Why I couldn't build Jev at OpenAI — Diogo Almeida, TypeSafe Co-founder & CEO

1. 0:00 – 1:29

   Intro

   1. DADiogo Almeida0:00

      Like, how can AI be so unbelievably smart? How can we, like, solve Millennium Prize problems in math but still not automate even the most basics of works? Like, like really basic rote stuff, but we have this, like, supercharged engine of automation that just does not have, like, the right plugs and stuff to plug into all of this economically valuable work. And, you know, like, if the whole company of TypeSafe disappears, like, maybe it'll take, like, a year or two for people to, like, truly catch up. I actually don't know how long it'll take. If, uh-- If model quality matters, then we are gonna be in a very good position for a long time. But, um, it-it, like, it-it, it's done, right? Like, there-- Like, this has changed the path of, like, technological history.

   2. SPSpeaker0:44

      Before we get into today's episode, I just have a small message for listeners. Thank you. We would not be able to bring you the AI engineering, science, and entertainment content that you so clearly want if you didn't choose to also click in and tune into our content. We've been approached by sponsors on an almost daily basis, but fortunately, enough of you actually subscribe to us to keep all this sustainable without ads, and we wanna keep it that way. But I just have one favor to ask all of you. The single most powerful, completely free thing you can do is to click that subscribe button. It's the only thing I'll ever ask of you, and it means absolutely everything to me and my team that works so hard to bring The In Space to you each and every week. If you do it, I promise you, we'll never stop working to make this show even better. Now let's get into it.

2. 1:29 – 4:19

   Jev Launch Week and the AI Economic Revolution

   1. SPSpeaker1:29

      [upbeat music] Okay, we're in the studio. Uh, special occasion because this week, uh, Diogo, my, my good buddy, uh, launched Jev, and, uh, it's been taking over the complete timeline. Uh, how do you feel? What, what's it like to be you right now?

   2. DADiogo Almeida1:45

      Emotionally?

   3. SPSpeaker1:46

      Yeah.

   4. DADiogo Almeida1:46

      Um, never been worse.

   5. SPSpeaker1:47

      [laughs]

   6. DADiogo Almeida1:48

      Um, [laughs] like, uh, I'm a ragged corpse of a person right now because there's so much going on, and I'm, like, a technical CEO, so I have, like, a lot of fires to fight.

   7. SPSpeaker1:58

      Yeah.

   8. DADiogo Almeida1:58

      But, like, mentally, it's-- Like, I feel-- I, I say this all the time, and I've been saying this kind of for years in my over/under events. Like, I feel like the entire AI field is like one of those, like, carnival house of mirrors, and everyone is just insane and saying the weirdest stuff that doesn't make sense. And it feels like for just this week, like I'm, I'm on a better in sync with reality and like, oh, people see it now. AI can be so much more than what was once thought.

   9. SPSpeaker2:34

      Mm.

   10. DADiogo Almeida2:35

       And like, yes, we are going to make-- like, like, an AI-based economic revolution is back on the table, and, um, this is fucking awesome. You know? This is fucking awesome.

   11. SPSpeaker2:45

       [laughs]

   12. DADiogo Almeida2:46

       I'm so jazzed the developers get it. It's, it's, uh... Yeah, and I want to show my internal gratitude to the developers and-

   13. SPSpeaker2:54

       Yeah

   14. DADiogo Almeida2:54

       ... I'm so jammed about the community and everything. It's so great.

   15. SPSpeaker2:57

       Yeah, you were saying yesterday that you decided to prioritize the town, town hall and not a bunch of, like, you know, VIP inv- investor-type people because you wanted to make sure that they are the people that you get, get your most, uh, a-attention, right? The engineers, the developers.

   16. DADiogo Almeida3:12

       Yeah, it felt a little like, "Oh man, I'm talking to, like, really important people right now."

   17. SPSpeaker3:16

       Yeah.

   18. DADiogo Almeida3:16

       I probably shouldn't reveal who.

   19. SPSpeaker3:17

       Yeah.

   20. DADiogo Almeida3:17

       But it feels a little bit dirty for me to, uh-- I, I'm, like, perhaps overly genuine in things. Like, it, it feels, like, dirty if, like, in my gigantic calendar event of things to, people to talk to, you know, the community isn't one of those.

   21. SPSpeaker3:33

       Yeah.

   22. DADiogo Almeida3:33

       You know? And actually, in my ideal world, it would be, like, community all the time. I was thinking, "Should I host a town hall while walking to your studio?" And I'm like, "Nah, that's too crazy."

   23. SPSpeaker3:41

       Sure. Yeah. [laughs] Well, you, you guys have been hosting town halls on the Discord. Discord is now a hundred thousand people. Um, your, your Twitter is-

   24. DADiogo Almeida3:48

       I don't follow these stats.

   25. SPSpeaker3:49

       Yeah.

   26. DADiogo Almeida3:49

       So holy shit.

   27. SPSpeaker3:50

       Your Twitter is blown up. Uh, you know, uh, it was, it was really funny 'cause, like, at AIE, you were like, "Uh, follow me, please," and then you didn't, like, provide even your, your handle. So [laughs]

   28. DADiogo Almeida3:58

       I'm a noob. I'm a noob.

   29. SPSpeaker3:58

       You're such a noob.

   30. DADiogo Almeida3:59

       I'm a noob.
3. 4:19 – 7:23

   What Is Jev? System One Models and Programmable AI

   1. SPSpeaker4:19

      on everything.

   2. DADiogo Almeida4:19

      Thank you.

   3. SPSpeaker4:19

      We'll talk about more, uh, details a-as you have them, but let's, for people who are, like, living under a rock or just-

   4. DADiogo Almeida4:25

      [sneezes]

   5. SPSpeaker4:26

      Just want, like, the definitive thing, what is Jev?

   6. DADiogo Almeida4:29

      [sighs] Let me think about-- That's a hard one.

   7. SPSpeaker4:35

      [laughs]

   8. DADiogo Almeida4:35

      Um-

   9. SPSpeaker4:35

      Okay, and, and I'm happy to, like, uh, re-ask if you wanna correct-

   10. DADiogo Almeida4:38

       No, no, no. I'm happy to-

   11. SPSpeaker4:39

       Okay

   12. DADiogo Almeida4:39

       ... I'm happy to, like, just jam on it.

   13. SPSpeaker4:41

       Yeah.

   14. DADiogo Almeida4:41

       I will say, like, the first thing that I'm relieved about with this question is now I don't have to answer that question to my parents anymore, 'cause ChatGPT can just explain it.

   15. SPSpeaker4:49

       [laughs] Nice.

   16. DADiogo Almeida4:50

       Um, so, uh, the way I see it is we ne-need a new class of models. Um, we're not attached to naming that class of models. Um, our the most accurate name we've come up with is System One Models.

   17. SPSpeaker5:02

       Yeah.

   18. DADiogo Almeida5:02

       There will be reasons, but it's-- There's a reason why we don't call them decision models because, like, they will be-- Like, System One is beyond that. Uh, that's all I can say. Um, we didn't expect this to be our big launch, so we have stuff in the tank. Um-

   19. SPSpeaker5:17

       You should've said low-key research preview. [laughs]

   20. DADiogo Almeida5:20

       Um, it kinda was, right? It kinda was. But, um, we-- So there's a class of models that we describe them as, like, machine native, System One, large programmable.

   21. SPSpeaker5:30

       Mm-hmm.

   22. DADiogo Almeida5:30

       I think these are is the class of models, where the goal is for code to be the consumer. So, um, as opposed to, um, lar- um, pre-trained large language models, which are meant for, like, auto-complete of the internet or RLHF models like chatbot instruction-following models, which are meant to, like, reply to text, um, or RLVR. It's in a weird gray area with R-RLHF. Like, these are meant to have things that directly are consumed by code, hence the name TypeSafe. So the thing we really, really, really want is to have, like, AI Like, be as powerful as possible, and we think the way to do that is to integrate it with software. And we are designing everything, you know, beyond just the outside, the deep internals of the model to be optimized for software. So, uh, number one, Jev is our first large programmable model, um, or a system-one model, whatever you want to call it. Um, Jev is meant to be optimized for intelligence per dollar, uh, hence the name Jev. You know, uh, I mean-

   23. SPSpeaker6:32

       Jevons Paradox

   24. DADiogo Almeida6:33

       ... Jevons- Jevons Paradox, yeah. Um, and it's optimized for intelligence per dollar. I love this debate with people about what is the most important between reliability, cost, calibration, and speed. Um, and Jev is meant to be... Jev will be the name of models that will be on the frontier of intelligence per dollar. There's other ways to optimize it, like, uh, ML, or at least if you're good at ML, it's all about trade-offs. And, you know, we are just going all out on that.

   25. SPSpeaker7:00

       Yeah. And, and to me, like, calibration is one of the new things that people weren't talking about as much. We've done an episode-

   26. DADiogo Almeida7:06

       Mm-hmm

   27. SPSpeaker7:06

       ... uh, in the past, uh, with Clémentine Foreyo of Hugging Face, uh, where they were like, yeah, a-actually, um, y- they are just are... You know, and this is your, your argument about RLHF, is they're mode collapsing towards what you want to hear the most-

   28. DADiogo Almeida7:19

       Ooh

   29. SPSpeaker7:19

       ... or what is most likely, uh, instead of, like, their own internal confidence about

4. 7:23 – 11:58

   RLHF, Mode Collapse, Calibration, and Yann LeCun

   1. SPSpeaker7:23

      a, a, a thing.

   2. DADiogo Almeida7:23

      Can I soapbox on that for a second?

   3. SPSpeaker7:25

      Go ahead.

   4. DADiogo Almeida7:25

      Or-

   5. SPSpeaker7:25

      Yeah, yeah

   6. DADiogo Almeida7:26

      ... cool. Uh, like, um, uh, I've been heard that your audience is the most technical, so I actually want to get into that.

