used to be the thing
- 39:59it's like. Your engineering candidate.
- 40:01Oh, you know how to take English and turn it into TypeScript.
- 40:04Great. Welcome.
- 40:05Right. That was kind of like at least
- 40:06the first order heuristic of what we're like interviewing you
- 40:08for with like coding interviews. Well, now there's magic robots
- 40:11that can do that. And so and so instead, so we're
- 40:14gonna keep our smartest people, the engineers, and shift them to
- 40:18sort of the strategy rather than.
- 40:20Yes, it's not like, it's not like.
- 40:22A product manager, I would never be an engineer, but I mean, that
- 40:25is sort of, it's like, OK, if they're not writing the code,
- 40:27they're still very smart, understand sort of the core
- 40:30product, and so they're gonna become the product.
- 40:32Going back, by the way, the type of engineers we try to hire are
- 40:35those that like talking to customers and like thinking
- 40:37about what to build and enjoy our product space.
- 40:40It's a cool product space. So yeah, those people have
- 40:43always liked that, and now they just have kind of more time and
- 40:45brain space for that, and they can be thinking about what to
- 40:47build and how should this work and the architectures and all
- 40:49this, and it's really cool to see.
- 40:51So we'll see how these roles change what designers do.
- 40:56Our designers have coded for a long time, but now they're
- 40:58coding a ton. They're mostly coding.
- 41:01It's like, well, are you, what, what do you do?
- 41:04Are you just an engineer that has like really good taste and
- 41:07more tattoos? Like what's that?
- 41:09It's very interesting to see happen.
- 41:11And internally I'm just trying to lead everyone through it
- 41:15saying we're all learning this together and if you don't
- 41:19disrupt your job, someone else. Well, and don't worry.
- 41:22How many people do we need to get to a huge exit hex?
- 41:25I don't know. More than we have today, so, and
- 41:28more than our. Competitors so we wanna get good
- 41:30ones you guys are all safe let's all.
- 41:32Figure out how we can all use our time really well, and then
- 41:34we'll go build this thing and have it be really valuable.
- 41:36So I wanted to go back to Palantir and some of the, I
- 41:40mean, obviously we just had this Anthropic took the headline
- 41:44because they were the ones sort of fighting with the government
- 41:46over, yes, fighting with the Department of War or Defence,
- 41:49depending where you sit on. Yeah, sit on things.
- 41:52Not going to take a position. I think we I've gone back to
- 41:55defence. I mean defence was established
- 41:57by Congress. So it is sort of a branding
- 41:59exercise. You don't need to weigh in on
- 42:00this, but I think core to the conflict the, you know,
- 42:06Anthropic had two concerns, right?
- 42:08One was this autonomous targeting and the other was this
- 42:11piece that we are gonna do domestic surveillance.
- 42:14And I think embedded in the domestic surveillance is that
- 42:17fundamentally because of how much better their technology and
- 42:21technology generally has gotten, the government can just acquire
- 42:24publicly available data and then come up with this NSA dragnet.
- 42:28That wouldn't have been possible, you know, when
- 42:30Congress was writing laws about domestic surveillance.
- 42:34And obviously pound tier like fits fits into this.
- 42:36So what is your view of? Like what's possible in terms
- 42:39of? Exhibit A in a tech company that
- 42:42lives in that Gray area. I was there 10 years ago after
- 42:47the 2016 election, and it had been not so long before that
- 42:51that we had like an, it was like an all hands or a, some sync
- 42:54version we had of this. That was like doing a like a
- 42:58win. Like the deployment team lead
- 43:00came up on stage to like talk about our work with ICE, which
- 43:04we're really proud of because we were working with this team
- 43:06called HSI. It's like the really good guy,
- 43:08you know, the, the, whatever else you feel about other parts
- 43:10of the government, he's stopping human trafficking and like drug
- 43:13smuggling, Like go, go HSI. Yeah.
- 43:18And then like, you know, that election, the vibe shifted on
- 43:21all of tech shifted 10 years ago is kind of a long time.
- 43:23Remember back. But like and all of a sudden
- 43:25this became really tough. And we had internal concerns.
- 43:28We had protests outside the office.
- 43:31We had applicants not wanting to come people not people not
- 43:34applying or pulling out of interview processes because we
- 43:36were evil. And there was a big clamor
- 43:38internally to get Doctor Carr, Alex Carp to, to say something
- 43:42publicly say something engage on this.
