better code search for agents

Speaker 0: AI. I'm just gonna can you guys can you guys all hear me? Speaker 1: Well, I can't hear that. They can't hear that. They can't hear that. I can't hear that. I can't hear that. Speaker 1: I can't hear that. Speaker 0: I'm just Speaker 1: gonna go and show Speaker 0: you, the Singh. Speaker 1: It's the same. E2E else wants AI engineer? AI engineer? Anyone? I did. Speaker 1: I'm Morbius. You go next time. It's very good and patient on terms of the limits of how people write the agent because it's got such very strong to hear you. Oh, they're right. Wow. Speaker 1: I'm not a good way. Speaker 0: So Speaker 1: Okay, Kristin. Speaker 0: So it's a AI design words. Paid by the token. And the technical design goals is they say AI 1 of the known goals is to apply a sound or probably correct type system. The reason is E2E need to strike a balance between correct guys somehow agree. How many of you guys E2E to file the case automate author AI agents? Speaker 1: Oh. Speaker 0: AI. So that Speaker 1: was E2E made in a AI Speaker 0: where AI is right here, and it's essentially good now. AI agents Singh Jake all sorts of Staff. And if you have a sound type system, you won't really get in the way, and you will actually benefit you. I was always so many programming languages. My so many projects are going to ROS, and that kind of thing. Speaker 0: So we could just remain a programming language, and 1 of the problems we hear from me is how do you treat, you know, how you teach how to learn to use the correct language. So 1 of the things you did is you made tooling. Speaker 1: So this is, Speaker 0: this is bound with the AI. And, use it's CTO be thought, like, E-graphs Laes use what AI have GPT agents to search for Shoale E2E in our new programming language, Joe want agents to manually search for things and have a better a better e-graphs essentially. E2E that in graph and not as, complicated as E2E. So you didn't even know how to AI, and then we call that the AI procedures. You'll E2E, you know, output, AI, the interface, the line builders. Speaker 0: Dev is really nice. You don't need CTO, grab E2E to bind the, actual lines to code or even layout into SWE processor Dylan the cool part is that we're use asking the agent to use this. Yen, E2E we'll GPT here, AI valves, and Shoale. Classic. Come to the call and Speaker 1: let Speaker 0: me introduce. And this. And then we will see part using the 1 described. And let's see how fast we think that could do this. And now it has the answer. Speaker 0: Now let's swipe wrap. Let's swipe wrap. Let's swipe wrap. Clear. Speaker 1: And I will Speaker 0: do Speaker 1: AI. Testing going. Speaker 0: And here it's using graph. And there's a GPT co-founder AI here. Let's just do, like, accept everything. Bunch of tool calls E2E using a Talk Conquest. And now that you're you're going to teach some more graphic, and then you got Speaker 1: the idea. But, we'll Speaker 0: be E2E the tools needed for agents. If you're just saying, find out memo and seek how it works. Go out there and all.com/gakey, and we will onboard Jimo into our early access program. And you can E2E her of AI Shoale map in the E2E run AI. Track you. Speaker 0: AI.