   7. SPSpeaker7:31

      [laughs] Yeah.

   8. DADiogo Almeida7:32

      Um, and if-- I went through extreme precision to make sure everything in our launch video is accurate and real. Apparently, that's very unusual. One of the things that no one paid attention to was the downsides of RLHF, in particular, mode dropping. Um-

   9. SPSpeaker7:46

      Mode dropping or mode collapse?

   10. DADiogo Almeida7:48

       Uh, it's the same thing.

   11. SPSpeaker7:48

       Is that what-- How you gonna-

   12. DADiogo Almeida7:49

       It's the same thing.

   13. SPSpeaker7:49

       All right.

   14. DADiogo Almeida7:50

       And, um, I, I, I wanna have a blog on this eventually, but I, like, want to tell as many people this as possible 'cause I think it's a very interesting thing. So the spicy take, I believe in Yann LeCun a lot. Uh, I think Yann LeCun's takes are actually among the closest to-

   15. SPSpeaker8:05

       What, what about this? [laughs]

   16. DADiogo Almeida8:06

       Well, do you want to address this now or should I go on mode collapse?

   17. SPSpeaker8:09

       No, no, no, later.

   18. DADiogo Almeida8:09

       Okay, okay.

   19. SPSpeaker8:10

       Go, go, go mode collapse. I don't know.

   20. DADiogo Almeida8:11

       So I actually think that among takes, Yann LeCun's is among the most accurate. Um, but he has this very famous/infamous slide about, uh-

   21. SPSpeaker8:21

       The cake

   22. DADiogo Almeida8:21

       ... L- LLMs are doomed.

   23. SPSpeaker8:23

       Okay.

   24. DADiogo Almeida8:23

       You know, like that one where he's, uh, where he, like, has, like, a pie chart with, like, this tiny par-- tiny little thing-

   25. SPSpeaker8:28

       Oh, that

   26. DADiogo Almeida8:28

       ... and says that as you increase sequence length, um, the probability of it making an error goes to... Yes, this one, this one. I love this one, um, because it's one of these things that seems mathematically obvious but is obviously wrong, right? Like, it's mathematically obvious, but it doesn't empirically hold, and this is my favorite thing to teach people about. Like, where you, you-

   27. SPSpeaker8:50

       What's the disconnect, right?

   28. DADiogo Almeida8:51

       Exactly. And, um, may I or you want to tell me, uh-

   29. SPSpeaker8:56

       About mode collapse?

   30. DADiogo Almeida8:57

       Oh, no, no. Uh, so mode coll-collapse is related to this.
5. 11:58 – 18:50

   Programmatic AI, Refusals, and Safety Alignment

   1. DADiogo Almeida11:58

      I just think there's, like, so many diamonds in the rough lit all over the research world right now that haven't been polished because people don't know how to, like, do the right task, and I think that what our launch did, it, does it kickstart us as a company? Like, yes. Will it be great for us as a company? Yes. I think it's gonna be, like, even greater for this direction of, like, programmatic AI. You know, there, there was going to be, like, a gold rush On top of us for- 'cause like software is super fucking charged. But I think there's gonna be a gold rush parallel to us as well on like all the different ways we can expose things to make software more powerful so people can make even cooler stuff, and then we are back to like early internet energy. You know?

   2. SPSpeaker12:41

      Yeah.

   3. DADiogo Almeida12:41

      And, and I think that's why like, you know, the Twitter is just like, "Jev, Jev, Jev." You know? Um, it's, it's like, it is like a party.

   4. SPSpeaker12:47

      It's inspiring because it's, it's like so different than what we're used to, which is, "I'm sorry you can't do this, but we do scaling laws and only the big labs can do it." Right?

   5. DADiogo Almeida12:57

      That, that-

   6. SPSpeaker12:57

      [laughs]

   7. DADiogo Almeida12:58

      ... uh, actually, c- if I... I'll, I'll nick tangent if that's okay.

   8. SPSpeaker13:01

      Yeah, yeah.

   9. DADiogo Almeida13:01

      I, I think you might enjoy this. Might-

   10. SPSpeaker13:02

       Really? Five, five tangents in. No, it's, it's going to be so it's fine. Yeah.

   11. DADiogo Almeida13:04

       Oh, it's... Yeah, I get lost in all my tangents.

   12. SPSpeaker13:06

       It, this is gonna be horrible for the listeners to figure it out, but they, they're gonna figure it out. It's fine.

   13. DADiogo Almeida13:09

       Yeah, we can edit it out.

   14. SPSpeaker13:09

       This morning. Yeah.

   15. DADiogo Almeida13:10

       Um, [laughs] so popular thing on Discord, um, that people keep asking me, I haven't had the time to explain it yet, is why am I opposed to safety alignment and why do we not refuse? Um, I'm not opposed to safety as a principle, but I think that safety alignment is generally misaligned with users. Um, and, uh, refusal is just like obviously a type error. Like if you're a human being and you're chatting with like a, you know, a model or whatever, you're cloud coding, and a refusal happens like, "I'm sorry, I can't read DNA.py," um, that's an annoying time. It's anno- it's, it's annoying, right?

   16. SPSpeaker13:47

       Mm-hmm.

   17. DADiogo Almeida13:47

       But you can work with it, right? And you're forced to work with it 'cause of Stockholm Syndrome. Um-

   18. SPSpeaker13:51

       [laughs]

   19. DADiogo Almeida13:51

       ... uh, t- I have stories about that t- too. I, I need another tangent deep in here. But like if you ever want this in a dependency running in the background, what happens if that refuses? What if someone else is using that dependency? They don't know what that system is. Like, like you want the software to just stochastically break because a user sent like a weird message in there?

   20. SPSpeaker14:11

       Mm.

   21. DADiogo Almeida14:11

       Like that is, um, like straight up insanity. It, it, it's coming from a place of like people who do not understand software, do not understand programming.

   22. SPSpeaker14:19

       Mm.

   23. DADiogo Almeida14:19

       And like they are obsessed with like, I believe this, um, horseless carriage of like AI coworker instead of unearthing like the full power of AI.

   24. SPSpeaker14:30

       Fair enough.

   25. DADiogo Almeida14:30

       Yeah.

   26. SPSpeaker14:30

       Uh, you, you want something that is the core kernel that is, is usable everywhere.

   27. DADiogo Almeida14:34

       Yes. Exactly. Like the cognitive core, right?

   28. SPSpeaker14:36

       Yeah.

   29. DADiogo Almeida14:37

       And you need this thing to be s- like so general, so optimized for its use cases. You want it to be like, you know, you want it to work on all the future use cases, all the weird shit that people are doing.

   30. SPSpeaker14:48

       Yeah.
6. 18:50 – 22:12

   Why TypeSafe Rejects Public Benchmarks

   1. DADiogo Almeida18:50

      Hell yeah.

   2. SPSpeaker18:50

      I think it's probably takes two sentences from you about, like, you will not, uh, y-you're not being that restrictive about your API. Like, clearly-

   3. DADiogo Almeida18:56

      Oh, yeah. Obviously, yeah

   4. SPSpeaker18:56

      ... ideologically, you would take y- your role as a platform very seriously.

   5. DADiogo Almeida18:59

      Yes. Uh, yes. Uh, uh, I don't know what you're referring to, but, like, this was-- Uh, I, I've seen a couple of things about, like, benchmarking.

   6. SPSpeaker19:07

      Yes.

   7. DADiogo Almeida19:07

      Like, obviously, we're not stopping people from do-- Oh, man, I should be careful about what I say, I'm realizing. [laughs]

   8. SPSpeaker19:12

      [laughs] No, you said it, you said it publicly that-

   9. DADiogo Almeida19:13

      Yeah

   10. SPSpeaker19:13

       ... that was in the preview period. You didn't take it out for the launch.

   11. DADiogo Almeida19:16

       Yeah. Okay.

   12. SPSpeaker19:16

       And now you're gonna take it out.

   13. DADiogo Almeida19:17

       So the team is doing stuff that-

   14. SPSpeaker19:18

       Yes

   15. DADiogo Almeida19:18

       ... I'm not even aware of, so it's great to know the team communicated that.

   16. SPSpeaker19:21

       Mm-hmm.

   17. DADiogo Almeida19:21

       I asked them to check in with the lawyers about that.

   18. SPSpeaker19:23

       Yeah.

   19. DADiogo Almeida19:23

       Like, we are obviously not stopping people from doing that type of thing. I am extremely in favor-- So I'm extremely anti-public benchmarks. Um, I'm extremely in fav-- I'm medium about private benchmarks that are proxies. Um, you know-

   20. SPSpeaker19:38

       So a-are you worried about, uh, sat-saturation or, like, uh, training on public benchmarks, so it's, like, e-easy to cheat?

   21. DADiogo Almeida19:44

       Um, not only is it easy to cheat, there's a lot of ins... So, uh, I think that we are-- Or, or anyone who's, like, competition with us that, you know, vaguely there is-- Like, you could say, like, the-

   22. SPSpeaker19:57

       There's, like, 50 Jev clones. Yeah.

   23. DADiogo Almeida19:59

       Well, uh, sure, sure.

   24. SPSpeaker20:00

       Yeah.

   25. DADiogo Almeida20:01

       Uh, w-these-- Let's say that there's various competition.

   26. SPSpeaker20:02

       And we'll talk about those. Yeah.

