Jev: System One models for Prod, not God — with Diogo Almeida, CEO, TypeSafe AI

[gentle 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? Emotionally? Yeah. Um, never been worse. [laughs] Um, [laughs] like, uh, I'm a ragged corpse of a person right now- Yeah...

because there's so much going on, and I'm like a technical CEO, so I have, like, a lot of fires to fight. Yeah. 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. Mm. 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. [laughs] 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- Yeah... I'm so jammed about the community and everything. It's so great. Yeah, you were saying yesterday that you decided to prioritize the dom-- 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. Yeah. It felt a little like, "Oh man, I'm talking to, like, really important people right now." Yeah. I probably shouldn't reveal who. Yeah. But it feels a little bit dirty for me to, uh... I, 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. Yeah. 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, "No, that's too crazy." Sure. Yeah. Yeah. Uh, [laughs] well, you, you, you guys have been hosting town halls on the Discord. Discord is now at a hundred thousand people. Um, your, your Twitter is kind of- I don't follow these stats. Yeah. So holy shit.

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 like- [laughs] I'm a noob. I'm a noob You're such a noob. I'm a noob. But no, but that, like, that's, like, positive aura- Cool... that, like, you don't know how to pro- [laughs] promote yourself. Yeah.

Well, someone, like, called me out when I posted, like, "Holy shit, we're all three trending, trending topics." And then they're like, "That's a personal feed." That's a personal- And I'm like, "Oh, no" Of course, of course it would trend to you. Cringe. Yes. Yeah, yeah, yeah. 'Cause it's what you clicked on. Um, so okay. Uh, let's, uh, yeah, oh, so congrats on everything. Thank you.

We'll talk about more, uh, details a-as you have them, but let's, for people who are, like, living it under a rock or just- [laughs]... just want, like- Mm-hmm... the definitive thing, what is Jev? Oof. Let me think about-- That's a hard one. [laughs] Um- Okay, and, and I'm happy to, like, uh, re-ask if you wanna kinda address this one. No, no, no. I'm happy to- Okay... I'm happy to, like, just jam on it. Yeah.

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. [laughs] Nice. Um, so, uh, the way I see it is we neew- 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. Yeah.

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-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- You should've said low-key research preview. [laughs] Um, it kind of was, right? It kind of was.

But, um, we, so there's a class of models that we describe them as, like, machine native, System One, large programmable. Mm-hmm. 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, lo- 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 re- 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 type safe. 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, uh, System One model, whatever you want to call it. Um, Jev is meant to be optimized for intelligence per dollar, uh, hence the name of Jev. You know, uh, I mean- Jevons paradox... 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. 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- Mm-hmm... uh, in the past, uh, with Clementine Forget of Hugging Face, uh, where they were like, yeah, a-actually, um, y- they're just... Or, you know, and this is your, your argument about RLHF, is they're model collapsing towards what you want to hear the most- Ooh...

or what is most likely, uh, instead of, like, their own internal confidence about a, a, a thing. Can I soapbox on that for a second? Go ahead. Yeah. Or- Yeah, yeah... cool. Uh, like, um, uh, I've been heard that your audience is the most technical, so I actually want to get into that. Yeah. 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- Mode dropping or mode collapse? Uh, it's the same thing. Is that what you- It's the same thing. All right. 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 Jan LeCun a lot.

Uh, I think Jan LeCun's takes are actually among the closest to- What, what about this? [laughs] Well, do you want to address this now, or should I let you go on mode collapse? No, no, no, later. Okay, okay. Go, go, go more, like, on mode collapse. I wanna- So I actually think that among takes, Jan LeCun's is among the most accurate. Um, but he has this very famous/infamous slide about, uh- The cake?...

the la- LLMs are doomed. Okay. You know, like that one where he's, it, where he, like, has, like, a pie chart with, like, a s- tiny part, tiny little thing- Pull it up... and says that as you increase sequence length, um, the probability of it making an error goes in. 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 y-you- What's the disconnect? Exactly. And, um, may I or you want to tell me? Uh- About mode collapse? Oh, no, no. W- uh, so mode coll-collapse is related to this. Yeah.

The disconnect happens because if you are in a mode covering or a calibrated distribution, um, you are, like, not, you are not overly punished about having outliers. So you'd expect, like, you know, something, some amount of the time you'd be out of distribution, some amount of time you'd be in distribution. That's what happens when you cover the distribution.

This was, like, models before GANs. They made blurry images, right? Mm-hmm. Um, instead, GANs mode drop. They, like, drop the minority classes and just do the really common ones, and this is why this effect doesn't happen, right? Like, instead of b- in order to generate really long strings, um, without making errors, they need to, like, be extremely conservative because it- really easy to see when an error happens.

It's very hard to see when, like, a subtle thing that looks correct happens, and that calibration is, like, total poison into, like, the probability distributions of strings. Yeah. And, uh, the, it's a, it's a nuanced take and, like, I, I think that th-this is why this doesn't happen, and this is why strings are so bad at, uh, decision-making or, uh, you know, overloading the string models are for decision-making is, is, like, a, a bad time.

And while we're on the topic of Jan, do you agree that his fix i-with which is, like, a world model, like a JePA-type, um, embedding thing is the right solve? So bas-basically, like, the, one of the reasons that it could fail is because you're trying to reason over token outputs and then, and then just looping back again and, uh, going, keep, uh, continuing going until you reach, like, a end of sentence.

Um, like, is that, uh... And his solve is JePA, right? Yes. Which is, like, joint ambition, uh- Yep... uh, joint embedding prediction. Uh, so, like, is that the solve or, you know, like, uh, do you have a, do you have a take on, on that? Oh, man, I probably shouldn't talk- [laughs]... too much about the insides of ML.

But I will say that my, my brand, other than unhinged, is practical. Mm-hmm. You know, like, even my take here is practical, and, like, I'm, am I a scaling law fan? Um, depends. It depe- you know, it's, it's, it's, like, it's a, scaling laws tell you how much better you get at a thing for amount in. Mm-hmm.

A scaling law does mean exponentially more resources for normally sublinear gains, which looks to be a bad investment unless those, like, linear gains are, like, really, really valuable. Um, but it's all, to me, it's all about, like, what can we do with what we have to make the biggest possible fucking difference? I can curse. Yeah. Yeah. Yeah, yeah, yeah. Yeah. We're, we're at, we're a poof for adults.

Hell, hell yeah. Uh, and also we have a scaling law thing if you wanna go into that later. [laughs] Oh, I, I could if we, we... It's, see, that part is not super relevant- Yeah... right now. I actually, if you wanna go into my Bitters lesson, I think that that's more relevant. Okay. But, like, to me, I'm all about, like, pragmatics, and I think that the JePA stuff is really cool early research.

I really love awesome research. Is it practical yet? Um, probably shouldn't say. Um, but, uh, like, there's just a lot of... I just think there's just, 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?

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

That, that- [laughs] Uh, actually, c- if I... I'll, I'll make a tangent if that's okay. Yeah, yeah. I think you might enjoy this. Really? Might d- Five, five tangents in. No, this is good. This is fine, yeah. Oh, yeah, I get lost in all my tangents. This, this is gonna be horrible for the listeners to figure it out, but they, they're gonna figure it out. It's fine. Yeah, we could edit it if it helps. This morning, yeah.

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 bottle 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- Mm... right? But you can work with it, right? And you're forced to work with it 'cause of Stockholm Syndrome. Um- [laughs]... uh, d- 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? Mm. Like, that is, um, like straight up insanity.

It i- it's, it's coming from a place of, like, people who do not understand software, do not understand programming- Mm... 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. Fair enough. Uh- Yeah... you, you want something that is the core kernel that is, is usable everywhere. Yes. E- exactly. Like the cognitive core, right? Yeah. And you need this thing to be s- like so general, so optimized for its use cases.

You want it to be like, uh, you know, you want it to work on all the future use cases, all the weird shit that people are doing. Yeah. You know, we obviously didn't train on any of that stuff. Is it surprising that it works? No, 'cause we trained on weirder stuff, my friend. [laughs] Um, so but one tangent up about like safety alignment. Okay.

Safety alignment makes sense for a product, in my opinion, for like ChatGPT and Claude. Like it, um, what, what safety... what makes safety and capability alignment different is capability alignment is like about doing what the user wants. That is sick for software engineers. They want their thing to do the thing, and the more predictable it is, the less they have to test it and play around with it. Jeb is not anywhere close to that yet.

It could be, but like there's so many more nines of reliability that we want in order to make it so good, like a database query, that you don't even have to think about it. It is just there when you need intelligence. Um, but safety alignment is like the opposite of instruction following. It's when you want to follow someone else's instructions, like OpenAI Anthropic's- The lab's values- Exactly... stack. Yeah. Exactly.

And this makes a lot of sense for our product. Again, like ChatGPT should do... Y- you, you sh- like if they don't want to like, uh, do like some, uh, not safe for work role play with ChatGPT, that's on them because like maybe that's, you know, f- what their users who have like parents and kids want.

L- like that's fine. But in an API, that's nuts, right? Um, l- like that's completely unacceptable because like people need to like program around this and that is, that's so anti-user that it's... It, it, it-- I'm... Huh. I, I, I, I, I can be an angry person, so, uh, I should try to calm down. It, it's, uh, people get your passion, and I think it's really good.

Uh, the, the one pushback I'll give you is like what if we use it to kill people, right? Like that, that is the actual like... N- the not safe for work thing, it's private, personal, whatever. But like, yes, like we will use it in war, and like that, that is, uh, something that companies can reasonably prefer their APIs not be used for. [laughs] I get that.

I, I, I think that there's like pragmatic places where that opinion can be held. I don't think the foundation of like a general purpose technology is that place, uh, personally. Mm. Like would I prefer that our stuff is not used to kill people? Obviously. Would I prefer it's used for like all sorts of like great stuff in the world? Obviously. Will I put my thumb in the scale for that? Yes.

Would, will I do it at the technological layer? Absolutely not, because that will fracture the intelligence. Every single time you mean it to over fit to some weird stuff, you're fracturing its intelligence more and more, and like these things are fractured to the like... They're, they're so dar- darn fractured right now. Yeah.

So, uh, uh, and, and as a furthermore thing, um, to me it's like I think intelligence will be more like a database than a coworker. Like I don't think it's up to databases to add checks on whether or not they're used for like, uh, what's something that's not great? You know, like CIA... Actually, I don't know what the CIA does really.

You, you can imagine. [laughs] You can imagine. Uh, k- killing people who are not even bad or whatever. Mm-hmm. And like I don't think it's the database's responsibility for that, and furthermore, like a thing that has been weird to me is when people like sign up for our thing on Slack and they're like, "Hey, we're gonna deploy this.