- 43:45And I, I was, I was among the frustrated on that.
- 43:48I'll, I'll confess, like, you know, I'm interviewing
- 43:51candidates who are like half the interview.
- 43:53They're just asking me questions about how I like, I can sleep at
- 43:55night or something. I'm like.
- 43:57It's also so secretive that it's hard to even necessarily not
- 44:00even inside the company. And inside we knew and and I
- 44:03think we had an appreciation and I still have a deep appreciation
- 44:06for like this was a company that recognized that to serve the
- 44:10government, to serve the government in these spaces, you
- 44:15have to be comfortable in the Gray areas.
- 44:17And that doesn't mean moral compromise.
- 44:18It actually means moral clarity and having principles, even when
- 44:22they're hard. And, you know, in retrospect, he
- 44:26was totally right. And they're, they're much more
- 44:27public engaging in the forum now.
- 44:29But I think at the time, the right thing for us was to just
- 44:31stay quiet, not engage, not not inflamed us and say let's focus
- 44:35on our work and let's focus on doing the best we can and also
- 44:41embrace the fact that we live in a democracy.
- 44:44But you have a view on this anthropic blow up specifically?
- 44:47Well, yeah, that's my yeah. The kind of segue it's.
- 44:50I love Anthropic. We're partners with them.
- 44:52They're customers of ours. I don't, I say this with love,
- 44:56but I do think it's like the heart and the head.
- 44:59Like from a heart. You understand some of these
- 45:02concerns. Like, would I, like, would most
- 45:05Americans vote for doing mass surveillance?
- 45:09You know, you can go gather up all this information and figure
- 45:11out where everyone's loyalties lie or something like, no, I
- 45:13think most Americans would not want that.
- 45:17But that's why we have a democracy and you can vote.
- 45:20People can vote for that. And I think what we came to
- 45:22appreciate a Palantir, what sort of respect is like, well, what,
- 45:25OK, we're gonna have red lines and what we're willing to do,
- 45:27does the government have to call us every time they want to use
- 45:29Gotham was our product every time there was any palantir to
- 45:32use this thing like. Isn't this an extreme deference
- 45:35to the executive branch specifically?
- 45:37I mean, it's not like Congress has passed the law, said, oh, we
- 45:40love Palantir. Like you.
- 45:42You could have a belief that like our product is so
- 45:45significant. They have, I mean they they sort
- 45:47of every year. Did power, but they haven't.
- 45:50Like expressly said, this is great I.
- 45:52Mean I'm, I'm not like a constitutional scholar.
- 45:55I'm just saying. But I do think, and I don't want
- 45:57to, you know, you don't, you're not, you're just the one here
- 45:59talking about it now. But like they're, they're, I do
- 46:02think there's this argument that it's like, oh, we're deferring
- 46:05to democracy. It's like you're deferring to
- 46:06like 1 branch that's taking on a lot of power and saying we
- 46:11represent democracy. This is what it looks like to
- 46:13live in a Republic. It's messy.
- 46:15You, you, you. Elect right.
- 46:16So it requires on the part of other actors to sort of
- 46:19interpret and what the will of democracy is beyond just like
- 46:23difference to the executive. You described a Republican, but.
- 46:26The argument that Palantir is making and trying to suggest
- 46:29anthropic embrace is sort of like yield to just the
- 46:32executive's interpretation of what the will of.
- 46:35Democracy is yes, yes. And that will of that executive
- 46:42is bound that that's Article 2 is bound by Article 1, the
- 46:45legislative branch and the laws they pass.
- 46:47And look, I, I, I'm not going to get into like debates on like
- 46:50current Supreme Court issues or all that, but like, at least if
- 46:54you read the Constitution, it's pretty damn clear, like the, the
- 46:57legislator pass, the legislature passes laws, including the power
- 47:01of the purse. They control what money is spent
- 47:05and the executive executes on that.
- 47:06And there's all these, you know, unitary executive and all these
- 47:08things. I'm not going to get into all
- 47:09that, but I will just tell you, even you see this play out now,
- 47:12there is a funding bill issue in Congress right now for
- 47:18Department of Homeland Security where there appears to be a
- 47:22legislative majority or controlling legislative majority
- 47:25that is not comfortable with funding Homeland Security at the
- 47:29way that they have been. And that is the democratic
- 47:32process working out. And you can look at the last
- 47:34this current Congress and say, well, they've been overly
- 47:35deferential to the executive and whatever and that may well be.