   27. DADiogo Almeida20:03

       Or let's just say that there's l- there-- Let's just assume that there's an industry two years from now of people who are doing similar things to us. The thing that we are selling is intelligence per something, per, like, dollar or per second. Um, the-- No o- Like, people obsess about the cost and the speed. Um, I believe that that is-- It's cool, but, like, the thing that matters is the intelligence. Like, the cost and the speed are, like, th- are bad things. You know, you're paying them for something, and you need the thing back, and the intelligence is what truly matters. The problem with intelligence is that there's a je ne sais quoi to it, right? Like, the, like, uh-

   28. SPSpeaker20:36

       Mm

   29. DADiogo Almeida20:36

       ... the, the good model smell. Like, the thing that happened after we launched of, like, two hours later that actually went way bigger than the video, which was like, "Holy shit"-

   30. SPSpeaker20:45

       This is actually usable
7. 22:12 – 26:28

   The Bitterest Lesson: Data, Compute, and the Right Task

   1. DADiogo Almeida22:12

      Hell yeah. Um, t- like, you know, Suchin says that algorithms beats compute very roughly. Um, data matters way more than compute, obviously. And doing the right task, having the North Star is, is, is the hardest, most important thing. This has happened, um, in LLM land twice so far, right? Maybe 2.2 times. You know, there's RLHF, which, like, shifted the task to instruction following. No one realized that that was possible. Um, RLVR did, like, a tiny little, like, edit to the, to the direction, and now us, right? RLCD. We have a new task, and the goal is, you know, programs in the loop. And, um, yeah, data matters so, so, so-

   2. SPSpeaker22:57

      Right

   3. DADiogo Almeida22:57

      ... unbelievably much.

   4. SPSpeaker22:58

      So, so-

   5. DADiogo Almeida22:58

      It's like I can't, I can't emphasize it less.

   6. SPSpeaker23:00

      Yeah. You consider yourself a data lab rather than, like, a model lab. Is that-

   7. DADiogo Almeida23:04

      Absolutely

   8. SPSpeaker23:04

      ... something-- That's the wording you guys use?

   9. DADiogo Almeida23:06

      Yeah, yeah, yeah. Uh, we are-- We will always, like, care so much about data. Um, to me, model capabilities means data. Data is so unbelievably complicated, and that is what gets nines. Like, you have no idea how, how much data can shift everything. Data is so important.

   10. SPSpeaker23:26

       Yeah.

   11. DADiogo Almeida23:26

       Holy crap.

   12. SPSpeaker23:27

       Um-

   13. DADiogo Almeida23:27

       So if people are looking for a job, um, we are hiring infinite data people. Actually infinite.

   14. SPSpeaker23:33

       [laughs] What is a good data person? Like, um, you know, clearly somebody who cares about reading through the, the transcripts of, uh, whatever. You said, for example, that y- all your data is synthetic.

   15. DADiogo Almeida23:44

       Yep.

   16. SPSpeaker23:44

       But, uh, that's only, like, the scratching the surface, right?

   17. DADiogo Almeida23:47

       Yeah.

   18. SPSpeaker23:48

       Like, it's not... Like, synthetic, so what, right? Synthetic, but we have a people with a lot of taste and a lot of care looking at, looking at these, articulating what's wrong, going back, regenerating. Is, is that what a good data person is these days?

   19. DADiogo Almeida23:59

       [sighs] Let me try to figure, figure out how to... Like, it's, it's super complicated, and, like, I literally onboard the data people with a Talk that I assume is longer than this podcast will end up being. So I will try to say, like, the high level of it. So number one, we don't do the kind of synthetic data that people ki- well, I'll do a-- actually, number zero. Um, data and synthetic data depends on your task. Like the shape of your data, uh, s- the shape of your task changes the data. Like RLVR's data is kind of environments, right?

   20. SPSpeaker24:32

       Yes.

   21. DADiogo Almeida24:33

       Um, RLHF's is the human feedback, you know? Um, each task has its own unique kind of data, and we, of course, have our own unique kind of data, right? Um, so number one, we have that. Uh, number two, um, the thing I, the reason why we don't want to train on our users' data even if we could, right? Like, we could probably ask for any terms right now, and it'll, w- w- we-- I don't know if it would make a difference. We truly don't want that, um, because no matter what, the real world data has so much bias. There's like a power law of, like, people, like, asking the same things where you'll end up, like, overfitting to it and, like, fracturing to it and all of that. And number two, we are, like, aiming for, like, a complete sci-fi future years from now where, like, these models are going to be, like, the general infrastructure, layers and layers and layers and deep down the stack to, to, like, things people can't even imagine. Like, I would like to think of our model, like, kind of like, like, you know, UDP as LLMs and TCP as our models. All sorts of stuff can be built on top of that, and we need to be able to nail those futuristic use cases such that software developers can actually build that futuristic stuff. And the way to do that is even if we had all of the data of the present, um, we would just overfit to the present, and then it wouldn't work. What we need is to, like, it, it, it almost feels like a, like, uh, they're, like artists. You know? They study this cognitive core. Our cognitive core is, like, way less jagged than anyone else's, and then they find the jaggednesses, and then they address them surgically in a way that, a-and you can never perfectly do this, right? But they do it in such a way that it addresses it in every single possible, like, dimension, past, present, future.

   22. SPSpeaker26:14

       The general case rather than the specific case.

   23. DADiogo Almeida26:16

       Exactly. And, like, that requires a lot of intelligence every time.

   24. SPSpeaker26:19

       Um, okay, so we mentioned a little bit, you, you, you sort of criticized my thinking as our, like, very RLVR influence, which is, like, very fair. Um, let, let's, let's actually mention RLCD.

8. 26:28 – 30:11

   RLCD vs. RLHF and RLVR

   1. DADiogo Almeida26:28

      Ooh.

   2. SPSpeaker26:28

      Um, which obviously you have some secret sauces. To my knowledge, you've never actually published a paper or anything like that on it. No, right?

   3. DADiogo Almeida26:34

      No, not yet.

   4. SPSpeaker26:35

      Um, but, like, what, what should people get from this? Like, what, uh, can you give people some confidence that you're just not just making up jargon for the sake of sounding cool, right? Like, uh, one thing for me is, like, calibration I do think is, uh, to me, like, well understood because we've covered it in, on the podcast.

   5. DADiogo Almeida26:51

      Yeah.

   6. SPSpeaker26:51

      But I, I don't know what you mean when you say RLCD versus what people are familiar with.

   7. DADiogo Almeida26:55

      Great. It's a great question.

   8. SPSpeaker26:56

      Yes.

   9. DADiogo Almeida26:56

      And actually, I will give a related question.

   10. SPSpeaker26:58

       Okay.

   11. DADiogo Almeida26:58

       Um, what is RLHF?

   12. SPSpeaker27:00

       Okay.

   13. DADiogo Almeida27:00

       Right? And actually, RLHF means multiple different things, right?

   14. SPSpeaker27:03

       Okay.

   15. DADiogo Almeida27:03

       Like, there's the RLHF of the original, I think it was, like, Paul Christiano teaching a robot to backflip or something like that. Uh, wasn't there something like that?

   16. SPSpeaker27:11

       Was that it?

   17. DADiogo Almeida27:12

       That, that was the original R-

   18. SPSpeaker27:13

       I, I referenced the PPO paper, but I don't, I don't know.

   19. DADiogo Almeida27:15

       Um, and, uh, so, uh, PPO is not necessarily from human feedback, if I recall.

   20. SPSpeaker27:20

       Okay.

   21. DADiogo Almeida27:20

       Um-

   22. SPSpeaker27:20

       That's true

   23. DADiogo Almeida27:21

       ... but I, but I believe it was, like, an OpenAI alignment work that could teach hard-to-specify outputs like a backflip. I'm not a hundred percent sure. Um, and then there was actually learning to summarize. You know, this was work, um, by a bunch of the, the team that helped with, um, instruct-- a-and co-authored, um, the instruction following paper, which was teaching doing PPO on language models.

   24. SPSpeaker27:44

       This is, this is the, uh, uh, sorry. Uh, I'm trying to, trying to [laughs]

   25. DADiogo Almeida27:48

       Yeah.

   26. SPSpeaker27:48

       Trying to manipulate this, this thing. Uh, this is twenty seventeen.

   27. DADiogo Almeida27:52

       Yeah.

   28. SPSpeaker27:52

       All right.

   29. DADiogo Almeida27:52

       Um, I-I'm not a hundred percent sure, but, like, that looks quite right.

   30. SPSpeaker27:55

       Yeah.
9. 30:11 – 41:24

   Why Powerful AI Still Hasn’t Automated the Economy

   1. DADiogo Almeida30:11

      But to me, like the pre-Jev world was a tragedy because it, it, it sounds arrogant. Uh, uh, hear me out.

   2. SPSpeaker30:17

      No, no, no. I strongly believe you. [laughs]

   3. DADiogo Almeida30:19

      Cool. Uh, it sounds arrogant, but like I felt this way since long before I even had-

   4. SPSpeaker30:23

      Yeah

   5. DADiogo Almeida30:23

      ... a company.

   6. SPSpeaker30:23

      And I can, I can vouch that, uh-

   7. DADiogo Almeida30:25

      Yes, I've been talking about this for so long

   8. SPSpeaker30:25

      ... you, you said this at Arounder for like three years.

   9. DADiogo Almeida30:27

      Yeah, I've been talking about this for so long. And, and I've been saying it because I thought it would've been easier. Um, they say they do not do things because they, it, they're-

   10. SPSpeaker30:35

       It is

   11. DADiogo Almeida30:35

       ... easy. They, 'cause they thought it was easy. So-

   12. SPSpeaker30:37

       Yeah, exactly. Exactly

   13. DADiogo Almeida30:37

       ... something like that.

   14. SPSpeaker30:38

       Exactly. Exactly.