Can we deploy this thing?" I am just like, "My brother, we are an API. You are a developer. It's none of my business," right? Like, uh, like you shouldn't know what the whole task even is- Yeah... because it should be decomposed into small things. We shouldn't be able to know what the downstream users are doing, and that is like a good boundary to give software engineers maximum power.

Ideally, they use it for the good stuff, and ideally, we can like help them, and like we've talked about like doing open source and charity and all of that. We have absolutely no time for anything else right now. But like they will get any of that bias out of the technological layer as long as I'm in charge. Yeah. That's great.

Uh, while we're on the topic, let's also briefly talk about your privacy stuff, terms of ser- terms of use, which, uh, got a little bit of- Ooh... uh, misunderstanding. Uh, I just wanna clarify that upfront. Hell, yeah. I think it probably takes two sentences from you about like you will not, uh, y- you're not being that restrictive about your API. Like clearly- Oh, yeah. Obviously, yeah...

ideologically, you would take a, your role as a platform very seriously. 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. Yes. Like obviously we're not stopping people from do-- Oh, man, I should be careful about what I say. I'm realizing- [laughs] No, you said it, you said it publicly that- Yeah... that was in the preview period.

You didn't take it out for the launch. Yeah. Okay. And now you're gonna take it out. So the team is doing stuff that- Yes... I'm not even aware of, so it's great to know the team communicated that. Mm-hmm. I asked them to check in with the lawyers about that. Yeah. 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- So are, are you worried about, uh, sat- saturation or like, uh, training on public benchmarks? So it's like e- easy to cheat. 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- There's like 50 Jeff clones, yeah. Well, uh, sure, sure. Yeah. Uh, oh, the, these-- Let's say that there's some competition. And we'll talk about those. Yeah. 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- Mm-hmm... 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-" This is actually usable. It, well- Yeah...

it, it's, you know, you know, like, beyond that. Yeah. You know, like, the... Hoo. The, the, the launch was crazy, and people could really sense how hard we care about that, and that's truly what I think the long term of this is. And I think public benchmarks are antithetical to this. Like, they are a way to get people trust in intelligence because intelligence has a, has a je ne sais quoi. But the public benchmarks are extremely, extremely gameable. Even if they try not to, they still will.

You know, like, l- back in the old days, every lab had a team to collect data that looks like MMLU to make it look better, which is just benchmarking, benchmarking with extra steps. Mm-hmm. So, um, I, I believe that in the long run, it needs to be vibes and trust until you put it into a workflow and evaluate it for that workflow and measure it and have your own sense of, like, how it does on the exact workflow that matters.

And our job is to keep moving the nines of reliability. This is, like, an ever-present part of our, l- l- like, of what we need to be doing as a company, and we need to do everything to have people know that this is something we care so much about. You know, like, if we wanted to, we could have released Jeff, like, a year and a half ago if we wanted it to be dumb. Oh. Uh, it, like, like, the, you know, my bitterest lesson, right? Like, architecture and, uh, yeah.

I'll bring it up- Hell yeah... since you, since you talked about it, uh, here. 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- Right... unbelievably much. So, so- It's like I can't, I can't emphasize it less.

Yeah, you consider yourself a data lab rather than, like, a model lab. Is that- Absolutely... something-- That's the wording that you guys use? 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. Yeah. Holy crap. Um- So if people are looking for a job, um, we are hiring infinite data people. Actually infinite. [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. Yep.

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

[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 is 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? Yes. 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 l- 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- 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, over fitting 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 over fit to the present, and then it wouldn't work. What we need is to, like- It, it, it almost feels like, uh, like, I-- 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. The general case rather than the specific case. Exactly.

And, like, that requires a lot of intelligence every time. Um, okay, so we mentioned a little bit, you, you, you sort of criticized my thinking as R- like, very RLVR influence, which is, like, very fair. Um, let, let's, let's actually mention RLCD. Ooh. 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? No, not yet.

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. Yeah.

But I, I don't know what you mean when you say RLCD versus what people are familiar with. It's a great question. Yes. And actually, I will give a related question. Okay. Um, what is RLHF? Okay. Right? And actually, RLHF means multiple different things, right? Okay. Like, there's the RLHF of the original, I think it was, like, Paul Christiano teaching a robot to backflip or something like that.

That, wasn't there something- Was that it? That, that was the original R- I, I referenced the PPO paper, but I don't, I don't know. Um, and, uh, so, uh, PPO is not necessarily from human feedback, if I recall. Okay. That's true. 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 100% 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 and, and co-authored, um, the instruction following paper, which was teaching doing PPO on language models. This is the, uh, uh, sorry. I'm trying to, trying to- [laughs] Yeah. Trying to manipulate this, this thing. Uh, this is 2017. Yeah. All right. Um, I, I'm not 100% sure, but, like, that looks quite right. Yeah.

If it has, like, a robot doing backflips or something like that, that might be it. [laughs] Yes. Okay, cool. I guess I, I got it right. Hell yeah. Um. There you go. Yeah. That's the one. So the idea was kind of like can you do, like, ill-specified things with it? So that's like version one. Version two was, uh, the learning to summarize work, um, uh, that, uh, like, OpenAI did, which is actually, like, PPO on language models to do something somewhat ill-specified.

This is, like, a-another thing that people refer to as RLHF- Mm-hmm... which I did not co-author. Mm-hmm. Oh, Dai was there. Cool. Um. [laughs] Hell yeah. And Radford. Uh. Yeah, yeah, yeah. Uh, shout outs to Alec and Ryan. Love them. Yeah. Um, but, uh, the thing that I refer to RLHF is the, um, oh, man, of, I, I'll get into that. You have comments on that, yeah.

I have comments on that paper, but, like, we're so many, uh, tangents deep. Yeah. Um, so the thing that really got, uh, to, to me, the thing that I'm calling to RLHF is the task of instruction following. It's not about the PPO. That part doesn't matter. It's about, like, setting a North Star of this is a valuable direction. It's kind of like the Bitterest Lesson, Lesson North Star. Ah. And for us, RLCD is this new task.

Um, and it is not-- I, I don't see it as jargon. Like, I, I try to communicate with precision. It's just that, hey, here's another North Star. Just like DPO and all of its like, um, you know, descendants also do RLHF despite not using the algorithm in that paper. Mm-hmm. And so cl-clearly stating the North Star is, um, being program, uh, programmable AI is, is, is one, uh, word that I really ca-catch on to, uh, removing the human in the loop- Yes, yes, yes...

uh, from, uh, because RLHF is tuning- Yes... for this so that you can automate everything. Yes. Everything that we can. Uh, did I miss anything else in the, in the thesis of, like, what the North Star is? There is-- That is, that is right. Uh, I'm overly nuanced in my communication. The one nuance is that we need to be practical. We need to be aware of what language models can do really well. You know, like what AI can do. Mm-hmm.

Right? Like, um, there could be programmatic types that are, like, sick AF, but if you, if the technology is not ready for it, it to, it's not a tragedy if that's not out in the world. Yeah. But to me, like, the pre-Jev world was a tragedy because it, it, it sounds arrogant. Uh, uh, hear me out. No, no, no. I strongly believe you. Cool.

Uh, it sounds arrogant, but, like, I felt this way since long before I even had a company. Yeah. And I can, I can vouch that, uh- Yes, I've been talking about this for so long... you've said this at Rounder for, like, three years. Yeah, I've been talking about this for so long. And, and I've been saying it because I thought it would have been easier.

Um, they say they do not do things because they, it, they're easy. It is. They, 'cause they thought it was easy, so something like that. Yeah, exactly, exactly, exactly. I, I thought it, this whole project would take a week. Um- [laughs]... and I was unbelievably wrong, so I am so sorry to everyone at OpenAI 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 and underdeliver 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, 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-- Like this has changed the path of like technological history.

Yeah. And like, we will be exploring that space as a field. Yeah. I, I think-- I, I def-definitely agree with that. Um, you've created possibilities. 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 Jeff 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's-- there are like five other model types that you should be exploring and like let a thousand flowers bloom. Ab- absolutely. Right? Like, like- And some of that- Early Internet- Some of which you will probably also be- Of course, yes. Yes. Early internet energy. [laughs] 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. Yeah. Right? It's like creation is back on the menu. [laughs] You know? Though it's gonna be a wild-ass world and, you know, buckle up. [laughs] I'm so, so jazzed about that. Yeah.

Uh, a- I mean, and, and now you have the, the funding and the, the momentum to do whatever you, you, you envision there, uh, which I, which I think is like very gratifying to see you have after, you know, so long of, uh, of saying these things- Yeah, yeah... but like actually, actually show the world. I know. I just... Such a, such an interesting thing to be a tease the whole time.

Like my talk like felt like it was a cliffhanger 'cause I didn't say how the automation would occur. Yeah. Um, uh, Sean reviewed our manifesto, and he's like- [laughs]... "It's a little bit vague in these parts." [laughs] And you know, like, "What's step one? What is, uh, what..." You know, "What is the intelligence product?" Well, I asked you for a model, and you were like, "Yeah, model coming." Yeah, yeah, yeah.

And like, uh, well, I, I just-- I mainly objected to the word composable- Mm-hmm... but build product on God is fantastic. Thank you. Yeah. I, I-- we, we've really rallied around that. I'd like to think we're not entirely a cult like some companies are. [laughs] But like, we are like jazzed about what, what we're doing and like we are, like by brand is being practical, and like we are all like so super-duper practical. Yeah. It's really great. Yeah.

So here-- and by the way, here, here is the, uh, the step, the, the, the secret master plan, right? Yep. The shape of machine native composable AI. It was your idea to make a secret master plan. I suppose. It's a, it's a Elon thing. When he started Tesla- Yeah... he was like, "Here's just what we'll do." But I, I, I'm giving official credit to you. Oh, thank you, thank you, thank you. Yeah. Thank you.

Um, but like, uh, you know, you, you should have told me you're [laughs] you're, you're also gonna do this model launch, uh- [laughs] 'Cause you, like you told me, you told me half of the story, and then the other half, you didn't have the Doom demo at the time. Yep. You didn't have any numbers to give me. Yep, yep, yep. I was like, "Oh, what?" Well, the problem is I don't believe in benchmarking. Exactly. Right? Exactly.

So like it is a thing that you need to feel, and like I think that this is the way to build long-term trust, even though it like hurt, it hurt us a lot, you know? Like, like last year when we did fundraise, no one believed us. Mm-hmm. You know? Like, like and they wanted just benchmarks and stuff, and we're like, "We're not gonna do that. We are principled.

We're gonna stand by our guns. That rewards bad actors. I don't give a shit, you know, like what you want. Like this is who we are, and we are standing by that." So sorry is my- No, yeah. Yeah. Well, in, in some ways, I think like, uh, choosing the hard path, it-- but it-- you end up making the company that you wanna work in. Yep. Right?

