Joe Lonsdale

It was kind of a cat and mouse game because every time you'd figure out what the bad guys were doing, they'd change their method. So we tried to use AI. It wasn't nearly as good 20 years ago. The AI could detect some things, but you really needed the human intelligence layers. You'd have a machine learning layer and you'd have these tools to let people see what was going on.

And you'd keep just iterating, staying ahead of them. And the tools were good enough. We cut down the fraud by about 90%. And that made it profitable. And then PayPal was sold to eBay. And so the anti-fraud thing was a big piece of what made PayPal work. And we got to know these Secret Service guys. And these are good guys. A lot of them are good old boys. They're not tech guys.

They're just trying to figure out what the heck this internet thing is and how to deal with it and how to catch the bad guys. And so they'd come to us for advice. And I got to know quite a few of them. And they'd come to me for advice on other stuff. And we started chatting with them, because cybercrime was a new thing, and you're helping them out.

And then 9-11 happened, and then we kind of saw the government spend billions of dollars trying to build new tools to help them do better, stop future stuff. And the stuff they were building horrified us. It was stuff that was based on principles from maybe like 20 years ago. And we're like, wait a second, guys. Silicon Valley exists. We've done all these new things with all the top talent.

And it was just so disconnecting. We realized this is actually really scary, because our country's spending... actually tens of billions of dollars on this stuff that doesn't work, that's 20 years out of date. It's just completely not what a top software culture is. And that's when we realized, well, we've got to figure out how to get involved and how to fix this. And how did you get involved?

We started Palantir with Stefan and I. I got about a bunch of my friends who were in PhD computer science programs one summer to come and sketch and draw it up with us. And they all thought we were totally crazy. So we couldn't convince them to join. A few of them joined a few years later. What is Palantir?

At a very high level, Palantir is an effort to take the very top technology culture in Silicon Valley and apply it to solve the most important problems in these institutions that didn't have tech cultures or the intelligence and defense world. But what is it actually doing? There was really initially four pillars of Palantir. It was data integration,

search discovery analysis, knowledge management, and collaboration. Each of those is a really big, hard product. So what happens if you have a government department? At the time, government was spending, say, $36 billion gathering data. So you're you. It's your job. There's like 5,000 databases. There's all sorts of signals and humans and other things coming in.

There's all sorts of rules about how you access this database, what you're allowed to see, depending on the context. What the hell do you do sitting in the middle of that? That's a weird, crazy problem. And you're a smart guy, but you're not a computer scientist. And so our job is to empower you.

Our job is to hook up to all the databases, integrate it so it all could be seen together, let you ask simple questions like, okay, take this guy we found next to Sam Bin Laden, show me any links to anyone else around him based on these contexts. Okay, now take those guys and monitor them. Do they show up in any databases? What do we know about them?

and just be able to kind of iteratively explore and analyze while not breaking the rules on what you're allowed to see and bringing things in more easily. So it's a hard problem to solve.

So Palantir today is different than Palantir then. Palantir then was all about organizing this information to extend human intelligence into this massive amount of data. Because there's no way that any single human is going to be able to keep 5,000, 20,000 databases of stuff in different formats in their mind at a time.

So you're going to have to organize it in a way you can interact with it, ask questions, preserve your investigations, share with others, collaborate. So that's a problem. Now it turns out that organizing all this information in all these ways, is very powerful to then apply AI on top of it.

As AI has gotten to be more advanced, there are now a lot more predictive, a lot more magical AI-like things that it could do thanks to that. Palantir was lucky in a way to have a top technology culture and to be solving these data organization ontology workflow problems, we call them, that when AI came along, it was really powerful to add AI to it and go even faster.

To be able to take all the data and figure that out. We first started working with a bunch of the special forces groups and we really helped them figure out some of those hard problems and partnered with some really smart guys that taught us, here's the data you should be looking at. We brought it together with our partnership and did it.

And then the Army Brigade said, well, we need this too, because they needed it badly. And they couldn't pay for it, of course, because they have some giant bureaucratic process. We just gave it to them. We said, just show us the lives you're saving. That's all we want to see. That's really inspiring to our engineers. And they started showing us the lives they were saving.

And then it came up for bid. It's called the Defense Ground Control System, DSIGS, remember? It was called back then, I remember. And of course, some general gave it to his friend for like $5 billion at some other company. And everyone protested, like, we're using Palantir. We don't want to wait years for this giant contract. So we ended up suing the government.

And I never sue anyone, but Palantir had to sue because they purposely just gave it to their friends, and we won. And it took years, but they eventually used us. You know what was shocking to me?

is when they finally eventually used Palantir, I'd met a bunch of the guys, and they all said, it's the best thing ever, it's amazing, we thought you guys were fake, we thought you guys were liars, because the people who were competing against us had just talked shit about us for years. But we finally got in and made it work. Man, that had to be... That had to be pretty enraging, too.

It's pretty frustrating when you're saving lives and you're doing it for free and you have all eyes forward to play people just working their asses off to really help and then they treat you like crap.