   15. DADiogo Almeida30:38

       I, I thought it, this whole project would take a week. Um-

   16. SPSpeaker30:42

       [laughs]

   17. DADiogo Almeida30:42

       ... and I was unbelievably wrong. So I am so sorry to everyone at OpenAI [laughs] that I thought... I was like, "Man, I'm solving this right now." Um, but, uh, like I think that the tragic thing is when... Well, I think overpromise, under-delivery is tragic too, and like AI is super extreme on that axis, and I think RLVR is like the main... Well, both RLVR and RLHF are extreme perpetrators of this. Um, but like it, to me, it's like, it's just, there's just so much potential there. Like AI is clearly so smart. I, smart. I love this in my talks, you know, when I ask people like, "How can AI be so unbelievably smart? How can we like solve Millennium Prize problems in math, uh, but still not automate even the most basics of works?" Like, like really basic rote stuff that like the, you know, it, it, it doesn't take like extremely smart people to do this. It's not a satisfying job. Like there's other things these people could be doing, but yet we need them to do like this ba- like super basic, non- unsatisfying stuff because, you know, like we can't automate it yet. But we have this like supercharged engine of automation that just does not have like the right plugs and stuff to plug into all of this economically valuable work. And, you know, like if the whole company of TypeSafe disappears, like maybe it'll take like a year or two for people to like truly catch up. I actually don't know how long it'll take. If, uh, if model quality matters, then we are gonna be in a very good position for a long time. But, um, it, it, like it, it, it's done, right? Like there, uh, like this has changed the path of like technological history.

   18. SPSpeaker32:18

       Yeah.

   19. DADiogo Almeida32:18

       And like we will be exploring that space as a field.

   20. SPSpeaker32:22

       Yeah. I, I think I, I def- definitely agree with that. Um, you've created possibility. So I think, you know, if I can paraphrase so that people can un- also understand, uh, you should not take the success of TypeSafe and Jev as just like, well, you know, that is a new, new model type. Now we're done. We go back to business. Like, no. Like, actually, there sh- there are like five other model types that you should be exploring, and like let a thousand flowers bloom.

   21. DADiogo Almeida32:44

       Ab- absolutely.

   22. SPSpeaker32:45

       Right?

   23. DADiogo Almeida32:45

       Like early internet-

   24. SPSpeaker32:46

       And some of that, some of which you will probably also do.

   25. DADiogo Almeida32:47

       Of course, yes.

   26. SPSpeaker32:48

       Yes.

   27. DADiogo Almeida32:48

       Early internet energy.

   28. SPSpeaker32:49

       [laughs]

   29. DADiogo Almeida32:50

       I think it's back to tech utopia. You know, it's no longer like, oh man, like sometimes my coding agents work, but the, all of the best ones are hoarded internally.

   30. SPSpeaker32:58

       Yeah.
10. 41:24 – 49:40

    Reliability, Robustness, and Determinism

    1. SPSpeaker41:24

       of a kind.

    2. DADiogo Almeida41:24

       And nines.

    3. SPSpeaker41:25

       And nines.

    4. DADiogo Almeida41:25

       It's, uh, like, like-

    5. SPSpeaker41:26

       Which uptime is in my-

    6. DADiogo Almeida41:27

       Oh, uh, but, but that's part of it. But like there's reliability in, um, like how intelligent the thing is, like how consistently does it do the thing that you want? And I think that like the, the big reasoning models are very smart. In my opinion, they still lack reliability.

    7. SPSpeaker41:42

       Mm.

    8. DADiogo Almeida41:42

       I think there's many use cases where you, they look like they should be smart enough to automate their work. There is economic incentive at, to automate that work, yet still they're not reliable enough as at an intern because they're optimized for different things. Uh, so like I think that there's the reliability of being able to like trust the outputs and also we are... Like, like there are dimensions of reliability that we are not yet at that I'm like so excited by.

    9. SPSpeaker42:06

       Yeah.

    10. DADiogo Almeida42:07

        You know, like I want to automate the easy work before the hard work. You know, like, I think that that's just a common sense thing to do. Um, but to me, we will be sufficient... I don't know if there's a such thing as sufficiently reliable, but I wanna get so good that people don't even need to try the model to know that it'll work. It's like, that's, like, what flow state is in programming, right? Like, I'm just, like, writing queries because I need intelligence in here. And, you know, like, when, for non-trivial branching, I can just write it in in, like a, like, like a TypeSafe SystemOne query and then get the results out and it just branches accurately. Like, that would be so, so good. Like, that's the, that is the dream.

    11. SPSpeaker42:41

        Yeah.

    12. DADiogo Almeida42:41

        And that is, like, uh, going to be, like, a long, long slog.

    13. SPSpeaker42:45

        Yeah. We're gonna go into your API designing a little bit-

    14. DADiogo Almeida42:48

        Ooh

    15. SPSpeaker42:48

        ... just, just to give people examples and, like, maybe paths not taken, that kind of stuff.

    16. DADiogo Almeida42:51

        Mm-hmm.

    17. SPSpeaker42:52

        Um, one thing up the front that I do wonder about in terms of reliability is I notice that there's no seed. Uh, there's no, uh... And, and so basically, same input, do I always get the same output?

    18. DADiogo Almeida43:03

        Hmm. So-

    19. SPSpeaker43:05

        And if not, why not?

    20. DADiogo Almeida43:05

        Oh, great question. So this is actually, like, a common question we have between, um... So reliability is actually a catchall. Like, whenever AI can't automate something, it's due to some form of reliability. Could be, like, type safety. It could be determinism. It just could be, like, it's, uh, it's jagged, right? So, um, reliability is a catchall. Uh, I just think that it's also a catchall for, like, what the North Star is. Um, re- determinism is, like, same inputs, same outputs. I do believe that this is, like, slightly interesting for unit tests, but I believe that to be the wrong North Star. I, I believe robustness is what people... I, I don't wanna tell people what they really want, 'cause that would be a little arrogant of me.

    21. SPSpeaker43:48

        [laughs]

    22. DADiogo Almeida43:48

        I believe that that is, like, the more important property. Um, uh, you want, um, given similar inputs, get similar outputs. And it's kind of wild how unreliable LLMs are.

    23. SPSpeaker44:00

        Hmm.

    24. DADiogo Almeida44:00

        Uh, like, a, a way that we test this is you put, like, UUIDs in, in, like, little-

    25. SPSpeaker44:05

        Yeah, yeah

    26. DADiogo Almeida44:05

        ... I think they're called nonces-

    27. SPSpeaker44:06

        Mm-hmm

    28. DADiogo Almeida44:06

        ... in, in the prompt. And what you want is similar outputs from all of those 'cause it's truly semantically the same question, and that is the part where you really want... Th- like, like, like, that robustness is where, like, people get, like, burnt with AI making decisions. So I think that is the, a super-duper important property. We could also have determinism. That is, that is a thing that can be available. As far as I can, like, mentally model for programmers, like, it, it, it, it could be valuable for some use cases, so, like, please educate me, um, in comments or you. But, um, my s- in, in general, it's easy... Determinism is something you can, like, trade off for better cost. You know? Like, we are, we are constantly wanting to be on the intelligence per dollar frontier. We are doing, like, absolutely disgusting things to be there. You know, like, this is, um... I, I, I shouldn't say this, but no one's here to stop me.

    29. SPSpeaker45:05

        [laughs]

    30. DADiogo Almeida45:05

        Um, you know-
11. 49:40 – 55:33

    Model Versioning, LTS, Speed, and Intelligence per Dollar

    1. SPSpeaker49:40

       it.

    2. DADiogo Almeida49:40

       We will not change our models when we deploy them. That is insane. We care about developers. Like, like, like it makes sense if you're... So doing something like that, a- again, this is the problem with a for, first-party product and an API. It makes-- You can do whatever you want in a first-party product, right? Like, like more power to them, whatever gets that experience. That is fine. With an API, you obviously can't do that. But I will say that we, uh, plan to move a lot faster than many people are used to model providers, uh, doing things. So we will be launching new models a lot faster than people think, and we are not promising long-term support for the models because we think that there's lots of improvements to have. So, um, there is a world that we might temporarily LTS what is right now Jev 1.13.0. We might do that because so many people are using it, and I know developers hate breaking dependencies. The alternative is fracturing our fleet, and that is a very bad vibe for everyone. It's gonna be hard.

    3. SPSpeaker50:41

       Yeah, you can't have like 100 different versions of the model. Yeah.

    4. DADiogo Almeida50:42

       Exactly, and if we're iterating very fast, there would be a lot of those versions as well.

    5. SPSpeaker50:46

       Yeah.

    6. DADiogo Almeida50:46

       So, um, we, we do want to have not just a LTS-supported thing eventually, long-term support. Um, we want a really sick way of doing that. We have like research stuff cooking in that direction, and I think it's gonna be the most pro-developer thing ever.

    7. SPSpeaker51:03

       Yeah.

    8. DADiogo Almeida51:03

       Um, but it is not yet our current models, and I'm not promising that we will be able to keep the exact same models. They will get smarter every time for sure.

    9. SPSpeaker51:12

       Yeah.

    10. DADiogo Almeida51:12

        And my sense is that even our model iterations, where it already is smart, um, it, between model versions the, the changes tend to be even smaller than the string models calling them twice. But when, when it, we go from like, you know, jagged to like wow, um, that is where the big deltas are.

    11. SPSpeaker51:30

        Yeah. Uh, one, one thing, one thing that's beautiful about LTS same models is that actually you can also port them to other silicon.

    12. DADiogo Almeida51:37

        Mm-hmm.

    13. SPSpeaker51:37

        Uh, I don't, I don't know if you've thought about this.

    14. DADiogo Almeida51:40

        No comment. [both laughing]

    15. SPSpeaker51:41

        Okay.

    16. DADiogo Almeida51:41

        So I care about intelligence per dollar.

    17. SPSpeaker51:43

        Yes.

    18. DADiogo Almeida51:44

        Right? Um-

    19. SPSpeaker51:44

        But speed.

    20. DADiogo Almeida51:45

        What?