Otherwise, if you sell out, then you're just working in like OpenAI but with my people, right? Which is like [laughs] Yeah. Yeah. I mean, like I'm-- I, I don't have too many regrets on that, obviously. Yeah. Like it worked out so unbelievably well and, you know, like I, uh, uh, I was emotional last night when I was talking about like the reasons I left OpenAI.

Um, and because like it actually had to change my wording after the launch. Um, my, uh, my, my phrasing was, "If an AI winter did happen and I did not do every fucking possible thing I could to like avert that, I would see myself as personally responsible both for, you know, the, the RLHF direction," which I think really widened overpromise versus underdeliver, and also not going all in on this because I think this is, this is where value is going to just be like printed.

So, and it was really cool because I feel like the, the AI winter I'm worrying about is averted. You know? Like AI will be useful. It'll be used for automation. Mm-hmm. It's been less than a week, and like the numbers are already undeniable- Yeah... that it's like being used for real work and like there's... It, it's, it's the Wild West. Yeah. Yeah.

Can you sh-- Uh, just, just if you have top of your head, what numbers are you seeing? Like what's, what's like sign-ups? Like what-whatever you can share. I'm actually not super on top of everything. Like the team is the ones who are telling me all of these things. Yeah. And I'm sure it's like changing every day, right? It's, it's, uh- But like- It's kinda nuts...

if, if there's a milestone that you're like, "Well, yep, that's something we're hoping for. We reached it." I will say a milestone that we've passed is [censored] tokens per day. Nice. Um, and this is not like fleeting [censored] tokens per day. Yeah. This is like even at night, like it's constantly churning, so you know machines are calling it and not just people trying things out. Mm-hmm. So that is, that is so cool. [censored] tokens a day is a lot. Yeah. Um, so surpassing that is awesome. Sign-ups to me don't really matter, and actually, this was like a bit of a mistake we made, if I'm like totally honest.

People on Twitter were calling us like marketing geniuses and all of that, and, um, that was just us. We don't have a marketer, also hiring. Um, and we were just being our genuine, goofy, like irreverent selves, and we w- we were just like offboarding people off the wait list so hard. Um, I Our platform team is so unbelievably, belly cracked.

I think we have more n- up nines of uptime than Anthropic while having- [laughs]... the most unprecedented launch ever. Like, that is kind of nuts. Yeah, yeah. So, like, props to them. Yeah. Um, and, uh, the thing we didn't reali- So number one, wait lists, wait list signups don't matter for, like, a developer platform, in my opinion.

You know, uh, I would guess that a large number of them are not even developers. So they go in, they try some queries, and a lot of people don't get it because they are not programming, right? Like, they're just like, "What? This is not a chatbot. Where, where's my ChatGPT 2?" Mm-hmm. Right? But if like, I, I haven't exactly calculated this, my sense is that if every single human being in the world, like, just wrote a couple of queries, that would be a rounding error compared to, like, one power user's for loop that is just, like, creating value.

Yeah. And the thing we are- didn't realize with a wait list is, like, we can just w- off- offboard anyone off the wait list. It doesn't matter. The scary part is rate limits and then once people start getting value from that, uh, then they just want tons and tons of rate limits because this is what software is, right?

Like, you spend effort upfront to specify your rote task, and then this rote task creates more value than it takes to put in, and then now that you have that- Set it and forget. Yeah. Exactly, yeah. You run it in the background. You make it a dependency, uh, to, like, other things. You can make, like, higher level stuff and, uh, like you just create so much value in the world.

You know, early internet people probably did not imagine, like, the wonder of early 2000s internet, which is still not early internet. But, like, it's, it's through, uh, no offense, composability- No, no... um, that all of the crazy stuff happens, and I just really wanted to emphasize that in our manifesto. We are going for emergence. We are going for, like, being the catalyst.

We're wanting to empower people, and we are going to do whatever we can for that, be it, like, Discords in our town hall with me wearing a garbage bag or not. [laughs] Um, and, uh, and podcasts and, and, you know- Hell yeah... getting, getting- Absolutely... like, 'cause I want the long form, right? Yeah.

It is like, yes, we'll get past the, some of the superficial things, and then we'll go deep. And- Hell yeah... and people will really trust and understand your mission and, like, you know, the, the people that, uh, will resonate, that will end up joining you or, or, you know, uh, uh, buying you, uh, and- [laughs] No, no, b- uh, sorry. As, as a, as a customer. Oh, as a buy- [laughs] Okay, okay. As a customer, as a customer. Yeah, yeah, yeah. Sorry, sorry. That was funny. I'm sorry. Uh, sorry, I didn't, I didn't mean to say that.

Um, but no, any- one, one, one version, one very flattering version of this, like, 36 million views of your launch video. Cool. Um- Up to 38 now. Uh, yeah, rounding error. Yeah. Uh, you know, uh, Navio still has got 74. Fable 5 got 57, so a- like, as far as, uh... And, and, and I, I didn't, I didn't do the stats for, like, original ChatGPT, like- Yep, yep, yep... which there was no video. Yep. So, like, up there, right? Yep.

Like, as, as far, as far as, like, if you were to launch a NeoLab in 2026, I think you're, like, number one right now, which is, like, pretty crazy. [laughs] Yeah. Well, I actually would rather... I do have the shirt, like, your favorites NeoLa- favorite NeoLab's favorite NeoLab. Uh-huh. Um, I don't give a shit about being a NeoLab. I think being a NeoLab...

Actually, we have a lot of, like, swag that's being a parody of a NeoLab. One of them, one of them I have is, like, NeoLab with product. [laughs] Which actually is not a NeoLab. Like, I don't care about that, really. Yeah. Um, what I care about is being a reliable dev platform. Yes. So, uh, appreciate the comparison- Yeah...

but, like, hopefully we transcend past them, and we go back into, like, a thi- you know, like, you know, a revolutionary moment for developers and, like, this stable thing that people can rely on and trust. Yes. Uh, I mean, to that end, um, I mean, I, I think that's one thing that, uh, really impressed me about you guys is, is that, yes, you, you do talk about reliability.

I thought it was mostly about calibration, which like we, you know, we talk about RLCD. Uh, but actually it's also about just, like, uptime and, and, uh, scalability and- Mm-hmm... all those things, right? They, they all sort of, of a kind. And nines. And nines. It's, uh, like, like- Which uptime is in my- But that, 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. Mm. 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 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. Yeah. 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 such a 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 system one query, and then get the result out, and it just branches accurately.

Like, that would be so, so good. Like, that's the, that is the dream. Yeah. And that is, like, going to be, like, a long, long slog. Yeah. We're gonna go into your API design in a little bit- Ooh... just, just to give people examples and, like, maybe path not taken, that kind of stuff. Mm-hmm. 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? Hmm. So- And if not, why not? 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. It- 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. [laughs] I believe that that is, like, the more important property.

Um, uh, you want, um, given similar input, get similar outputs, and it's kind of wild how unreliable LLMs are. Hmm. Uh, like a, a way that we test this is you put, like, UUIDs in, in like little- Yeah... I think they're called nonces- Mm-hmm... 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, 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- 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 shouldn't say this, but no one's here to stop me. [laughs] Um, you know- You have- you s- you s- you sign off on your own PR. [laughs] That is not how it works at this company.

[laughs] Um, I believe for this week, um, my chief of staff, K, is the most powerful person in tech. Yeah, and shout out to K for organizing this. Holy sh- Yeah. Holy shit, she is so fucking competent and powerful. She's incredible. Um, uh, I mean, she sucks. Don't poach her. [laughs] Um, but, um... So I try to be a bit more filtered, but, like, people are telling me, "Don't call it a Frankenstein's monster of models," but because that has like negative implications.

I think Frankenstein's monster was, like, the good guy in this whole... I mean, it was innocent, right? I didn't read it, okay. [laughs] I'll, I'll confess. Okay. That, that one facial expression, I- There's a- My cards are on the table... decent Jacob Elordi movie if you wanna see- I-... sort of the- You-... the adaptation. Anyway.

Th- y- you have no idea how little time I have right now. Yeah, yeah, yeah. My priorities are sleep, you know? Um- Developers, developers, developers. Developers, yes. [laughs] Uh, developers, developers, developers. [laughs] Um, but, um, yes. I- I- we, we do, like, absolutely disgusting things to be on the Pareto curve of intelligence per dollar, and we are going to keep doing that. Yeah.

We're gonna be doing crazy-ass stuff, and I think people really need to think outside of the box. Like, like, part of the reason why surprising is, like, people are thought inside the box, and we continue to do that. Um, as of right now, we are obviously the best at this, and we want to continue being the best at that whole thing. Yeah. So, um, wait, where did, where did we tangent from? No, so, so I- Yeah...

asked you about, uh, will you have seeds and determinism? Oh, yes. So- And then you basically defined reliability and, like- And robustness, yes... from how you see it. Yes. But, like, determi- well, like- I have an, a robustness example that's, that's, uh, real quick I can show you. I would love that. I would just say one thing. Yeah. We can make a deterministic model. Exactly.

Like, like we're hap- if people can convince us that that is a valuable thing to do- Yeah... and we don't have a gigantic GPU shortage- Yeah... um, we can happily make all of these models. We live to please, um, and rev- and, and rev- revolt, revolute. Uh, uh- [laughs] You will throw over everything, except you'll do it in a nice way. Yeah. Uh, and fun- and have fun with it.

So, so like, uh, determinism could be on the cards. Yeah, yeah. It just gets you less intelligence per dollar. Yeah. Uh, well, just having seen the trajectory of OpenAI and Anthopic, uh, you will... Uh, uh, now just trust me now that you will be peer pressured into doing it. Uh, so, like, just people will want it even if they, if you tell them they don't need it, they'll still want it. So, like, yeah.

That, that's the TLDR of- Okay. Yeah. Okay. Yeah, yeah. I will love to... Maybe one day we will see how that happens. Yeah. I've been told I'm, um... They, they, they say that part of our brand is being unshakable. Oh. And they say that that's just the nice way of saying stubborn. Stubborn, yeah. Yeah, exactly, and I'm a very stubborn person. I don't think we could have done it. Yeah. Yeah. No, but so, like, I, I...

Okay, but I tr- I, like, have argued with you before. Yeah. And I know that- And you've been right about developers every time. [laughs] So okay, I give up. You win. You win. [laughs] I, I'm sold that I've argued with you before. [laughs] No, no, I'm just saying, like I, I think that you can hold your ground while also, like, if I give you the right evidence, you can, uh, not...