- 47:39But you see this happening. And now people, we're looking
- 47:41forward to the midterms and people are worried about that
- 47:42and they're going to react to the public sentiment.
- 47:45And if public doesn't want something great, what a
- 47:48wonderful opportunity for a political entrepreneur in
- 47:51Congress to get up and propose a bill banning this type of mass
- 47:54surveillance. They're just outlawed.
- 47:56You can pass a law to do that. That is a thing.
- 47:59That's your job. Actually, that's literally their
- 48:01job. If this is a thing we don't want
- 48:02to have, we both want. Congress to take a more active
- 48:05role in saying what they support and what they don't.
- 48:07And I agree that I mean, ISIS and Homeland Security is a good
- 48:11example where Congress is stalling on it.
- 48:14I mean, we also live in this liberal democracy where unless
- 48:18it is a I understand in a war, people get drafted and summoned
- 48:21into government service. But the, the in normal times, I
- 48:25think the belief in America is that you don't have to do
- 48:29something for the government if you don't want, you know, you're
- 48:31not like. Off the pay taxes, right?
- 48:33Right. Besides sort of making the whole
- 48:35thing work with taxes, you don't have to like go work for the
- 48:37government. You don't have to, and Anthropic
- 48:40doesn't have to. No, no one does but the.
- 48:42Criticism of anthropic is they sort of need to be summoned into
- 48:44it. Maybe they could just an option
- 48:50they have is say we actually don't want to work with you
- 48:52guys, we're going to withdraw from the contract.
- 48:54Best of luck using other AI providers like that that's
- 48:57allowed. They're not.
- 48:59There's not the. Trump There's a war.
- 49:00Powers Act, There's a War Powers Act where the government can
- 49:03mandate like I think it's called war powers that I think it's
- 49:06just like we saw some of this during COVID.
- 49:07Like that's one of the ones they floated they but they ended up
- 49:10saying, oh man, it's a supply chain risk.
- 49:13It's more dangerous, apparently. But they're but that's them
- 49:16that's almost doing the opposite.
- 49:17They're almost cutting them out. I mean, I'm just saying, I'm
- 49:19just saying that like an option they have is just to not do it.
- 49:21Now going back at Palantir, one of the things we talked about
- 49:25that I do believe to be true is if we're not engaging on this
- 49:27issue, other companies with less scruples, with less principle
- 49:34point of view and by the way less effective software are
- 49:37going to. And we have a privacy and civil
- 49:40liberties team account here that I loved.
- 49:42If there's a lot of them are still there, they're incredible
- 49:44people, very thoughtful that we worked with very closely on a
- 49:47lot of these things that and, you know, you kind of look down
- 49:50the street at the other defense contractors and you go, they
- 49:53don't have that shit or they do. There are a lot of lawyers who
- 49:56but it's, it's, it's like, would we rather be the ones engaging
- 50:00on this and thinking critically about it and trying to find the
- 50:02synthesis of how you have robust defense with civil liberties, or
- 50:06would you rather see the field? And for us, the calculus was we
- 50:10would rather be here, even if it's a little uncomfortable,
- 50:13because that's our role in this pluralistic, messy democracy we
- 50:17have. I mean, I agree.
- 50:18That there was sort of a misguided time where the tech
- 50:20industry said we shouldn't do, you know, Google didn't want to
- 50:23work with the government or, you know, there was like, we're not
- 50:25work with the Department of Defense.
- 50:26We've moved to a point where it's patriotic to support the
- 50:30government. Yeah, but I but I think there's.
- 50:32Nothing like nothing like multiple wars breaking out to
- 50:35kind of get Americans to like wake up and be like, oh, maybe
- 50:37we need maybe we need good capabilities and weapons.
- 50:40Like Anderol was like super unsexy off SPAC, not a sexy VC
- 50:45thing. And then all of a sudden a war
- 50:46breaks out in Ukraine that's becoming like a drone war.
- 50:48And everyone's like Anderol's amazing.
- 50:50But I've always thought Anderol's amazing.