    21. SPSpeaker51:46

        Speed as well.

    22. DADiogo Almeida51:47

        We'll see.

    23. SPSpeaker51:48

        Yeah.

    24. DADiogo Almeida51:48

        We'll see. Honestly-

    25. SPSpeaker51:49

        I mean, it's a whole part of the inference tech tree that is like, I mean, exploding in the past year, right?

    26. DADiogo Almeida51:53

        Yeah.

    27. SPSpeaker51:53

        Like that you can, you can move to like a Cerebras-

    28. DADiogo Almeida51:56

        Yep

    29. SPSpeaker51:56

        ... um, an Etched or whatever and get like the hundred, hundred thousand times speed up.

    30. DADiogo Almeida52:01

        Yeah. Like I, uh, I think that intelligence per second is like a different metric, and we've even talked about like things like intelligence per dollar times second and like metrics like this. My guess on like Jevons paradox occurring, or at least the Jev series of models, and the thing I like hunt people down about internally is like I don't care how much smarter it is, it needs to be in the pre-different tier. So like, like that is what the brand of Jev is. It is the best thing at intelligence per dollar. For intelligence per second, we'll see. I, I think that it's an intriguing thing. I know that there's many industries that are like extremely dependent on real-time stuff, and they will like, like intelligence per second means tons of dollars for them. But we'll see. I, I, I, I'm, I, I would love to like do both and like have the market correct me either which way.
12. 55:33 – 59:57

    Inside Jev’s API and Programming Primitives

    1. SPSpeaker55:33

       choices-

    2. DADiogo Almeida55:33

       Ooh

    3. SPSpeaker55:33

       ... mostly because this is the only podcast that will ask you these, these kinds of questions.

    4. DADiogo Almeida55:36

       Oh, hell yeah. Hell yeah.

    5. SPSpeaker55:37

       Um, so you have three primitives. Um-

    6. DADiogo Almeida55:39

       Yeah

    7. SPSpeaker55:39

       ... choice, score, null. First of all, null, where is that from?

    8. DADiogo Almeida55:43

       Um-

    9. SPSpeaker55:43

       Is that-- is this, like, a term in the, in the literature or what?

    10. DADiogo Almeida55:46

        Uh, now-

    11. SPSpeaker55:47

        I don't know

    12. DADiogo Almeida55:47

        ... now it is.

    13. SPSpeaker55:47

        Yeah.

    14. DADiogo Almeida55:48

        Um, we debated this a lot.

    15. SPSpeaker55:51

        [laughs]

    16. DADiogo Almeida55:52

        We debated this a lot. Um, it is, you know, it is bool-ish, right? Like true/false. Um, it is-

    17. SPSpeaker56:01

        But it's continuous.

    18. DADiogo Almeida56:02

        Yes, exactly. Uh, so, so first, the origin of the name is Bernoulli.

    19. SPSpeaker56:08

        Ah.

    20. DADiogo Almeida56:08

        Yes, so it's-- th-that's why it's even spelled that weird way. That is, like, a subset of the name Bernoulli from, like, a Bernoulli probability-

    21. SPSpeaker56:15

        Right

    22. DADiogo Almeida56:15

        ... which is actually what that is. Um, so that is the origin of it. We were debating this a lot. We liked Pbool, we liked Pool. We, we, w-we were wanting to call it, like, a pool party, but then no one let me. Um, you know, we had, like, a bunch of, like, other arguments about that, and null, we figured, was, like, the best thing. The-- our, our rationale, and, like, this was actually the same thing with Jev too, um, is that we think that we are, like, an irreverent, insane bunch, and programmers don't care. You know, like, if Jev is just gonna be a string, we didn't expect it to catch on or even have puns or anything like that, right? Actually, there was a lot of hate on the name internally. They've all apologized except for one person. Um-

    23. SPSpeaker57:02

        Still holding strong.

    24. DADiogo Almeida57:02

        Yes, uh, our mutual friend. Um-

    25. SPSpeaker57:05

        Okay, okay.

    26. DADiogo Almeida57:05

        Yes, yes, yes, yes. Um-

    27. SPSpeaker57:07

        I respect her for that.

    28. DADiogo Almeida57:08

        Yeah. Uh, yeah. Uh, she wanted Jev to be called Meow.

    29. SPSpeaker57:13

        She would, of course.

    30. DADiogo Almeida57:14

        Yes, of course.
13. 59:57 – 1:19:57

    How to Build with Jev: Structure, Decomposition, and Small Decisions

    1. SPSpeaker59:57

       Yeah. Um, any other-- So, uh, any nuance you wanna go through for-- L-literally, this is for the Jev people who are, like, deciding to really invest in Jev. You are the expert, right? Um, I'm just, like, wanting to provide more background for them on, uh, API choices, um, you know, how they should use some of these, these things like legends, confidence, um, how critical in your testing, um, you know, like how, how... Like just any sort of like pro tips that you, like, want to offer people-

    2. DADiogo Almeida1:00:25

       Hell yeah

    3. SPSpeaker1:00:25

       ... when we're down at this level.

    4. DADiogo Almeida1:00:27

       Thank you. I love this. Um, no, no, no.

    5. SPSpeaker1:00:29

       This is why we're here.

    6. DADiogo Almeida1:00:30

       This is-- Hell yeah. I didn't expect this, and actually, I-- no one has asked me this, um, in probably, like, months when I was, like, onboarding, like, our DevRel.

    7. SPSpeaker1:00:39

       Okay.

    8. DADiogo Almeida1:00:39

       Um, so sick. Um- So our model is designed for being like deep in the insides of computer programs in the future. We like unironically believe that this will be much more massive than anything people are even considering today. And our model might not be ready for that, but we are like continuously working for that future. It will never be good enough at these shallow tasks. Um, uh, uh, sorry, it'll never be g- like, like we're not just gonna cle- keep on climbing the shallow tasks. We want to be deep in the guts of programs because that's how you make software powerful. All the s- all the types inside of our, um... This is an actually an output.

    9. SPSpeaker1:01:16

       Mm-hmm.

    10. DADiogo Almeida1:01:16

        Um, but all the, um, all the parts, uh, of like the, uh, the, the input, like the state, the instructions, the criteria, all of them can be structured JSON objects.

    11. SPSpeaker1:01:27

        Mm-hmm.

    12. DADiogo Almeida1:01:27

        That way, like programs can like insert them in the right spot, and you don't need to like put things-

    13. SPSpeaker1:01:32

        Pieces

    14. DADiogo Almeida1:01:32

        ... into templates. Exactly. So if ever... I, I think people don't read into this part enough, and they think it's all strings, and that's, that, that's fine. Um, but these are all meant... Like, like I would say that if you're using like a template, like turning it into like a system message or something, you are thinking in like the old way. You know, we should be making things as easy for computers to understand because that structure is truly there, right? Like, it would be weird in like, like a programming language to have like all of your numbers in, then you pass it into like... You turn it into a string. Normally, you do that for printing when, when you have a human in the loop, right? But for like within the computer, you want to be passing like nested structure that is semantic all around, and we are really gonna be optimizing our model. It, the model's pretty optimized for this, but the thing is every different nested level of structure is harder to reason about, and we want-- we are really cooking hard in that direction. I think people should keep cooking in that direction because it makes the code like so much more legible and beautiful and like agnostic to like the, the implementation details. It's like, here is my state. You know, like here's my function state. Like think of it, think of it as like an AI function. Which subsets of my state, which is like all the variables you have available, should I pass in here? System messages are like disgusting global variables where you just put everything in there, and you put all those-

    15. SPSpeaker1:02:50

        Slop, yeah

    16. DADiogo Almeida1:02:51

        ... instructions at once. And then y- you know, like you s- you hope that every single instruction gets nailed instead of asking the questions in parallel.

    17. SPSpeaker1:02:58

        Okay.

    18. DADiogo Almeida1:02:59

        And, uh, also, I would recommend, I, and I truly say this not from like a, like it makes me money perspective, um, I truly recommend asking lots and lots of questions. Break them down, make them smaller, and like really decompose. Like no matter if the models can do it today or not, I believe that the biggest like, um, s-saving grace of like what's happening this week will be people's code bases, AI code bases are gonna be so much better. You know, like if you decompose problems into simple decisions, every single one of these things is extremely evaluable. Like, like a AI beforehand is big system message, and then maybe you have like another big AI-

    19. SPSpeaker1:03:38

        Big output, yeah

    20. DADiogo Almeida1:03:39

        ... to see like if it actually does this. That's nuts, you know?

    21. SPSpeaker1:03:42

        [laughs]

    22. DADiogo Almeida1:03:42

        It's kind of crazy. Like, and it, it, it, it's... That was our Stockholm syndrome, right? But like that's kinda crazy. Like if you wanna say like, "Hey, don't read this sub-directory or don't pass any API keys to DeepSeek or whatever else," like that should be programmatically, basically guaranteed. And you know, you'll never have guarantees of any machine learning model, but like by breaking it down, you can actually s- you can actually measure it, right? Like-

    23. SPSpeaker1:04:06

        Yeah. You can verify that it was actually called or-

    24. DADiogo Almeida1:04:08

        Yes, and like our model, our model, like the, the interface itself is so verifiable. This should be like a sigh of relief.

    25. SPSpeaker1:04:14

        Mm.

    26. DADiogo Almeida1:04:15

        You know, like it's, it's, uh, it- it's just gonna lead to way better engineering.

    27. SPSpeaker1:04:19

        Yeah. I think I get that. Um, and, and so, you know, one of the reasons people didn't use to do this in the past is because they would just call a, a small LLM, right?

    28. DADiogo Almeida1:04:28

        Yep.

    29. SPSpeaker1:04:28

        And it's still too slow, it's still too expensive versus chunking everything that... I've done exactly this myself.