You can sort of throw away your priors and be like, "Yep, like that actually makes sense to me." Yep. And so, like, y- you know, just, just trust your own gut on this. Yeah. Yep, yep, yep. Um, I'll bring out some- But I suspect, though, that we will be GPU constrained for a very, very long time. Very long, yeah. And anything that has less intelligence per dollar means it consumes more GPUs- Yeah...

for the same intelligence, which is... You know, like our goal is not to onboard companies. Like, it's, it's valuable, but, like, our goal is to have people, like, experiment and do weird shit, and we need, like, we need to, like, get, uh, it to as many hands as possible and, like, starting, like, the California Gold Rush for that.

Hmm. Yeah. I think there is right now, yeah. Yeah. Um, just a word of caution. I mean, I'll, I'll just say it- Ooh, hell yeah... because somebody's thinking about it right now. Mm-hmm. Which is when you say things like, uh, "We will not commit to deterministic models. We, we will, uh, we'll do whatever it takes for intelligence per dollar, and we are, are- we are facing GPU constraint," people are thinking you may quantize your models, right?

Like s- like whatever you had at launch, you may quantize down in, to reduce the quality, um, in order to, to free up, uh, memory or bandwidth or whatever, right? Mm-hmm. And so, uh, you should probably, uh, have some kind of promise, which you don't have to make now- Yep... about, like, we will uphold model quality at launch. People, like... So this is like when people, when, when we...

I, I mean, you were at OpenAI when you launched- Yep, yep, yep... all these, all these APIs, and even Claude as well. Like, uh, when they first launched the models, the model strings, um, did not stay the same model at all times. Yep. Right? You have versioning in your models. That's great. Yep, yep, yep, yep, yep.

But, like, you should, you should publicly commit to some kind of, like, once a thing is launched, we don't change it. We will not change our models when we deploy them. That is insane. We care about developers. L- l- 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 'cause 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- Yeah, you can't have, like, a hundred different versions of the model.

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

Um, but it is not yet our current models, and I'm not promising that we'll be able to keep the exact same models. They will get smarter every time, for sure. Yeah. 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. Yeah. Uh, one, one thing, one thing that's beautiful about LTS-ing models is that actually you can also port them to other silicon. Mm-hmm. Uh, I don't know if you've thought about this. No comment. [laughs] Okay.

So I care about intelligence per dollar. Yes. Right? Um- But speed. What? Speed as well. We'll see. Yeah. We'll see. Honestly- I mean, it's a whole part of the inference tech tree that is, like, I mean, exploding in the past year, right? Yeah. Like, that you can, you can move to, like, a Cerebras- Yep...um, an Etched or whatever and get, like, the hundred, hundred thousand times speed up. Yeah.

Like, I-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'm, I would love to, like, do both and, like, have the market correct me either which way. Yeah. You know? Yeah. Like, like, uh, I, I, I, I would love to be informed by people. Yeah, totally. Uh, and it's not, it's not just real- about real time, right? It's also about scale because, um, at, at scale, every m-microsecond is just multiplied by billions and trillions of times.

It depends on how background it's running, right? Yeah. Like, if it's, like, a big background, like, database MapReduce query, the latency might not matter so much as, like, the cost- Yeah...to get intelligence from it. But, like, if it actually is, um, something more real-time, like user-facing, you have budgets, uh, like between one hundred milliseconds and one millisecond that are, like, totally magical.

And actually, even if you were below a hundred milliseconds, if you could half that time, that means you can get double the intelligence or sequential intelligence calls to have like a, like a phenomenal experience. Yeah. So, uh, right, that is definitely happening right now. It is super-duper cool. I love the intelligence per second use cases, but I don't think that that will be Jev's niche. Okay. Yeah, fair enough. Yeah.

When thinking about the promise of faster and cheaper- Mm-hmm... [clears throat] um, typically the other, uh, the, the trade-offs that other models are offering is faster but more expensive. Yep. Right? And so you're-- Like, one of the reasons I was thinking about why is Jev resonating so much is that you've done the faster but cheaper side of the quadrant- Yeah...which is very, very unoccupied.

Uh- Yeah...while holding intelligence, like, somewhat constant. Uh, yes. It, uh, that, that, that's a very load-bearing statement while holding intelligence constant. That's the hard part, right? Like- Which unfortunately, like, so basically you, you refuse to have the, to like, like do any public benchmarks, or you don't like any public benchmarks about it.

And I will-- I've actually pre-committed- But you need some internal sense. Say again? You need some internal sense of this in this- Oh, of course. Yeah. Of course. We have, we have our own internal evals, for sure. Yeah. But it takes a lot of discipline not to game those, and it needs to be, like, a top-level priority to not game them. Yeah. Of course we do that, right? Yeah.

Like, how else can we make the guarantee that our models are in the pre-different tier of intelligence per dollar, right? Mm-hmm. Like, we're not flying blind in there, right? If we're doing, like, completely weird things with different costs or whatever else, you know, like, how do we compare them? We plot them and get, you know, we try to figure out, like, what is the best for the users.

Yeah. So we for sure measure them. I'm not anti-measuring. Um, but it, it, it's extremely dangerous when you have, like, any alternative incentive, and this is the one thing that, like, I kind of rule with an iron will. Maybe my coworkers might think I rule many things with an iron fist. But to me, like, not shitting ourselves, uh, uh, um, about how smart our model is is one of the most important things there. Yeah.

Like, we need to be truth-seeking. Yeah. Yeah. Uh, agree. Agreed. Um, okay, I wanted to go over some, uh, details on the, uh, API choices. Ooh. Um, mostly because this is the only podcast that will ask you these, these kinds of questions. Oh, hell yeah. Hell yeah. Um, so you have three primitives. Yeah. Um, choice, score, null.

First of all, null, where is that from? Um- Is this, is this like a term in the, in the literature or what? I've never heard- Uh, now, now it is. Yeah. Um, we debated this a lot. We debated this a lot. Um, it is- You know, it is bool-ish, right? Like true/false. Um, it is- But it's continuous. Yes, exactly. Uh, so, so first, the origin of the name is Bernoulli. Ah. Yes.

So it's, that's why it's even spelled that weird way, that is like a subset of the name Bernoulli from like a Bernoulli probability. Right. 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, we 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- Still holding strong. Yes, uh, our mutual friend. Um- Okay, okay. Yes, yes, yes, yes. Um- I respect her for that. Yeah, uh, yeah. Uh, she wanted Jev to be called Meow.

She would, of course. Yes, of course. Okay. You, you in there, you in there, [laughs] But, but like, um, yeah, null is... We, we had to make a new concept for this thing, 'cause if it was a bool, it would be confusing to people. So actually, all three of these are actually new concepts. Okay. These are not types that exist in programming, and that was intentional because they map very closely to types, but they're not quite that.

A score is not an int, so if you had like instructor or pedantic or whatever map ints or floats into scores- Mm-hmm... you'd get a little bit cooked, you know? And, and like we, we were really erring on the side of clarity over the side of like making people like easily understand what's going on. I mean, don't you worry about that?

Don't you want things to integrate directly into things that people are already using? Yes. Yes, we do, and actually, I think that, you know- You have integrations with like other SDKs and stuff. Yeah. But you have... Sorry, you have your own SDKs. Yep. But typically, for example, as a developer relations person, I would be very obsessive, like, "Yes, here is how you use, uh, you know, Jev with Instructor." Yep. "Here's how..."

You know, that kind of stuff. We might have that somewhere. [laughs] I'm so behind on everything that- Someone would do it for you in the community. Oh, yeah. Yeah. Now that, now that you're successful- Yep... people will be like, "Oh, that's cool." Cool. But like, you know. Anyway- I, I don't see that as binary either. Yeah. I actually see success as a score- Yeah... and there's always more to climb- Yeah, yeah...

in like how much we can like be there for our community. J- just to be clear. And, um, I'm... This, this section is stressful because I didn't review the docs- [laughs]... and they're constantly changing. Okay. But, but- But to me, scores- Yes... do exist. So, so, so scores are similar to like LM judging. Mm-hmm. Right?

So like if you want to call it like a judgment, I guess you could, but like that, that is like the, the, the way people con- already use this type of thing, right? Like maybe a null could be like a probability, but everything for us is a probability, and a choice is actually closest to a function call, but a function call is like an extremely disgusting thing that, um, if you want OpenAI juice, uh, sauce, uh, uh, tea, the, the...

We should go back into that later. Like if, like, like a choice is just like the right way of explo- uh, of exposing like a switch match statement- Yeah... within code. So, so it like maps cleanly to an enum. Yep. And you can choose to hydrate it into a function if you want. Yes. And like in the enum, choice is the important part- Yes... of that.

And, uh, like actually, I think these map all into like programming primitives where like choice maps into like a, like a switch statement on an enum. Oh. Um, nulls map to if statements. Yeah. And scores map to sorting or thresholding at a greater than or less than. Okay. And this has been always what the vision is. Like there will be more types, and they will map into programming primitives. 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- Hell yeah... when we're down at this level. Thank you. I love this. Um, no, no, no. This is- This is why we're here. 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. Okay. 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 we're not just gonna cl- 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. Mm-hmm. Um, but all the, um, all the parts, uh, of like the, the, the input, like the state, the instructions, the criteria, all of them can be structured JSON objects. Mm-hmm.

That way, like programs can like insert them in the right spot, and you don't need to like put things- This is-... 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, 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 instructions at once. Slop, yeah.

And then y-you know, like, you s- you hope that every single instruction gets nailed instead of asking the questions in parallel. Okay. And, uh, also I would recommend... I, I-- and I truly say this not from, like, a, like, a it-makes-me-money perspective. Um, I, 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, l-like, a AI beforehand is big system message, and then maybe you have, like, another big AI- Big output, yeah... to see, like, if it actually does this. That's nuts, you know? [laughs] And it's just kinda crazy. Like, it, 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 subdirectory," 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- Yeah, you can verify that it was actually called or- Yes, and, like, our model, our model, like, the, the interface itself is so verifiable. This should be like a sigh of relief. Mm. You know? Like, it's, it's, uh, it's just gonna lead to way better engineering. Yeah, I think I get that. Um, and, and so, you know, one of the reasons people didn't used to do this in the past is because they would just call a, a small LLM, right? Yep.

And it's still too slow, it's still too expensive versus chunking everything that... I've done exactly this myself. Yep, yep, yep. Right? Like, I, I benchmarked, here's a pipeline that throws everything in system prompts and then just gets one big output versus break it down into 100 different things. It was slower, more expensive- Yep... not as good.

Yep, yep, yep. Right? Uh, th-th-then that happens all, uh-- Yeah. It's-- a-and it's, like, super inconvenient. It's unwieldy. Why not just put it all together? You kind of end up repeating some stuff between questions. Yeah. So it's, like, maybe, like, you know, inefficient or something like that. But then it results in something that is very hard to rely on. Yeah. And software doesn't need to run in the background.

It would break my heart if our stuff couldn't run in the background. Is there a way to break things down that you guys have found that works versus, uh, what you thought worked and doesn't work? Interesting. Because, like, people are just gonna be exploring this, you know, now that you've said it. Like, they would use this as a reference and be like, "Okay," like, "that's how I'm supposed to use Jeff." Yep.