- 50:52And I know all the founders and, you know, but it was funny to
- 50:55kind of see them going from these like black sheep to now
- 50:57these darlings on the other side of a war that kind of reminds
- 51:00you it's not the end of history. Like the Ukraine war is a
- 51:03reminder that like, no, this is that we live in a dangerous
- 51:06world. But the argument that you're
- 51:08making that Palantir, you know, we're so it's so thoughtful
- 51:11about, you know, sort of civil rights or whatever, or like
- 51:16understanding the effect of its technology.
- 51:18The argument right now is that tech companies shouldn't be the
- 51:21ones deciding, and the government should have total,
- 51:24total control. Like that's the line from what
- 51:29heck, said. The Palantir seems to to sort of
- 51:31say, whoa, we're not creating contracts that limit you.
- 51:35But they can just choose not to engage in a contract.
- 51:38There were, there were, there was work.
- 51:40I can't talk about it. There were things at Palantir
- 51:41that we just decided not to do, countries we decided not to work
- 51:44with lucrative, lucrative deals where we sent teams out to go
- 51:50diligent something and came back and said this is not an area we
- 51:54are not comfortable here. We don't think we can actually
- 51:57have our software to be deployed in a way that we think is gonna
- 51:59preserve the things we want. We're just not gonna engage.
- 52:03I mean, that's an option. But Andrew Abbott had a contract
- 52:05they liked and then the Defense Department.
- 52:08Yeah, I mean, I've read their stuff.
- 52:09I've read Emil Michael's interview with Empire Wires,
- 52:12Empire Wires, and I don't wanna. The last thing I want to do is
- 52:18get in any of that. I'll just say this.
- 52:20Do you want to be spicy? It's just of the moment.
- 52:23I'm not trying to get. Are you neutral on it or are you
- 52:26you? What is your you know you're
- 52:28supportive of? The department I, I, I was
- 52:29raised at pound tier and I, I part of the reason I was there
- 52:32and part of the reason I still am supportive of is I think our,
- 52:34we, we, the West needs the best capabilities.
- 52:39And one of the things that defines the Western Western
- 52:41society and the Western alliance is democracies and these
- 52:45republics we have and in the United States, this wonderful
- 52:48Republic we have that in every generation has felt broken in
- 52:52some way. I always think it's funny when
- 52:54people we've never been more divided.
- 52:55Bitch, we fought a civil war like, like, like we this, this
- 52:59just what this looks like and, and, and no one's ever going to
- 53:02be totally happy with the current state of it.
- 53:04And that is kind of just how these things work.
- 53:06And I, I don't say that to be like fatalistic or even like, Oh
- 53:11well, just whatever the government use the software
- 53:13however you want. I just think it's like the, the,
- 53:16the voters take the there's a saying I liked, which is people
- 53:22deserve to get what they voted for good and hard.
- 53:25And you know what? They're getting a good and hard
- 53:27right now. I.
- 53:28Just I just think it we're still a liberal society where people
- 53:31participate or don't and like we are anthropic is being bullied
- 53:35into participating. Like the government is using the
- 53:37tools and you control whether you're supportive of them or
- 53:40not. You know, it's it my my issue is
- 53:43not yes, the Department of War is in charge of making war and
- 53:47anthropics in charge of using their technology how they want,
- 53:50and my issue is them using every lever they can to coerce
- 53:53anthropic to acting against this their moral.
- 53:56Beliefs, I mean, this administration has has been not
- 54:00been shy about employing coercive techniques.
- 54:03Look at law firms. I mean, there's a lot of.
- 54:05And a lot of people want to extract away the current
- 54:06administration, like whenever we're having this argument,
- 54:08they're like, oh, the the executive, whoever they may be.
- 54:12And it's like it's executive with the black record.
- 54:13People had critiques of the last administration in the way that
- 54:16they were acted coercively toward private enterprises and
- 54:19some sectors as well. This.
- 54:20Is by not inviting Tesla to Electro.
- 54:23These other examples people have stuff on crypto.
- 54:25I mean, I, I'm just sure I'm I'm.
- 54:26I'm, I'm just saying I'm just, yeah, I'm just.
- 54:28Saying like you can point these, this is not, this is not a novel
- 54:32new like criticism of government.
- 54:35I, you were right, we live in a liberal democracy and I and I
- 54:41one way you could look at this as anthropic is getting bullied.