    30. DADiogo Almeida1:04:32

        Yep, yep, yep.
14. 1:19:57 – 1:34:50

    The Inverse SaaS-pocalypse and AI Disappearing into Software

    1. SPSpeaker1:19:57

       Mm-hmm.

    2. DADiogo Almeida1:19:57

       It's gonna be fucking awesome. It will not cause mass unemployment, but it will cause, like, a whole bunch of awesome shifts, and wor- the world will be a lot better. And also, like, I'm really tired of AI always being the foreground character, uh, of things. Like, I think that the world should just be more delightful, and AI should just help with that.

    3. SPSpeaker1:20:17

       Hmm.

    4. DADiogo Almeida1:20:17

       You know, and I-

    5. SPSpeaker1:20:18

       Just, like, disappear into the background.

    6. DADiogo Almeida1:20:19

       Exactly. You know-

    7. SPSpeaker1:20:20

       Yeah

    8. DADiogo Almeida1:20:20

       ... like, uh, the, I, I say this in my talks. Like, how can it be that 2019 software, like software, SaaS, whatever, uh, super-duper valuable, right? It's 2026 now. How is the software basically exactly the same despite AI being so freaking awesome, other than sometimes having a chat box on the side, right? Uh, that, that, like, that kind of works but doesn't allow you to make decisions that the companies have stakes in because they, they can't be trusted to make decisions. That to me is nuts. You know, there, there's so much economic incentive for this, and I, I think it's going to be, uh, like a, like an inverse SaaS-pocalypse. I think SaaS is going to be super charged by this. They're the ones who are, like, most in the know of what things are valuable to automate, and it's gonna be, like, a crazy time.

    9. SPSpeaker1:21:06

       Yeah. I, I think, I think so too. It's a s- it's a beautiful thing that you've unlocked, you know?

    10. DADiogo Almeida1:21:10

        Yeah.

    11. SPSpeaker1:21:10

        Um, you mentioned one thing here, which I don't know if it's, uh, like, directly here, which is what is a system one problem and what is not? What, what is a system two problem? Like, you know, um-

    12. DADiogo Almeida1:21:20

        Fuck.

    13. SPSpeaker1:21:20

        [laughs]

    14. DADiogo Almeida1:21:20

        That's a hard one.

    15. SPSpeaker1:21:21

        [laughs]

    16. DADiogo Almeida1:21:22

        That's a hard one, my friend. Um-

    17. SPSpeaker1:21:24

        'Cause people now are just trying to Jev everything, right?

    18. DADiogo Almeida1:21:26

        [laughs]

    19. SPSpeaker1:21:26

        Which, like, probably is gonna fail.

    20. DADiogo Almeida1:21:28

        Love Jev everything.

    21. SPSpeaker1:21:28

        Right?

    22. DADiogo Almeida1:21:28

        [laughs]

    23. SPSpeaker1:21:29

        Um, but, like, some things are gonna be good.

    24. DADiogo Almeida1:21:31

        Jev everything is pretty funny.

    25. SPSpeaker1:21:32

        Yeah.

    26. DADiogo Almeida1:21:33

        Uh, it's a pretty funny way of doing it, saying it. Um, the-

    27. SPSpeaker1:21:36

        Yeah

    28. DADiogo Almeida1:21:36

        ... so I'll tell you the truth.

    29. SPSpeaker1:21:37

        Yeah.

    30. DADiogo Almeida1:21:38

        Um, the truth is that this is an empirical problem, just like scaling laws are, are an empirical thing. You know, like, why doesn't, like, robotics really work right now despite all the money being spent on it?
15. 1:34:50 – 1:40:17

    Computer Use, Dark Data, and Jev’s Biggest Use Cases

    1. SPSpeaker1:34:50

       stuff.

    2. DADiogo Almeida1:34:50

       Whoa.

    3. SPSpeaker1:34:50

       Is this, is this what you were referencing?

    4. DADiogo Almeida1:34:51

       I've never seen... I haven't... I- I've seen... [inhales] I saw, like, a airline browser use thing.

    5. SPSpeaker1:34:58

       And, um, inside this new note, let's make the title say hello.

    6. DADiogo Almeida1:35:02

       Wow.

    7. SPSpeaker1:35:02

       Great. Great. Okay. Um, let's move on, and can you open the Arc Browser? And once you're there, can you Google search Norbert Wiener? Um, now can you open up x.com? Is this this kind of use case?

    8. DADiogo Almeida1:35:17

       Oh, the voice use cases. This is actually the first one I've seen. This is-

    9. SPSpeaker1:35:20

       Oh, okay.

    10. DADiogo Almeida1:35:21

        Wow.

    11. SPSpeaker1:35:21

        [laughs]

    12. DADiogo Almeida1:35:23

        Oh, wait, wait, wait, wait. Oh, like, can you go, can you go back a second? Can you go back a second?

    13. SPSpeaker1:35:26

        Um...

    14. DADiogo Almeida1:35:27

        Rumors claim Anthropic engineers wor- worship Claude as God. Wow.

    15. SPSpeaker1:35:30

        [laughs]

    16. DADiogo Almeida1:35:31

        Wow. Dang. That's pretty funny. Um...

    17. SPSpeaker1:35:35

        And here you are building prod.

    18. DADiogo Almeida1:35:36

        [laughs] Abso... Wow, this is sick.

    19. SPSpeaker1:35:41

        Uh, yeah, so, so c- clearly you can operate the whole computer with voice, with Jev as a decision model.

    20. DADiogo Almeida1:35:46

        So just like I'm anti-benchmarking, I'm also anti demos. I want to make sure that it works reliably. I love people are playing with it. This is super fucking sick, have no doubt. I want to see this. I wanna see it be used. I want our team to play with it. I wanna find the weaknesses, and I wanna solve that.

    21. SPSpeaker1:36:04

        Yeah.

    22. DADiogo Almeida1:36:04

        And I would lo... Man, that looked really cool.

    23. SPSpeaker1:36:06

        [laughs]

    24. DADiogo Almeida1:36:06

        That looked really cool. I want that.

    25. SPSpeaker1:36:08

        Uh-

    26. DADiogo Almeida1:36:08

        I want that. Like, when my, when my wrists are sore, I, like, just whp- whisper flow everything. That'd be sick.

    27. SPSpeaker1:36:13

        Well, as, uh, well, you know, just, just to round out the use cases side-

    28. DADiogo Almeida1:36:16

        Yeah

    29. SPSpeaker1:36:16

        ... 'cause I, I do have to let you go.

    30. DADiogo Almeida1:36:17

        [laughs]
16. 1:40:17 – 1:42:29

    How Jev Could Reshape Coding Agents

    1. DADiogo Almeida1:40:17

       Um, Claude Code and Codex are, I believe, the, the winner, like the, the number one and two. I'm not entirely sure. I don't follow closely.

    2. SPSpeaker1:40:25

       Roughly.

    3. DADiogo Almeida1:40:25

       But like it's roughly that. But they're built around a single model world. You know, like, and, and that makes a lot of sense for them, right? Because, like, it has been a one-model game where it's like kind of like the same model but different intelligence that you're shopping.

    4. SPSpeaker1:40:38

       Mm-hmm.

    5. DADiogo Almeida1:40:38

       But all the open coding agents are like fucking jazzed right now because they're like getting their Jev on and-

    6. SPSpeaker1:40:44

       [laughs]

    7. DADiogo Almeida1:40:44

       ... like the thing is there's-- I'm sure they're trying a lot of weird stuff.

    8. SPSpeaker1:40:47

       Mm-hmm.

    9. DADiogo Almeida1:40:48

       But all the coding agents are kind of roughly at like approximate parity, right? Because, like, there's not so much you can do with a while loop. But the moment one person finds one killer use case that, you know, you can only do with that coding agent, everyone will flock to it because they have like a monopoly on that thing. But all the open coding agents will be able to copy that, right? The end. But I don't know what the Claude Codes and Codexes will do because they are built around that one model world.

    10. SPSpeaker1:41:16

        Single model, yeah.

    11. DADiogo Almeida1:41:17

        And like I think that's gonna be like a really interesting thing. You know, like I would love to be able to integrate with them personally. Like, I want to integrate with everyone. Like, uh, they, they might make competitors eventually, I don't know. But like it is not me, my job as Sonfire Infrastructure to be opinionated on that, right? Like, I want to just serve the world. Um, but I, I don't know if they would do that and like I think it'll make the coding agent game super weird. You know? Like, I'm so excited for that. Uh, and like I'm sure, uh, I'm getting my team to review right now an internal document I made on design patterns I suspect will be useful for coding agents, so hopefully I can share it like right after I walk home. Um, uh, but like I think that there's just like such ripe area for exploration out in the world and like it's, it's-- Man, if I did not have this, I would love to experiment with coding agents right now.

    12. SPSpeaker1:42:10

        Yeah. I mean, and, and I'm sure the coding agent companies would love to work with you as well to, to figure that out. Um, yeah, I, I do think that there's still use cases for Claude Code and Codex with you guys-

    13. DADiogo Almeida1:42:18

        Of course

    14. SPSpeaker1:42:18

        ... um, which, um, uh, it's, it's easy to explore there. Um, okay. I mean, you know, we've, you've, you've been very, um, uh, obliging and sort of indulging in all these, all these things. I just wanna take you out of TypeSafe-

17. 1:42:29 – 1:49:32

    AI Safety, Frontier Pacing, and the Limits of RLVR

    1. DADiogo Almeida1:42:29

       Hell yeah

    2. SPSpeaker1:42:29

       ... just generally about... And you've, you've made very clear your position on, on the state of AI. Uh, give you m-more room on the alignment safety side of things.

    3. DADiogo Almeida1:42:37

       Oh, uh, did I not-

    4. SPSpeaker1:42:38

       Oh

    5. DADiogo Almeida1:42:38

       ... talk about safety alignment at all?