Um, then the question is, how do you break things down? [laughs] Interesting. I, I like to break things down into its, like, its smallest semantic unit. Yeah. Like, what is the lowest level thing? I try to never have... Uh, I've probably queried, uh, the, the model the most, uh, among anyone. Mm-hmm. And, like, I try to, number one, in my, in my queries, this is, this is a lot more like the way I prompt things.

Like, I make it really, really structured and explicit, and in the questions, I always-- I like the back ticks, but, like, it works for all of them, you know? Uh, like, be really clear what I'm referring to because we wanted the model to be really literal because when you program, you want things that instruction follow really, really well.

That is what the art of programming is and what AI does is expanding the things, the kinds of instructions that can be followed. So, uh, I'm a fan of doing that. I, I... Sometimes I'm a little lazy, and I, like, I have, like, more, like, hybrid things, but, like, I think that for, like, really big production things, you just want to, like, keep on adding more questions, and you wanna make it really easy to add more questions.

You know, be really, really precise about all of that breakdown and then have the code to have the exact behavior you want. If I could give, like, a tiny little example of this, um, is, um, like refusals, right? Like, uh, I'm not gonna talk about why we don't refuse. I might have done that already. Yeah. It's, like, all a blur. You did already, yeah.

Um, but, uh, like, for refusals, I don't think you should ask, "Should I refuse here?" You know? Uh, that's a really... It, it-- the, I think the answer will be pretty good because, like, that's a system one compatible task. But I think you're way better off, like, asking many different independent questions about, like, the different situations you can refuse about because instead of having to, like, just guess based on you, you know, you can actually specify what you want.

And beautifully, you know, like, and I think this is, like, truly, really beautiful, if you find a situation where it's like, "Oh, it didn't refuse because of this reason. I didn't specify this part of the task," that is awesome. That's what software engineering is about. Like, you fix the bug by adding that question in, adding the threshold, maybe remembering that as a test case, and now it is just solved forever.

Like, your software can't forget about that, like, in the prompt because of context rot. It is just there, and you can, like, just keep measuring that forever. You know, and if the models are not perfect at some of these things, you can choose what threshold you want for all of these factors based on real examples.

It's like, it's like ML without the ML, and you can just do it for, uh, for anything. And, like, there might be some things the model's not good enough yet, right? Like, I would-- I'm a little bit afraid when I see people doing trading with the models- Mm... like, um, automated trading. Mm. Um, it, it looks cool.

I th- I, I just think that people should leave it to the professionals. [laughs] Um, but, like, that's just a very hard, high-level task that maybe the models aren't good enough yet to figure out. Yeah. Uh, well, I, like, uh, even if they were, then they would It, it suddenly wouldn't be 'cause of efficient market.

But, like, that's one of those things where, um, you can, like, break it down into things and just evaluate them, and you might be like, "It's not smart enough at this. Maybe we don't deploy it yet for this version." Yeah. Or we make a trade-off, or we err on the side of safety. Or like, hey, the models are not good enough at, you know, like, detecting, like, this weird combination of, like, sarcasm with a VIP customer, that this is when we escalate to a human, and that's what confidence estimates are about too.

Okay. Uh, great, great answer. I think, uh, one thing I'll, I'll mention very quickly, which, uh, I don't expect that you have as too long of answer for- [laughs] You don't... is, uh, well, no, no, no, no. It's just- [laughs] Yeah... it's just, um, typically, like, you are still relying on thresholding as, like, the, the lever that the user can pull. Mm-hmm. But what if just the, the, the, the calibration is wrong, right?

Like, you're s- saying your, your calibration's perfect. But- I didn't say that... I, I mean, like- I didn't say that... so is it ki- perfect cali- and then good calibration means, like, lower value is lower, like, sort of probability lower. Higher value is probably higher. But it could be wrong. Of course, of course.

It could be locally misaligned. Yes. And so then I would want to fine-tune it or something, right, which you don't offer, but you could. Uh, I, I, I- We could... again, see, this is a short answer- Yeah. Oh... which is you, you don't have it right now. [laughs] Oh, do we want to offer fine-tuning, is it the question? That could be, that could be one version of it, or you could have a, a different knob. Yeah. Right?

Uh, where, like, because, like, n- right now you're, all you're saying is, like, if something's wrong, uh, skill issue, you should, you should just change the prompt again or break it down even further, or you change the confidence. Yep. Those are my two options. Yep. Right? And that doesn't feel super satisfying if your model is just getting it wrong. Yep. And it will, it will get many things wrong, to be clear. Right.

We have, like, a report issues button, complain to us in Discord. We want to make it a lot better. Every single model version will be, like, notably better. Yeah. We will stop shipping them quickly if they weren't getting big improvements. So number one, that is, like, totally reasonable. I think that that's simply pragmatic to admit that AI is imperfect at some stuff, right?

I do think we'll find use cases that they are, like, good enough at, and good enough kind of depends on the use case, right? Like, human beings can do a lot of work despite being bad at that work because their EV is quite high, and presumably with the right thresholding and everything, there probably is, like, large amounts of work that could be done even if mistakes are being made.

Um, on the question of fine-tuning, I could imagine, I could imagine it in the cards. I do have concerns because, like, uh, in the what people need versus what people want category, um, like, I think general models tend to be really... Like, like, again, there's, there's the je ne sais quoi of generality that making it good at, like, a million other tasks than this one narrow task might make it better at edge cases in that task, which I'm, I would be a little bit afraid of, you know?

Yeah. Um, I, I, I could imagine it is, is my answer. Uh, I'm endlessly practical on these things. I want everything. Like, my vision of the world is j- I, I... There's, there's so much we want to be building. Yeah. But also, like, I would not want to ship something that is, like, a giant foot gun, uh, like some other AI companies, which, yeah.

[laughs] Well, you know, so both OpenAI and Clau and, and I think even Gemini have rolled out fine-tuning and then took it back. Yep. Uh, which is an interesting, um, observation that pretty much fine-tuning is now in the domain of open source models. Uh. [laughs] Yes. Yes. Um, I do know about that. And, like, it was kind of crap, so, like, that's probably better- Yeah... that they took it down.

Yeah, so it could just be a foot gun, and then telling people that fine-tuning it is probably the wrong way to go is great. Another interesting answer could be that, like, well, our model is so different, like, you know, j- the scene and the quantization doesn't, doesn't apply to us. Yeah. Output tokens doesn't apply to us.

Fine-tuning also doesn't apply to us. [laughs] Well, a- actually, I, I, I'm super open to that possibility. Yeah. Like, uh, my, uh, this is not a promise. This is a desire. Just so to, to make it clear, I like to be really honest. Like, I think that, um, as intelligence per dollar gets cheaper, cheaper, cheaper, cheaper, I think that we could get really, like, small approximate things that hopefully are proxies for intelligence.

Like, is there a world where people don't write regexes anymore because, like, you know, the intelligence per dollar that uses AI is cheaper than, like, the complexity of a regex? You know, that would be kind of sick. I would love that, you know? And it might require fine-tuning for some of those narrow use cases to really get past the threshold. We will see. My hope is calibration gets that. Calibration plus a ca- cascade of models.

Like, if it's super confident, then maybe it's right. And, and if it's in the middle, then you do the next bigger model, and you chain off from there. I don't really know how that's gonna go, but, um, yeah, I, I, I could imagine it. And something that I could imagine too is, like, imagine you have, like, a, a series of, like, we own the en- entire Pareto frontier.

Something that a business might want to do or, uh, I, I think a hacker would be okay with dealing with a Pareto frontier of models. Maybe a business wants something more dynamic. You could imagine, like, having, like, a s- different sizes of models and to dynamically pick which model based on how smart it is on different parts of your stack.

And you could even imagine, because of how simple our thing is, you could imagine, like, some automatic fine-tuning on that. Yeah. Uh, not the promise in the slightest. I'm just, like, cooking on sci-fi. But you would consider different sizes of dev models, so to offer that gradient? Absolutely. Yeah, yeah, yeah. Yeah. Like, we, we... Like, how would I know how much intelligence people need? I don't know. Right?

Uh, yeah, I don't know either. [laughs] Demand is, demand is, uh, unlimited. [laughs] Well, yeah, people are telling us not to ship things right now because we don't need to ship things because again- It's good enough. Yeah. Yeah, but that's kinda lame, and, uh, I really like the saying, "This is something that I hope people hold me to," because it'll be hard to, to- To come back... to walk back from. Yeah. Yeah. Like, the...

I, I don't know if it exactly the saying that culture is what you do when the market doesn't reward it, and I really like that because I think that we are standing for something. Maybe in the future- What we're standing for is, like, so obvious that we're the equivalent of, like, boring, like, Visa or something like that, and, like, we're just, like, a utility that no one really thinks about, and I'll be wearing non-pink suits or whatever else.

[laughs] But I really want to be, like, rallying the world to this, you know? Like, I want to keep doing cool stuff bec- not because we need to, but because I want, like, people to realize that this is just the beginning, you know? Like, that wasn't even meant to be the opening salvo. That was, like, kind of like a, a, you know, low-key research preview or whatever you wanna call it. Yeah. And there, there's a lot more we can do. Yeah. We've-- With...

Like, machine native intelligence is gonna go wild. So not the only s- potentially not the only size, potentially not the only model that you guys launch, uh, you know, that you, you want to open people's minds- Absolutely not for any of those. Yeah. I want, I want to, like, meet whatever needs we can. Yeah. Right? Uh, like, uh, at-- But with, like, a giant caveat, I don't want to be like OpenAI's product teams that, like, like, throw stuff at the walls.

Like, I want it to be, like, in a, under a unified vision. Like, if you go back to the manifesto, like, everything needs to be under one of these three, three things- Yeah... in my opinion. Um- Uh, I'm not- Yeah... I'm not prepared to do this, uh- Oh, I'm sorry. [laughs] I'm sorry for asking. I, I can just, uh, just talk about it. Like, we have, like, three steps in our stuff. Yes. It sounds like a tease.

I want everything to go under one of these three things- Good, good... to keep pushing the boundaries and everything. Like, like, this is not-- these are not, like, checklists. These are, like, axes that we think build, like, the foundation of, you know, of, of, like, a new te-te-technological revolution, and I want all of the, all the bets we make to be somewhere in there, and we will be doing some weird, weird stuff model-wise.

[laughs] So, um, because machine native, right? Like, uh, humans don't need to totally get it. It needs to just be valuable. Uh, you know, th- w-with, uh... Just give people a tease or hints. Like, what, what does weird look like? What is weird? I, I'll give people a hint. Yeah. Um, some people are trying to call them decision models. Okay. Um, th-the-- our primitives are decisions. Um, I, I wouldn't do that, uh, because I think there's other types that are machine native that are not decisions. Okay. We'll leave it at that- Is that a fun-... and let people guess... a fun hint? Yeah, yeah. Yeah, yeah.

