Shawn Wang: recommendation

Transcript context

…Yeah, absolutely. Cool. And then I think like that's very classic, just supply chain problem that like all registries effectively have. And the, you know, there are different approaches to this problem. Like you can take the Apple approach and like vet things and like have like an army of, of both automated system and review teams to do this. And then you effectively build an app store, right? That's, that's one approach to this type of problem. It kind of works in, you know, in a very set, certain set of ways. But I don't think it works in an open source kind of ecosystem for which you always have a registry kind of approach, like similar to MPM and packages and PiPi. And they all have inherently these, like these, these supply chain attack problems, right? Yeah, yeah, totally. Quick time check. I think we're going to go for another like 20, 25 minutes. Is that okay for you guys? Okay, awesome. Cool. I wanted to double click, take the time. So I'm going to sort of, we previewed a little bit on like the future coming stuff. So I want to leave the future coming stuff to the end, like registry, the, the, the stateless servers and remote servers, all the other stuff. But I wanted to double click a little bit. A little bit more on the launch, the core servers that are part of the official repo. And some of them are special ones, like the, like the ones we already talked about. So let me just pull them up already. So for example, you mentioned memory, you mentioned sequential thinking. And I think I really, really encourage people should look at these, what I call special servers. Like they're, they're not normal servers in the, in the sense that they, they wrap some API and it's just easier to interact with those than to work at the APIs. And so I'll, I'll highlight the, the memory one first, just because like, I think there are, there are a few memory startups, but actually you don't need them if you just use this one. It's also like 200 lines of code. It's super simple. And, and obviously then if you need to scale it up, you should probably do some, some more battle tested thing. But if you're interested, if you're just introducing memory, I think this is a really good implementation. I don't know if there's like special stories that you want to highlight with, with some of these. I think, no, I don't, I don't think there's special stories. I think a lot of these, not all of them, but a lot of them originated from that hackathon that I mentioned before, where folks got excited about the idea of MCP. People internally inside Anthropik who wanted to have memory or like wanted to play around with the idea could quickly now prototype something using MCP in a way that wasn't possible before. Someone who's not like, you know, you don't have to become the, the end to end expert. You don't have access. You don't have to have access to this. Like, you know. You don't have to have this private, you know, proprietary code base. You can just now extend cloud with this memory capability. So that's how a lot of these came about. And then also just thinking about like, you know, what is the breadth of functionality that we want to demonstrate at launch?…

Stored transcript either side of the excerpt. The highlighted words are the published quote; the surrounding text is unedited source, never generated.