    6. SPSpeaker1:42:40

       Oh, you did. You did.

    7. DADiogo Almeida1:42:41

       Oh, maybe I think I didn't. I think maybe I didn't.

    8. SPSpeaker1:42:42

       You did. Uh, I, I-

    9. DADiogo Almeida1:42:43

       Oh

    10. SPSpeaker1:42:43

        ... just like, you know, I, I think that there's, there's a lot of, uh, you have a lot of researcher discussions. We have this every-

    11. DADiogo Almeida1:42:49

        Of course

    12. SPSpeaker1:42:50

        ... every NeurIPS.

    13. DADiogo Almeida1:42:51

        Yeah.

    14. SPSpeaker1:42:51

        What are people talking about? You know? Like, uh, so for example, um, I, uh, recently was at, uh, one of these researcher gatherings, and people are genuinely worried about the pacing, right? Like this, this whole topic about like we should slow down because-

    15. DADiogo Almeida1:43:07

        Mm-hmm

    16. SPSpeaker1:43:07

        ... uh, the public is like clearly not ready. Um, and, uh, I'm sure you have strong feelings.

    17. DADiogo Almeida1:43:13

        Um, I feel like this is the kind of thing that is a dangerous topic-

    18. SPSpeaker1:43:20

        Okay

    19. DADiogo Almeida1:43:20

        ... to talk about. Um, I'm happy to talk about it. I, I live for danger. [both laughing]

    20. SPSpeaker1:43:25

        All right.

    21. DADiogo Almeida1:43:25

        Our company brand is chaos. It's not Jev. It is irreverence and chaos.

    22. SPSpeaker1:43:29

        And you know, yeah. And, and like you were at OpenAI during like the one of the very first like- Very visible incident, just the blip, right?

    23. DADiogo Almeida1:43:37

        Oh, the- the-

    24. SPSpeaker1:43:38

        Which, like, and, and, like, you, the dominoes have gone down now-

    25. DADiogo Almeida1:43:41

        Mm-hmm

    26. SPSpeaker1:43:41

        ... to now every frontier lab has co-signed a document saying that they want to pace

    27. DADiogo Almeida1:43:45

        Um, interesting. I-- So comp- it's a very complicated, nuanced thing. I actually do want to write a response to this more formally. I do have, like, a little bit of a short version of my response-

    28. SPSpeaker1:44:00

        Yeah

    29. DADiogo Almeida1:44:00

        ... which is that, um, as you RLVR more, like RLVR is like, the, so RLVR is not actually about verifiable rewards. Like, that has been failing since before the reasoning revolution. Like, like, like, and that's the weird part about tasks, right? Like back when-- Oh, fun history. Back when RLHF was becoming a thring, thing, there were three different things that, like, are now called post-training, different efforts, and instruction following was by far the, like, the, the bastard child. Like, people didn't like it. They didn't want to take it into account. It was annoying. You know, like, I talked to the pre-training team, and I'm like, "Guys, this is the magic." And they're like, "We'd run so many model sweeps. You want us to wait for human evals to figure out which models to use?"

    30. SPSpeaker1:44:45

        [chuckles]
18. 1:49:32 – 1:56:48

    Why Diogo Wouldn’t Pre-Train with $1 Billion

    1. SPSpeaker1:49:32

       [laughs]

    2. DADiogo Almeida1:49:32

       I mean, like, it's, it's, it's like, you know, it's a cost-saving thing.

    3. SPSpeaker1:49:35

       Yeah.

    4. DADiogo Almeida1:49:35

       You know, instead of, like, having to pre-train again. Like, there's intriguing stuff. I actually think that Like, intelligence has a je ne sais quoi at every single level, and it's always super-duper fascinating. Like, I am a shape rotator, so I don't like finding that, but I love it when people find it and teach me about it. Um, but in it, you know, looking at the data, this thing that, uh, our data team is so good at that I'm not, um-

    5. SPSpeaker1:50:00

       Hmm

    6. DADiogo Almeida1:50:00

       ... I, it's-- I find it really, really fascinating. I love actually thinking about, like, how capabilities are, like, put into the model, like, over, like, the short term. You know, like, there's, like, the, the, the, the really rapid alignment of fine-tuning, and over the long term, after seeing it over and over and over again. Like, this stuff gets baked deeper and deeper and deeper and deeper into the model until it gets robust. You know, and that is, like, the, the North Star to surface, and, like, the system one stuff is the stuff that ends up getting robust. So I find mid-training to be, like, a fascinating thing. Um, I'm a fan of all forms of training. Uh, I'm a fan of all forms of, like, surfacing new types of intelligence. I wouldn't do it all myself because it's expensive. Um, and I have said privately and also... Should I say this?

    7. SPSpeaker1:50:50

       [laughs]

    8. DADiogo Almeida1:50:51

       Huh. Huh. You know, like, my philosophy is anything I sh- I, I should say in, like, private with, like, an investor, I should say in public with the people because that is, like-

    9. SPSpeaker1:51:01

       Power to the people

    10. DADiogo Almeida1:51:01

        ... my thing.

    11. SPSpeaker1:51:02

        Yeah.

    12. DADiogo Almeida1:51:02

        Yes. So m- the thing I've said, uh, before is if you gave me a billion dollars, I wouldn't pre-train. Um, I still believe that to be true. It is a very expensive thing when if you are, like, like, if you're an AI engineer, you can, like, slice and dice and do all sorts of stuff. You know, like, Frankensteining is not the most elegant, beautiful thing, but it solves problems, baby.

    13. SPSpeaker1:51:26

        Hmm.

    14. DADiogo Almeida1:51:26

        So, um, anything except pre-training.

    15. SPSpeaker1:51:29

        Hmm. Hmm. Yeah. Uh, amazing. Uh, yeah, I, I think one, one direction that I do think that is interesting, uh, just, like, synthesizing all your, all your commentary about these model things is, like, do we have a, a, a, a, a super model, uh, that has all these capabilities involved, or do we break them out, uh, in, uh, fur-further? I guess sort of like, um, uh, one way to put this is that OpenAI was trending the direction of the omni model.

    16. DADiogo Almeida1:51:53

        Mm-hmm.

    17. SPSpeaker1:51:53

        Right? Um, 4.0 was, was one of those. Uh, then for, for, for a f- brief period of time, there was always, like, there was, like, a kind of a main branch of the, this is the chat tune model, and this is the coding tune model.

    18. DADiogo Almeida1:52:04

        Those are t- completely different things. Those are extremely different concepts. I would like break that down a little bit. So multi-modality is a little bit different-

    19. SPSpeaker1:52:12

        Hmm

    20. DADiogo Almeida1:52:12

        ... um, because sometimes the other modalities help, sometimes they hurt.

    21. SPSpeaker1:52:16

        Yes.

    22. DADiogo Almeida1:52:17

        Like, you know, people are moving-- They seem to be moving away from speech, which is different than audio, um, because it seems to not generalize well to the other stuff.

    23. SPSpeaker1:52:26

        Hmm.

    24. DADiogo Almeida1:52:26

        This might get solved. I'm a fan of all of this. But these are, like, empirical real questions. Like, scaling laws are not about just throw money at it and it gets good. Scaling laws are pragmatically how good is a thing, you know? Like, uh, like, there are worlds where, like, no matter what you scale, it may not be good enough. So, um, y-y, you know, like, computer use is not currently solved is my understanding. Like, I'm hoping that we can be a, like, play a part in solving that. But, like, it-- there might be no amount of data we collect that will solve that. We might need better methods or something else like that. So, um, like, we-- you need to be, like, really practical in all of this. Am I a fan of omni models? I'm a fan of all forms of intelligence. But I will go straight into one thing you talked about, which is different from pre-training, which is post-training-

    25. SPSpeaker1:53:15

        Hmm

    26. DADiogo Almeida1:53:15

        ... because I hate fracturing intelligence. That is, like, the bad thing to me, and this whole, like, chat first reasoning mode is because, um, it forces the intelligence to be fractured. Like, when you're optimizing for chat, this tends to be, like, pure RLHF, and it's quite intrinsic in RLHF to do the stuff people, like, naturally complain about, right? Like, oh, I'm gonna-

    27. SPSpeaker1:53:36

        You're absolutely right. And-

    28. DADiogo Almeida1:53:37

        Yeah

    29. SPSpeaker1:53:37

        ... you know.

    30. DADiogo Almeida1:53:38

        Psychophancy? Psychophancy? I don't-
19. 1:56:48 – 2:03:10

    The OpenAI Story Behind TypeSafe

    1. DADiogo Almeida1:56:48

       Hell yeah

    2. SPSpeaker1:56:48

       ... because I, I do want to, uh, g-get you out. Uh, one is like, just reflecting on your two-year journey. It's roughly two years? Two point something?

    3. DADiogo Almeida1:56:54

       Um, with the company-

    4. SPSpeaker1:56:55

       Yeah

    5. DADiogo Almeida1:56:55

       ... I think that this is, like, more like a four-year journey.

    6. SPSpeaker1:56:58

       Yeah.

    7. DADiogo Almeida1:56:58

       But-

    8. SPSpeaker1:56:58

       Well, yeah. Actually, like, uh, I was thinking, remembering that, like, you had this, like, hero run around Thanksgiving. You were like, you were canceling everything because, uh, you were like, "Guys," like, uh, "everyone's on holiday. I'm gonna take all the OpenAI GPUs and g-go do this thing."

    9. DADiogo Almeida1:57:11

       Yeah.

    10. SPSpeaker1:57:11

        [laughs]

    11. DADiogo Almeida1:57:12

        That was a good time.