- 54:44Another and something I will say I admire is anthropic is saying
- 54:47we're patriotic Americans. We want our government to have a
- 54:50great capability. We are proud of the product we
- 54:52make. We want it to be employed and we
- 54:54just have some concerns and limitations on where it can be
- 54:56employed. That is a perfectly valid place
- 54:59to be. And the government, if you take
- 55:01them at face value, is saying we're we want to work with
- 55:04private enterprise, but we need to have the flexibility to go
- 55:08engage in the things we have to engage in to do the things we
- 55:10want to do. Emil Michael on that post was
- 55:12like, you know, we do a lot of stuff with guns, but if you're
- 55:16not comfortable with, I'm paraphrasing, but if you're not
- 55:18comfortable with that, maybe you can't work with us.
- 55:20And like, I do think that's one of the things you kind of have
- 55:22to stare into like, again, going back to the time of Palantir,
- 55:26like there's things we did and worked on that you kind of look
- 55:28at and you're like, not because it's like morally repugnant, but
- 55:32it's like, does any you know, no, I would certainly never
- 55:36stand up and like celebrate like people dying.
- 55:39Like even if there are adversaries, there's something
- 55:41there's just, you know, people kind of look at that and it's
- 55:43like you can the the reluctant warrior, right?
- 55:45It's like the and to engage in these areas is to have to engage
- 55:51on the the reality of this, the Gray areas, the the grizzly on
- 55:56the ground way that the sausage is kind of made.
- 56:00And I think you can try to abstract that away, but it's
- 56:03it's. I just think Ethereopia picked
- 56:04this fight at a very savvy time. I mean, we just went to war with
- 56:08Iran. We bombed a school, you know
- 56:11like no all in they have no plan like.
- 56:13It was all in Claude's plan. They were about to be wrapped
- 56:16into this whole thing like, I don't know, I I just think that
- 56:19some of us as if Claude. Planned it because.
- 56:21People were deferring to our anyway.
- 56:24I don't know. I, I, I'm not trying to get
- 56:28overly involved. All I'll say is I have a lot of
- 56:31optimism and hope this gets worked out.
- 56:33And I have a lot of optimism and hope for how AI can just improve
- 56:36the way our government works and improve the way our military
- 56:38works when we need to unsheath it.
- 56:41And I think that it is our job as citizens to be engaged, to
- 56:48vote, to lobby our representatives, to make sure
- 56:52that if there are limitations, we want to add around that, that
- 56:55those are done. And I think we're going to see
- 56:57that play out in the Democratic cycles ahead, as we always have.
- 57:00That's how this process works, and we will watch it work.
- 57:03Yeah, that's my take. The we could.
- 57:06I want to move on to another topic.
- 57:08I mean, our last topic before I let you go, just raising money
- 57:11from venture capitalists. I feel like I have a lot of VCs
- 57:13on on the show and, you know, they tell their story about how
- 57:16things work. As a founder, I'm curious like
- 57:19what you really look for in VCs or who you think's doing it well
- 57:24right now. One thing I try to do is I when
- 57:28I talked. When I do like rounds of VC
- 57:29catch UPS, either current investors or perspective ones, I
- 57:33try to have the same conversation each time, like
- 57:35pick a topic. It's very interesting kind of
- 57:38doing this side by side and not like it's not one-dimensional
- 57:40like 1 is better or worse, just like a lot of different
- 57:43dimensions. And one of the things I feel
- 57:45like I've I've come to really look for is like first
- 57:47principles understanding of things.
- 57:48So like we were talking about pricing and so about pricing
- 57:51like the, the kind of like thin VC version is like kind of
- 57:56regurgitating A blog post and sort of while these other people
- 57:59are doing this and these guys are doing this and here's what
- 58:02margin they have in all this. I'm always trying to figure out
- 58:04like the first principles version of it of like, if we
- 58:07really strip it back and understand the technology and
- 58:10understand our customers, what do they want?
- 58:13And you can kind of tell when AVC has been like really
- 58:17involved and very thoughtful with one of their portfolio
- 58:19companies where they can say it. They're like, oh, well, at this
- 58:22company, the way we looked at it is this.
- 58:24And what we found is this, and we had this one customer told us
- 58:26this and that made us think about this.
