Palantir builds for what clients need, not what traps them

Transcript

Transcribed by machine (Whisper) from the episode’s audio, so a word may be off here and there.

1. 22:41

   different than traditional SaaS companies? Well, first of all, it's important to remember, we are valued. We drive our whole business based on a set of beliefs that are like America superiority and so on. But then, interestingly, that's tethered to the value creation argument you're asking, which is at the end of the day, if you believe your client is noble, like I believe our US government clients are not only noble, but they're no more noble than I am. And I'm happy with myself, but they're doing missions that will change world history for the better. Okay. So if you know that the thing that's going to happen downstream from that is you're going to resist building parasitic products right because like i don't want a product that will make me despite what anyone but like anyone who's been a palantir will know this to be true

2. 23:37

   we i did we did not the audacity of just saying we're going to build a product in pg that's going to make you safer let's go away from the ambient noise uh um that will make you uh safer and more protected is audacious in a weird way that we didn't realize the most audacious part of that was saying we're going to build a product that's actually what the client needs because what you're supposed to be doing software is build the client that the client gets addicted to not the one they need and like this is like and this is absolutely a truism in in in software development of any kind and so the like the again the most radical thing we ever did was say yeah The client, or in this case, partner that we look up to has these needs.

3. 24:29

   We didn't even give them what they asked for. We gave them what they ought to ask for, which, again, is very artistic. It's like any rational business would have said, I don't know, client A, B, C, D wants to be, at the very least, you just give them what they want. That's kind of what integrators do. And then what very successful software companies do is say, oh, they think they want X. We'll fake give them X, but we'll give them this thing that they can't get rid of.

4. 24:53

   And then nothing else will work, but they'll just have to pay. And then I can hire 50,000 salespeople and get these wonderful economics. And then every single decision, like, again, the FDE thing, now everyone copies it. But it was the most radical, like, you had everyone making fun of this for 20 years because that meant you were going to get a one multiple. But there was no way to extend PG. like essentially PG, you can think of PG as like a precursor to AIP because you have text, free text that a computer could not understand. The NLP technology was primitive. The only way to extend that product to like the bounds of technology now was to have people doing the impulse. And then the only way to build the next product or deepen that product was to understand what the customer ought to want tomorrow and build that into the product today. And so like

5. 25:41

   Like that value creation, by the way, it ends up being very, very protective because and and then because the customer partner actually becomes they may not like you. Like some of our best clients globally, both in government and in commercial, we're like, I don't really like these guys. Yeah. But you know what? When you bring someone home safely, they learn to love you. And so and then that allows us to have this privileged position that we now have.

6. 26:10

   And now when we're launching, you know, when we, again, there's example, the most recent example that people will be more familiar with would be launching AIP. Okay. When we launched AIP, or I launched it in like an artistic, I don't know if you know this story. But in any case, I launched AIP in like the darkness of night. The darkness of night. Yeah. Okay. Yeah. On like pre-Easter holiday because it would have been massively resisted.

7. 26:37

   And I would say, as one of my favorite people said in any other company, the whole head of whatever part of the product that was going to be involved in this would have quit because it was just launched. Completely artistic, like this is where the world is going. We have all the components. We had ontology of foundry. But the components need to be brought together. Okay. The thing is, if you're working back from what will sell, you would have said, well, wait a minute.

8. 27:04

   Everyone's saying LLMs are going to solve all these problems. but I believed and it turns out that LLMs are commodity products and orchestration would be much more valuable than the product themselves and you need both we need the LLMs but in reality and that's part of the reason we have these numbers because it turns out we're right but like if you were working back from what experts think and what customers are asking for you do not launch that you launch something else