    12. SPSpeaker1:57:13

        And that was, like, the, the pre-TypeSafe-

    13. DADiogo Almeida1:57:15

        Yeah, yeah, yeah

    14. SPSpeaker1:57:15

        ... moment, right?

    15. DADiogo Almeida1:57:16

        I-- That might've been-- Was that when the coup was happening? I don't really know.

    16. SPSpeaker1:57:19

        Yes, actually. [laughs]

    17. DADiogo Almeida1:57:20

        Yeah, yeah, yeah. That sounds right. Yeah.

    18. SPSpeaker1:57:22

        [laughs]

    19. DADiogo Almeida1:57:23

        I remember. Oh, my God. I don't wanna... I, I'm not-- I, I don't think I have the time to spill the tea about the coup right now, but, um, that was really annoying. Um-

    20. SPSpeaker1:57:33

        The, the coup was annoying or the, the run was annoying?

    21. DADiogo Almeida1:57:35

        The, the coup was annoying.

    22. SPSpeaker1:57:36

        The coup. Okay.

    23. DADiogo Almeida1:57:36

        Yeah, yeah, yeah.

    24. SPSpeaker1:57:37

        Yeah.

    25. DADiogo Almeida1:57:37

        Uh, it... I will-

    26. SPSpeaker1:57:39

        Safetyists took over the company. Yeah. Anyway.

    27. DADiogo Almeida1:57:43

        Maybe next time we chat-

    28. SPSpeaker1:57:44

        Okay. All right, all right

    29. DADiogo Almeida1:57:45

        ... I'll dump tea about-

    30. SPSpeaker1:57:46

        Tea
20. 2:03:10 – 2:09:29

    Why Diogo Thinks Most Neo-Labs Are Getting AI Wrong

    1. DADiogo Almeida2:03:10

       Oh, yeah.

    2. SPSpeaker2:03:10

       And another question is, someone like you-

    3. DADiogo Almeida2:03:12

       Yeah

    4. SPSpeaker2:03:12

       ... is in the frontier lab right now who is frustrated not getting the, the funding or the, the resources, whatever, uh, the attention. Uh, what's your advice to them? You know, do, should they do what you did?

    5. DADiogo Almeida2:03:23

       Should they do it? Ooh, that's a fascinating question. Um, man, how do I do this without burning bridges?

    6. SPSpeaker2:03:35

       [laughs]

    7. DADiogo Almeida2:03:35

       Um, I-- my sense is that most n- unless there's some level of economics I don't really understand, um, I think most neo-labs are crap. Um, I, I don't want to see myself with that as peers. Um, like I, I don't really understand what's going on there. Like, is it becau- like number one, I don't really value researchers. Um, I value people who, like, like look at my bitterest lesson, right?

    8. SPSpeaker2:04:02

       The data, the task.

    9. DADiogo Almeida2:04:02

       I, I, I want-- well, not just that.

    10. SPSpeaker2:04:04

        Yeah.

    11. DADiogo Almeida2:04:04

        I, like, we need researchers, but we need them to give a lot of fucks about the right task, and that's the important thing, right? So it's actually, like the, it's, it's kind of backwards when people value pure research pedigree 'cause that generally doesn't create value. Um, so it, li-like number one, I, the, I believe in North Star tasks and doing cool, really useful stuff. Um, number two, um, because I don't value researchers, I don't, I don't recommend going the... Well, it clearly is profitable for someone, or it might be in this environment. So, like from a purely pragmatic perspective, I don't see creating neo-labs as want, something that creates value. It seems to destroy value because, like, they are, like, redoing work from scratch with, like, low probability of actually moving the frontier, and as far as I've talked to most neo-labs, they don't really have a direction. They tend to want money to play around with their experiments. Um, if they have direction, I'm super in favor of it, to be clear. So my advice for someone is it really depends on why you're doing it. You know? If you are a researcher who wants to play a-around with research, um, probably the labs are the best place to do that, TBH. Like, there might be other places. I, I, I don't really keep track of that politics. But I would just recommend not being that way personally. You know, like, I think it's better for the world with people being driven to solve real problems. Uh, and those problems may be exploratory, that's fine, but, like, ideally have principles that you stand behind. But if you think that you wanna do the right task, like abso-fucking-lutely, like please do. Like, please break this, like, unimind, you know, uni-unimodal, like-

    12. SPSpeaker2:05:50

        Hive mind.

    13. DADiogo Almeida2:05:50

        Yeah, exactly.

    14. SPSpeaker2:05:51

        Yeah.

    15. DADiogo Almeida2:05:51

        Like, like ev- you know, like again, this pacing the frontier is coming from, like, this one view of AI that looks like, you know, AI super genius that is incredibly jagged. And that is, um-

    16. SPSpeaker2:06:05

        Solvable

    17. DADiogo Almeida2:06:06

        ... you know, it's solvable, and it's weird, and it's, like, not matching reality. And, um, it, it's like, it's tragic, right? Like, like I, I, I think, like, all of these, uh, like the, like really unearthing technology I think is, like, just good.

    18. SPSpeaker2:06:20

        Yeah. For what it's worth, uh, you know, a-again, I, I'm trying to rep-accurately represent the position of the, uh, anthropic OpenAI folks I was talking to, SpaceX as well, by the way-

    19. DADiogo Almeida2:06:28

        Okay

    20. SPSpeaker2:06:28

        ... uh, is that, uh, it is, this is a political thing much more so than a pure X risk thing.

    21. DADiogo Almeida2:06:34

        Yep.

    22. SPSpeaker2:06:35

        Uh, so yeah.

    23. DADiogo Almeida2:06:36

        Cool, cool.

    24. SPSpeaker2:06:36

        Political positioning is-

    25. DADiogo Almeida2:06:37

        And, and that, and that's beyond my pay grade.

    26. SPSpeaker2:06:39

        Exactly. Yeah.

    27. DADiogo Almeida2:06:39

        That's well beyond my pay grade.

    28. SPSpeaker2:06:40

        Once they sh- once they told me that, I was like, "I get it. Uh, this is about the twenty twenty-eight, uh, election." [laughs]

    29. DADiogo Almeida2:06:47

        Oh, no.

    30. SPSpeaker2:06:48

        Yeah. [laughs]
21. 2:09:29 – 2:21:59

    Coding Agents Beyond the KV Cache and the Multi-Agent Future

    1. SPSpeaker2:09:29

       and cheap.

    2. DADiogo Almeida2:09:29

       Yeah.

    3. SPSpeaker2:09:29

       What is a second or third one that you wanna throw as a bone to someone else that you're not, that you want someone else to work on that you're not gonna work on?

    4. DADiogo Almeida2:09:35

       Ooh.

    5. SPSpeaker2:09:36

       Like, uh, just basically give people tasks.

    6. DADiogo Almeida2:09:39

       Give people tasks?

    7. SPSpeaker2:09:40

       Yeah, like, that your tasks, you-

    8. DADiogo Almeida2:09:41

       Oh, there's so many I want. Oh, what?

    9. SPSpeaker2:09:42

       You have picked your tasks, right? You know what I mean?

    10. DADiogo Almeida2:09:44

        What? Wait, I-- That's such a good question. Holy crap. Oh man, I'm so excited by that.

    11. SPSpeaker2:09:48

        'Cause you're, you're gonna be, you'll be f- so the next, like, 50 years, you're gonna be busy doing your thing.

    12. DADiogo Almeida2:09:52

        Hell yeah. Okay.

    13. SPSpeaker2:09:52

        [laughs]

    14. DADiogo Almeida2:09:53

        So let me give, like, a fun one and a not fun, uh, l- l- and, like, maybe a valuable one that's also fun. Um, my, my fun one is I think games could be so freaking cool if they were intelligent. Like, when I see people play around with, like, like Ali's Doom demo where, where, like, you can, like, get NPCs to control stuff, like, you know, like, like that was just really, like the-

    15. SPSpeaker2:10:17

        Yeah

    16. DADiogo Almeida2:10:17

        ... like a proof of concept. I think some really cool stuff could be made. It looks really, really cool. You know, like, um, like, I, I'm a big Stardew Valley fan, you know, and, like, it's, it's really static and it's still compelling. Like, I feel like there's a lot of cool story that could happen. You don't need to call, like, Jev in the game loop. It's probably too expensive for that. But even, like, simple, like, state machines for NPCs, I think it could make, like, such a compelling world. Oh, man.

    17. SPSpeaker2:10:42

        Mm.

    18. DADiogo Almeida2:10:42

        Um, and man, a little sad that I can't work on these types of things.

    19. SPSpeaker2:10:46

        Yeah.

    20. DADiogo Almeida2:10:47

        My, my life path is a little bit set right now, and I'm, um, uh-

    21. SPSpeaker2:10:51

        Yeah, but you can call someone else to work on it.

    22. DADiogo Almeida2:10:52

        Yeah, yeah. That's cool.

    23. SPSpeaker2:10:52

        And then, and then you can, like-

    24. DADiogo Almeida2:10:53

        Yeah

    25. SPSpeaker2:10:53

        ... feedback on it.

    26. DADiogo Almeida2:10:54

        And, um, the thing that I would really, really like to explore is, like, coding agents free from the tyranny of the KV cache. Like, like, it might not be as good as true coding agents are, but I think there's just so many weird things to think about. Th- that, that's why I wrote the article, "KV Cache Rules Everything Around Me."

    27. SPSpeaker2:11:14

        Mm.

    28. DADiogo Almeida2:11:15

        Um, uh, believe it or not, I don't think anyone has used the phrase on the internet, "Cache rules everything around me," C-A-C-H-E, um, when I, when I Googled it. Um, so-

    29. SPSpeaker2:11:26

        Okay

    30. DADiogo Almeida2:11:27

        ... uh, like, I wrote this 'cause I wanted to tell people about, like, this is how coding agents w- agents work and how the KV cache works-

Episode duration: 2:22:21