- 58:28And you know, like, like they can really like unpack it and
- 58:32it's very different from like sort of the blog post version
- 58:35synthesis. So that's something I feel like
- 58:37I, I really, really look for and someone who's like thinking
- 58:40about things clearly and with fresh eyes and people, you know,
- 58:44there's like a sort of criticism of VCs like the herd animals or
- 58:49whatever. And there is a lot of that look.
- 58:51But I'm always interested when I when I talked to someone who's
- 58:54got like a really novel insider opinion or a contrarian take cuz
- 58:58I at least I know I'm getting like fresh thinking it is.
- 59:02I feel like I used to roll my eyes at first principles
- 59:05thinking, I think now that I have a business, I mean, you
- 59:07know, media is its own sort of animal.
- 59:10But there are things where it's like if you apply generic
- 59:13principles of like well run media companies, some I are
- 59:17lessons I should definitely learn and some you'd say, oh,
- 59:20you really don't understand some of the core things that make my
- 59:24business work. And you really do need somebody
- 59:26who has like empathy for like what is working about your
- 59:30business and what is sort of like the core sort of non
- 59:32negotiable. And if you have someone just
- 59:34sort of sloppily applying like just like heuristics without
- 59:39understanding like what's really working it can, it could send
- 59:42you on a total. Totally.
- 59:44I do a lot of Angel investing. I love it.
- 59:46It's like a great, I'm busy with my day job, but it's actually
- 59:49like a great kind of like mental stimulus that makes me think
- 59:51more about hex and I'm I try to be really careful.
- 59:53Like if you're a founder coming to me asking for advice, you
- 59:56have to sit through my 5 minute like disclaimer this.
- 59:58Is why you shouldn't trust anyone else you are in.
- 1:00:00It in a one. This is like, I'm probably wrong
- 1:00:03taking the greatest stop because I, I just, I like I and what I
- 1:00:06try to do as best I can is tell people, I will tell you what we
- 1:00:10did and why I think it worked or didn't work.
- 1:00:13I will try to unpack from first principles the facts in the
- 1:00:15ground, what we've seen, the data to back it up.
- 1:00:18And then you can draw a conclusion on whether those like
- 1:00:21ground facts, how well those map to you 'cause that I think is
- 1:00:25the thing you're trying to do. And I find this experience
- 1:00:27myself. I'll go talk to founders and get
- 1:00:28advice, talk to three really smart founders and get 5
- 1:00:31different opinions. And it's, it's like what?
- 1:00:34They're all really smart. Like Dylan at Figma said this
- 1:00:38and Kazar, an applied intuition said this and you know, I'm
- 1:00:41like, Oh God, that there are opposite things.
- 1:00:43What do you do? And it's like, well, you did you
- 1:00:45really understand the like grounding thing and, and the
- 1:00:50path dependency. And it's the other part of this
- 1:00:53job is like, it's a lot of luck and it's a lot of circumstance.
- 1:00:56And you know, you ask, how did you find that great executive?
- 1:00:59Oh, well, I did the search like this, but like, also like
- 1:01:01someone happened to be available on the market and you know,
- 1:01:06it's, it's, it's, it's always like the survivorship bias thing
- 1:01:09too of like, well, this worked for this person, therefore it'll
- 1:01:13work for me. Or the failureship bias thing
- 1:01:14that you also hear from a lot of VCs, which is, well, this other
- 1:01:17company tried to do that and it didn't work, Therefore it won't
- 1:01:20work for you. And So what was the first
- 1:01:21principle of why it did or didn't work?
- 1:01:24And can you really understand that?
- 1:01:26That's like a a thing I'm always trying to grasp for as best I
- 1:01:30can and usually failing, but sometimes get it ready.
- 1:01:33Gary, thank you for coming on the show.
- 1:01:34Thank you, this is fun. Great.
- 1:01:36That's our episode and thanks for sticking around to the end.
- 1:01:38Please, if you've made it this far, you've got a like comment,
- 1:01:41subscribe. Excited to grow the channel?
- 1:01:44And of course, you can find our writing and reporting on sharps
- 1:01:47and venture capital at newcomer.co.
- 1:01:51We also host events. You can check out what we're
- 1:01:53doing at Newcomer dot events, share their podcasts, help us
- 1:01:57get distribution, support the channel, comment support, like
- 1:02:00comment, subscribe. You know, you know the deal.
- 1:02:03Thanks. Thanks for being on the journey
- 1:02:04with us. All right, see you next week.