I think it's a, I think it's a pretty fun hint. Yeah, yeah. There's people... Look, there's, there's people saying, like, "I've done this before. I made a decision model a year ago." Like, "Geoff is not new and Geoff's not cool." Yeah. But, like, I, I think, uh, there's the categorical, like, here's what you're establishing is possible.

There's the, uh, performance of, like, well, actually the-- for the benchmarks and the numbers that you're getting, you are still beating ev-- as far as I can tell, you're still beating every single clone of you out there. I, I don't care about the benchmarks- Exactly... just to be clear. Exactly. So, like, uh, like, even if we were w-winning or losing, I want to denounce them. You've established the category, right? Yep. Yeah. Yep, yep.

Uh, but al- but also I think this, this nuance between decision models and system one, uh, I think is actually the thing that you're trying to- Yes, and I just wanna make software engineers super powered. Yeah. Right? Uh, l-l-like, with AI. Like, and, or, or, like, the tragic thing to me is, you know, in that AI winter direction, I think, like, it's, it's, it is so sad that AI was so powerful yet so underutilized. Like, um, um... It's a thing that gets me emotional, but, um, man. Like, I think that that is-- I, I don't want to, like, just be, like, pure techno-optimist, like all technology is good.

I think what was happening now was, like, a travesty. Like, it's... A-and, like, there's... You know, I just want to, like, open up those possibilities for people. [laughs] Yeah, I'll stand it there. I, I, I, I, I've, I've cried too much these last few days- [laughs]... to, to, to want to do it on the record. Yeah, yeah, yeah. No, it, it- Yeah.

I appreciate you sharing a little bit of that, and I, I think people can see that you're very authentic and- Yeah... passionate about this. Um, th... You know, you don't necessarily get that from a name like Typesafe AI, but, like, uh, I think once people immerse themself, themselves enough in, like, here's the genuinely different direction you want the world to go- Mm-hmm...

and, like, actually you have done, like, the, the hard part about going from zero to one on the, on the thing, then, like, like, now let's all go to-go together in, like, the, the new direction, right? Yeah. Yeah, yeah, yeah. Yeah, yeah. But I, I don't kn- I am sure that I won't think... Maybe I will think that the hard part was done, perhaps. I think that there's going to be many more hard parts. Um, like, if, you know, all sorts of stuff gets automated and we finally see GDP growth and, like, you know, it's like, you know- Yeah... a jazz party every day, then maybe the hard part is done. But, like, I, I don't th-think so.

And, like, I really, really think that people focus too much on speed and cost and not enough on reliability. Okay. Like, reliability is what makes it delightful. Like, reliability is what, like, allows you to trust it. You have this line, uh- [laughs] Yeah... TFP growth rating three percent in five years. Hell yeah. I've never seen- Hell yeah. Let's fucking go. [laughs] I've never seen- Yeah... a lab care about TF-TFP growth.

But, like, that is what an economic revolution is, right? Like, it's actually extremely consistent with what the OpenAI charter used to stand for. You know, it was talking about, like... I think the charter is the same, but they've kind of tried to move definitions around to, like, you know, hundred billion in profit or something like that. Not that I hate on OpenAI. It, it wasn't, like, a...

Yeah, it wasn't a well-defined term what AGI is, right? They tried to do it. Yeah. Yeah. Right? Like, doing majority of the world's economically valuable work, and they should have to answer the question: How can it do millennium price problems in math and zero of the world's economically valuable work, like rounding error?

You know, like, I think that all models are roughly tied right now at zero. There's some chance that, like, we have started already. But, like, I would guess that it's not yet one percent. Um, and I think that that will show up in... Like, when it does happen, it will show up in the economic statistics. Mm-hmm. 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. Mm.

You know, and I- Just, like, disappear into the background. Exactly. Yeah. You know, like, like, uh, the-- I, I say this in my talks. Like, how can it be that twenty nineteen software, like software, SaaS, whatever, uh, super-duper valuable, right? It's twenty twenty-six 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 SaaSpocalypse. I think SaaS is going to be supercharged by this.

They are 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. Yeah. I, I think, I think so too. It's a s- it's a beautiful thing that you've unlocked, you know? Yeah. 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. Fuck. Um, [laughs].

That's a hard one. [laughs] That's a hard one, my friend. Um- 'Cause people now are just trying to Jev everything, right? [laughs] Which, like, probably- Jev everything... is gonna fail, right? [laughs] Um, but, like, some things are gonna be good. J-Jev everything is pretty funny. Yeah. Uh, it's a pretty funny way of doing it, saying it. Um, the-- so I'll tell you the truth. Yeah.

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? Mm. I don't think it's about, like, spending more money necessarily. The, the empirical results m- just might not be there, right?

Um, so empirically, I believe that these, like, pre-trained super condensations of intelligence are fundamentally system one thinkers. I think that they truly... L-like, system one is the closest thing to describe what LLMs are strong at. RLVR has done incredible things for system two thinking. I am at awe. It is super freaking cool.

Like, I don't think that it's going to result in AI doom in the slightest. Um, not, not zero percent, of course, 'cause I think zero percent is miscalibrated. But, like, it, it's, it's really cool what they've done, and they've really pushed it to the limits. Well, maybe they don't think so, not the limits limits.

But, um, like, it is, it is a weird thing for models to do, and they are very fragile at this. Like, think about how people used to talk about AI back in the ChatGPTJs. Like, wow, it's really general. It can do a lot of general things. And, and, and then, but it's bad at math problems and, like, GSM8 grade school math. Um, and then now look at how people talk about RLVR. It's so fragile. It's so jagged. You know?

Like, it can-- why can it do this, like, really weird thing? And actually, you know, math is not just spiky, it's fractal, right? This is because RLVR is, you know, like, if we talk about, like, what is the North Star for each thing, RLHF is please humans, right? That is what the human feedback is. RLVR is optimize benchmarks.

You know, that l- everything that goes into the RLVR category literally is a benchmark by definition 'cause a benchmark is programmatically verifiable, simple outputs that can, like, do well. And, you know, RLCD is make it reliable for, you know, programmatic use. And, um, yeah. That, that, I'll- Yeah... yeah. This, this, uh, maybe I'll, I'll offer some thoughts, and then you can sort of, um- Mm-hmm...

correct me if I'm wrong. Um, one, for example, one thing that I've been thinking about is also... Uh, so I, I threw Jev at a bunch of things when you gave me access- Ooh. Yeah, yeah... uh, on day one. Um, and, uh, multi-hop reasoning, right? Like- Yep... uh, s-so single hop, fantastic. Mm-hmm. Like, state, state of the art. Uh, you should never use anything other than Jev for single hop. Yep, yep, yep.

Multi-hop, it's gonna start to falls down. Yep. And it's, like, kind of mon- monotonically increasing as you increase the hops. Yep, yep, yep. Right? So, oh, yes, back to that empirical question. It depends on what we can, like, pull out of the models. Yeah. Right? So we want everything. Like, we want to unearth as much intelligence as possible, period.

Um, the models-- like, I see us as, like, unlocking and smoothing and sculpting the intelligence while, like, adding new capabilities and, like, you know, like, like filling in gaps in it. Um, and we will be filling in, like, you know, more and more and more and more of these gaps over time. But the reality is that we are in the business of unearthing properties.

Those properties are actually a function of what is available from, like, these, like, you know, these conden- condensed cores and, like, Frankensteining them all together to have all of the properties of everything. Yeah. You know? Um, but the reality is we are in the business of unearthing as many capabilities possi- as possible, and system one just happens to be the description of what works.

And everything that works in that paradigm will be system one-ish. You know, like, I, I am... Like, like there is a reason why we don't do what's called latent reasoning, reason-reasoning in strings. Yeah. I think the reasoning... Like, the, what models do really well is reasoning within the models. It's not totally complete. It doesn't do great at all. Wait, la-latent reasoning is reasoning in strings?

I thought latent reasoning is reasoning in, in, inside the model weights. The, the re- That I think that's- I don't know... people used to call that- I just wanted to clarify... continuous reasoning. Okay. I'm not entirely sure. Okay. It was called latent reasoning because, like, it used to be that the reasoning traces were secret. Mm-hmm. So they're kind of like a latent variable for the answer. Ha. Yeah.

So what's secret is now shifted. They, well- Yeah, yeah... it, it's still secret for OpenAI Anthropic, right? So no reasoning Jev- Yes... as far as you will ever do it, right? Because that, that, like, violates the whole promise of system one. I-- my promise is to do whatever necessary- [laughs]... for machine native stuff. Yeah.

I, I could imagine there are, like, there are some forms of reasoning that are less slow, inefficient, and fragile that I, that are, like, totally on the cards, just to be clear. So pragmatic person, I'm not making promises on, like, you know, like, methods. I'm making promises on, like, the, what my ROI North Star is, and I'm going to fight for that.

Like, like, you know, like, like this launch didn't happen, and we are still, like, hungry for our place in the world. That's great. Yeah. Yeah. Um, I, I think the other thing that, uh, vision is another one that it's like a, a big, like, you know, capa-capability that you don't have. But maybe it doesn't ever belong in sy-system one.

I, I think I have a pretty good vision. Uh, what? Sorry? I think I have a good vision. No, no, no. Sorry- I am kidding. I'm kidding. [laughs] Oh my God. Yeah, yeah, yeah. Um- Because people obviously, the first thing they want is vision because of the Doom demo, but also just, like, everything, you know, o- other than text, is vision. Everything is in the cards- Yeah... in my mind. Okay.

Like, like, um, a- and actually, this is, like, a debate we have. This... Man, your audience is probably, like, the great one to have in this debate. There's a question about, like, how much do we try to, like, give people what they think they want, which is what we did in stealth for two years, we just knew that this is obviously going to be valuable, versus give them what they say they want, right?

A- and, like, uh, there's a lot of dimensions of this, right? And y- you know, like, context length is an example of this, right? Um, every single model, including ours, I actually think as far as I can tell, ours is, like, by far the best- The longest context. Yeah... at not degrading in long context. Yep.

Um, but, like, the other providers are just like, "Whatever people wanted, let's just give them the stupid thing." And, um, like, we need to figure out a balance for this because- Yeah... you know, like, if you take the, the, the former side too far, give people what they want, you end up with, like, anthropic nanny state-style thinking, which is very, like, anti-developer.

While, like, the pro-developer route would be like, give them what they want, but developers are... Like, we don't want to put the burden on them to figure out the je ne sais quoi of intelligence. So we are trying to, like, figure out this navigation of, like, how quickly to release things, uh, to still, like, have our, like, brand of trust, and also, like, teach our... treat our users like adults- Mm-hmm...

that can make informed decisions that, uh, don't need, like, nanny stating on top of this stuff. Yeah. I think that's fair. Uh, and- Yeah. And, and we don't know the answer, to be honest. Like, uh, we'll, we'll have to figure it out. It's gonna be... That's probably going to be, like, one of my biggest debates over the next couple of days. Yeah. Because, like, we have a lot of stuff.

Again, we didn't expect it to pop off, so we were like, "We need some follow-up launches." So yeah. Yeah. I don't know, I don't know if you didn't expect it to pop off. Like, I... You put... I, I saw the work that you put in. Like, I have never seen you lock in so hard as- [laughs] It was like the last two months, basically, right? Like- Well, well- Like-... that's also because- [laughs]...

my chief of staff made me lock in. [laughs] Yeah. So- Like, it's like, uh, I have never... I thought I worked hard before. Yeah. And- No, but, like- Yeah... you were showing up at our writing workshops, and I was like, "What are you doing here?" And, and, like, oh, you- It was useful. It was great... you, you clearly, like, were very intentional about your launch. Yep.

And the work showed, and like- Yeah... congrats, like, you know- Thank you. Thank you... all the kudos. Like- Um, I hope to keep locking in- Yeah... is my, is my sense. Yeah. I, I want to... Like w- like, I think that we've passed many great filters for the, the, the tech world, what we're wanting, but, like, there's still gonna be a bunch more. Yeah.

And, like, holy smokes, am I excited to fight the good fight. Yeah. It's exciting. Um, before we broaden out to, uh, uh, topics outside of TypeSafe- Ooh... I just wanted to offer, uh, any other things that you think, like, underrated or misunderstood about what you have launched. Underrated or misunderstood? Yeah. You have pan outs...

Uh, sorry, patterns here, uh, maybe, maybe you wanna go into that, um, model jaggedness, any- anything. Yeah, you know. Give me one- Yeah... noodling of it. Oh, man. I, I would rant about all of these. I, I really shouldn't. I really shouldn't. Okay. Um- And, you know, people can come go to your Discord if they want. Yeah. Uh, uh, people put a lot of love into the cookbooks- Yeah... is what I will say.

The cookbooks have, like, some fire stuff. We had considered putting a bunch of these things, like, in the main launch blog post, but it got kind of long and unwieldy and, like, very power usery. But, like, we really, really, um... I, I'll be frank, like, before the launch, um, every... Like, like what we're saying sounds, like, sounds like, like this weird alien tool, why would anyone need this?

You know, it was a very weird thing. We were very worried about teaching people about, like, this new frontier. It obviously succeeded, but, like, we put a lot of work because we thought the education would be, like, a gigantic bottleneck for us. Um, I pr- it probably worries and it's no lo- probably no longer a problem because people are doing things, like, well beyond what you could ever expect. They'll show you what- Yeah... how to use your model. But... Exactly.

But, like, they, yeah, and their use cases are, like, kinda cooler than ours. Like- [laughs] Like, there's a bunch of stuff where I'm like, "Man, if that was our demo, holy shit, that is way cooler than what we were showing." Um, like the, the computer use stuff, holy smokes, is it cool. Um, but like, like, we put a lot of love into this. This is not like AI-generated trash, as far as I know. Yeah. We put a lot, like, it's, like, a lot of love in here. Yeah. Fair enough. And, like, each of these are...

Like, like, there's real alpha there. Okay. Like, these are inspired by solving real customer problems that existed, and we went through the work of, like, helping them do cool-ass stuff. Yeah. How much, uh... While you're talking about this, right, how much validation did you do before launch? Like, what, you know, what was that process like? What was that process like?

Like, clearly you did some, but obviously, you're not getting in touch with as many people as you are today. Yes, of course. But, like- Um, I actually think that the reception was pretty bad, and, like, actually for the non-technical people in the team, they were really worried. Yeah. You know, like, there was a lot of fear.

It's like no one really gets this. And, like, you know, they don't want it. We're, like, selling, like, a vitamin and not, like, a painkiller. Like, should we have FDEs to, like, write the software around solving that problem? Yeah. We had almost no revenue before launch. Um, it was kind of like, uh, like we...

Like, the technical people were, like, obviously true believers, right? Like, we knew that this was sick. Its pro- computational properties are, like, off the charts on, like- [laughs]... so many axes that we're like, "Yeah, obviously it's gonna be huge." I was definitely super afraid, which is why I locked in super hard.

But, like, the most common thing was the, like, I would say, like, more than half the people we had play with it just did not get it. And, like, the people who did, uh, like, were like, "Man, this is really cool, but how do we get this through procurement and stuff like that?" You know, it was like, it was, like, quite a, uh, quite a battle, and we just knew, like, okay- Our target market is gonna be developers. People will find the use cases, and that way everyone is gonna FOMO in and, like, I, I don't want to rub in people, like, changing their minds with the facts changing. Mm-hmm. I do want to p- call into question, like, the concept of product market fit, you know?

Yeah, but, like, because, like, there was a product, there was a market. Like, we were like, "Hey, do you want to use this?" And people are like, "I don't know, really know if it solves our problems." It explodes and everyone's like, "We need as much rate limits as we can. Can we literally give you GPUs because we are constrained right now?" Yeah.

So of course, uh, you know, like marketing is an element of it- [laughs]... of course, but I don't even think it's about marketing. I think it's about, like, passionate developers who've, like, you know, our souls basically resonated at the same frequently, and that frequency, and that got, got everyone else excited too. Yeah.

And I'm hoping as well that, like, we as a company will be eternally, eternally, eternally grateful to those developers. Like, not, and not just like, you know, like, um, the, the companies that, like, are, like, start off with developers and, like, go to enterprises. Big oil market, yeah. Exactly. Yeah.

And like, I'm, like, even thinking about, like, how can we launch things that are better for... Oh, man, I don't know if I should say this- [laughs]... but I will. Better for developers than enterprises. Exactly. Okay. How do we do that? Like, how do we empower them? Um, and I have cooks, I have cooks. But, um, it's a, it's a very weird thing to do and, like, I don't know how else I can show my thanks and loyalty to that.

You know, like, a- and, and that's why I did, like, the, the dyeing my hair yesterday. It was like- [laughs]... it's like I wanted to talk to them 'cause it felt dirty to me during our company's, like, most important times not to keep talking to them. Good. Uh, well, I mean, that's why- one of the reasons- Yeah... you're here.

Hold me to that, please. Yeah, yeah. We will, we will. I, I try to be principled. Yeah. Quote me on this. [laughs] Uh, call me out. You know, have the pitchforks out if I change. Uh, I was just gonna briefly show the computer use stuff. Whoa. Is this, is this what you're referencing? I've never seen... I haven't... I- I've seen... [inhales] I saw, like, a airline browser use thing.

And, um, inside this new note, let's make the title say hello. Wow. 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? Oh, the voice use cases. This is actually the first one I've seen. This is- Oh, okay... open up the photo booth. Wow. [laughs] Oh, wait, wait, wait, wait.

Hold on. Can you pause- can you go back a second? Can you go back a second? Um- Rumors claim Anthropic engineers wor- worship Claude as god. Wow. [laughs] Wow. Dang. That's pretty funny. Um- And here you are building Prod. [laughs] Abso- wow, this is sick. Uh, yeah, so, so c- clearly you can operate the whole computer with voice, with Jev as a decision model. So just like I'm anti-benchmaxing, I'm also anti-demos. I want to make sure that it works reliably. I love people are playing with it. Mm-hmm. 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. Yeah. And I would lo... Man, that looked really cool. [laughs] That looked really cool. I want that. Uh- I want that. Like, when my, when my wrists are sore, I, like, just whip- Whisper flow everything. That'd be sick. Well, as, uh, well, you know, just, just to round out the use cases side- Yeah...

'cause I, I do have to let you go. Um- [laughs] Uh, who's, uh, who are the, who are the bigger companies that have reached out and have surprised you with what they want to do? Just, you know- I am so out of touch for that. Okay. People have shown me screenshots of companies, and from what I've seen, it's all of them. Um- Yeah.

Mo- mostly, like, you know, for, for those people who work at larger companies and they're not doing this kind of work, uh, I just wanna give people examples of, like, you should go look that up, look that up, look that up. Oh. So, uh, like, I think demos are super-duper sick. Obviously, the coding agents are, like, gigantic use cases. Mm-hmm. Like, they are, like, also super sick. Oh, Cog is all about Jev right now. Oh, hell yeah.

[laughs] Can I, uh, can I give a little bit of a, a tangent about coding agents, if that's okay? I- Yes, please... uh, uh, let me- We love coding agents here... give me a second. Give me a second. Okay, actually, r- I'll come back to coding agents. Let me describe, like, the big families of use cases. Yes. Like, we've mapped this out from first principles, like, long before release.

They are what we call dark data. Like, people hoarded big data, but they would not throw a LLM at it 'cause it was too expensive. So large companies l- adore this. They have, like, piles of data that they wish they could analyze, and this is, like, a data scientist's wet dream. So this is, like, this is a giant one.

Like, I think this plus, um, coding agents are the m- big money makers because that's what the all-- where all the volume is, right? Um, there's the real-time stuff, you know. Like, people who need, like, intelligence in the loop. They-- Like, I would guess that every CEO, if not CTO at those companies, um, knows how much better their product gets with every, like, ten milliseconds shaved.

Yes. Uh, and, like, to- Especially e-commerce, yeah. Yeah. Oh, o- or, like, assistant-y things, you know. There's many AI assistant-y things and, like, as far as I can tell, they really love it. Um, again, I'm not in the front lines of customers right now, so I just get-- know what my team tells me. But, like, this, I'm so excited for this. I'm really excited for this for games.

I really wanna play, like, sick-ass auto battlers, where you're, like, commanding your team or, like, semi-auto battlers. I, I think that'd be so cool, but don't make it too good while I still have a job. Um, [laughs] and, um, the, you know, like, there's the, uh, the, the, what we call, like, verify everything. You know, like, verifying all LLM calls, kind of like observability.

Um, I think, uh, actually, on the note of docs, what people should be doing is, like, the parallel questions are very cheap. So if you have, like, big states you wanna ask many questions on- This right here? Yeah... put IDs on every, like, message and then ask a question about each ID. So, like, when you have, like, a long state. Uh-huh.

So that way you can, like, pay for the state once and ask lots and lots of questions about each message within it. I think that is, like, a, like, a great way that, like, saves money and is- Which, by the way, I always think, like, it's interesting framing system one, system two, because it, it basically makes the case that you should always make one or ten or a hundred Jev calls for every one reasoning call that you make Well, uh, maybe. Right. Well, I, I mean, I don't-- I would like people to spend less. Yeah.

You know, maybe you do like, you know, one half the reasoning calls and, like, ten Jev calls each or something like that, or whatever solves the problem that, like, couldn't have existed otherwise. Um, okay, uh, wait, uh, number four use case was, uh, what I describe as, like, smart software. Like, software that's intrinsically composable and, like, does, like, weird, fun stuff that could never happen before.

You know, like the programming language as Jev thing. I don't know if you've seen that. That is so cool. Man, if we knew how to give out credits, because, like, we're really early in our infra days, I would wanna give all these projects credits. Mm. Um, and I, I think that those are, like, how we've mapped out, like, the main use cases. Um, computer use has also come in kind of like the real-time direction as well, and, like, that's really, really cool.

If it is reliable, I am super jazzed about that. I suspect we can make the model a lot better at these use cases, 'cause, like, that came out of left field a little bit, so that's, that's really cool. Um, um, um, on the coding agent thing, and this is, like, a really surprising thing that is happening right now. Okay. Um, Claude Code and Codex are, I believe, the, the winner, like, the, the number one and two.

I'm, I'm not entirely sure. I don't follow closely- Roughly... 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. Mm-hmm.

But all the open coding agents are, like, fucking jazzed right now because they're, like, getting their Jev on and- [laughs]... like, the thing is there's-- I'm sure they're trying a lot of weird stuff. Mm-hmm. 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. Single model, yeah.

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, yeah, 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. 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- Of course... 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- Hell yeah...

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. Oh, uh, did I not talk about safety alignment at all? Oh, you did. You did. I think I didn't. I think that maybe I didn't. You did. Oh. I, I 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- Of course... every NeurIPS. Yeah. 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- Mm-hmm... uh, the public is, like, clearly not ready. Um, and I, I'm sure you have strong feelings. Um, I feel like this is the kind of thing that is a dangerous topic- Okay... to talk about. Um, I'm happy to talk about it. I, I live for danger. [both laughing] All right. Our company brand is chaos. It's not Jev.

It is irreverence and chaos. And you know, yeah, and, like, you were at OpenAI during, like, the, one of the very first, like, very visible incidences, the blip, right? Like- Oh, the- Which, like, and, like, you-- the dominoes have gone down now- Mm-hmm... to now every frontier lab has co-signed a document saying that they want to pace. 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- Yeah...

which is that, um, as you RLVR more, like RLVR is like, 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?"

And, like, everyone is like, you know, giving tons of, like, resources to, like, the code gen team, which, like, they did s- have some successes, but they were trying really hard to do RL on co, uh, like unit tests. And it didn't work, obviously, right? Like, you needed reasoning for that. So, uh, re-- So just to be clear, RLVR is not purely about the reward. It's about, like, the shape of everything too.

And part of it is that reasoning is included in here, like this latent variable that you're doing things. And when you're doing things, you're just letting the models do whatever they want in order to make them be as powerful as you can to answer the hardest problems. And this whole pace the frontier discussion I think is, like, a very narrow focus because it assumes that everyone needs to do more RLVR, right?

Which, um, like I obviously don't think I need to do more RLVR on our models. Mm-hmm. You know, I think zero is the optimal amount- [laughs]... for our shape. Hey, right? Like, come on. Yeah. Uh, you know? So, um, it, it's really, I think, a bit of a sleight of hand where they are saying that we actually want to keep doing the thing that looks dangerous because it does dangerous things.

You know, like people say like, "Oh, maybe the sandboxing was a problem," or whatever else. Um, yeah, I mean, obviously it is, and they could've easily solved that, right? But they chose not to because the more things you let the models do in this do anything category, the more powerful it is, right?

So, uh, like n- they're-- I think there's some, like, dissolution of responsibility there on, like, things that by design or non-design they're trying to make is just an assumption. You know, we must do RLVR and not just we must do it, we must do more and more and more, um, with giving the models, like, the power to do powerful, you know, do anything they want in the middle 'cause that teaches them to be powerful outside of it, and we don't want to limit those things well because it'll make it slightly less powerful on those things.

Um, so like if you assume all of that, they're like, "Oh yeah- That's the logical conclusion... we're heading into a dangerous world, guys." Right. Like, everyone is gonna be doing this, and this, this is the only way to make AI sick. So, um- So basically it's like, it's like i- these are all internally consistent but actually starts from a premise that has alternatives if you- Of course... think about it. Of course.

I, I think there's-- like, like, on the bitterest lesson direction, I think that there's very few people who've, like, made right tasks. You know, like new directions of AI. That is-- or new, new north stars. That is rare. Again, like, I think 2.2 times or something for LLMs itself, like RLHF and then RLCD. Oh. R- RLVR is, like, a.2 in my opinion, and I think that's generous.

Um, but or a.5, or, like, it could be one whole one. I don't really care. Um, but I do think that people are thinking very closed-mindedly about this type of thing, and, um, this-- the, the only people who are at fault here are the researchers because it's definitely not the populace. You know, like, they just assume that OpenAI and Anthropic are just doing the best they can, and they are not the experts who are aware of the true optionality available. Yeah. Uh, and that's fair. And, and like- Yeah...

you're, you're also doing your part in waking them up. Yes. Yeah, well, I, I'm doing my best. But, like, my goal is not, like, convince labs that there's, like, other directions- Yeah... to, to go down. My goal is have, you know, it's like spark hope in software engineers to start, like, actually automating things they've always wanted automated.

Um, I had this, like, article that I wrote that my team didn't let me write, uh, they didn't let me publish, about, like, the, the, the future I want of AI. And, like, there's, like, a lot of, like, little things. Like, remember do what I mean? Imagine if everything could do what I mean, 'cause, like, that, that, all-- oh, that demo was do what I mean. Mm.

Like, like you, like, like there's levels- Don't-- Yeah, don't do what I say. Hmm? Yeah. Don't do what I say, do what I mean. Yeah. And, like, we couldn't do what I mean yet because, like, computers are so basic and literal. But that computer use one was just that. And I think that there's, like, levels of smoothness that'll happen in the world that people just don't understand, and, like, the promise- [laughs]...

of, like, smarts all around are-- it's, it, it's, it, it-- I don't wanna overpromise. Uh, I don't think it's going to happen right now, but, like, we are gonna do whatever the fuck we can to make that happen. Yeah. Um, any other things on the sort of general shape of post-training? You know, um, you obviously, you have been very intimately involved. Uh, mid-training, is that, uh, something that you do have comments on?

I don't think we've ever talked about it. Mid-training. Um, I mean, it's all a spectrum. Yeah. Right? Like, am I- This is a curriculum, but, like, fancier. Yeah. [laughs] I mean, like, it's, it's, it's like, you know, it's a cost-saving thing. Yeah. 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, and, you know, looking at the data, this thing that, uh, our data team is so good at that I'm not, um- Hmm... uh, 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-run 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? [laughs] 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- Power to the people... my thing. Yeah. Yes. So m- the thing I've said i- 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. Hmm. So, um- Anything except pre-training Mm, mm. Yeah. Uh, amazing.

Uh, 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, in fur- further, right?

So like, um, uh, one way to put this is that OpenAI was trending in the direction of the omni model. Mm-hmm. Right? Um, 4o was, was one of those. Uh, then for, for, for a brief period of time, there was always like, there was like a kind of a main branch of like this is the chat tune model, and this is the coding tune model.

Those are t- completely different things. Those are extremely different concepts. I will, like, break that down a little bit. So multimodality is a little bit different- Mm... um, because sometimes the other modalities help, sometimes they hurt. Yes. 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. Mm. 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, 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, there, 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. Mm. 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- You're absolutely right. And- Yeah... you know. Sycophancy, psychophancy- Yeah... whatever word- Yeah, yeah... how- however how to pronounce that, overconfidence, hallucination.

Uh, like even the kind of style that excels in LM arena, bold, italicized, emojis. You know, like it doesn't answer the question simply. It gives like a long writeup, and then it asks you a follow-up question, so it feels more like a human talking to you. All of these things, um, come because strings are super weird. You know?

They are like weird-ass things, and you need to be miscalibrated. You need to like mode drop. You need to be hyper-confident in order to not go off the rails because the, the reward model will punish you so hard when it happens because it's obvious. Y- um, and then this, like, warps the probability space entirely, and it interacts with that of the reasoning models, right?

Because, like, it, you know, the, the models are like these sin- simple linear things that tend to cheat a bit. So I think that that's very different than exposing intelligence is, is my guess. And a lot of the art to intelligence is studying the subtlety that I think that, um, at least when I was in OpenAI, people were not really studying that because, like, they were just like, "Chat, chat, chat, chat," just like people are on with Jeff right now.

Yeah, you give me an ultima- you know, you give me an objective, I will just go optimize for that, right? Yes. Like, and it, it- But, but if you try, and there-- And, you know, like the saying is like you could have like two objectives, and you could just like optimize for both. But then that is literally the act of fracturing, right? Um, so yeah.

So, I mean, in some ways you ha- you are also factoring intelligence into system one, system two, but you just don't agree with the other people's fractur- fracturing, which is fine. Well, it, it- Which is fine... it's a little different. No, no, no. [laughs] If I could, if I could add-- If I could defend- Yeah... um, the system two tasks, um, number one, like we don't toss out the system two tasks, right?

Like you can try to make Jeff work on it, and there actually is an intelligent answer for that, which is unknown. You know, like my, like there, like there is better and worse behavior in the system two tasks, which would be like really low confidence, lots of uncertainty. Maybe some heuristics can like move the needle here and there. But we care about them too, just to be clear. I just think that that is not what the, what is the intelligence is native to.

So we're not trying to fracture anything like that. Um, and all fracturing makes the model dumb. You know, like if people, like get the model to say like it is OpenAI or Quinn or, uh, d- t- you know, like Claude or whatever else, I don't really know what it says this, these days. I am not going to put into the models that you are Jeff from Typesafe. That fractures it, right? Like, it, it's, like, like I, I don't want that. Like represent what do the internet thinks, right? Like be correct.

That is what I want because that's how you get the smooth, predictable intelligence. I, I mean, identity is a thing, I guess, that- And for an API-... it may be somewhat of a special- For a, for a first party product, yes. Yeah. But, like, for an API, I don't think so. Yeah. Okay. You know? Yeah, that's good.

Like, I, I don't, I-- Like people don't want, if they're making a chatbot with, uh, you know, ChatGPT, they don't want it to say it's ChatGPT. They want to say it's like Chipotlei or whatever, right? Well, you know, so the, the way that you also have to make up for it is you have the skill, right? Yeah. The, the, the Jeff skill. Yeah. Which, which is for coding agents to- Yeah, they do... work with Jeff.

Um, okay, a couple closing questions- Hell yeah... because I, 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? Um, with the company- Yeah... I think that this is like more like a four-year journey. Yeah. But- Well, I- Yeah...

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 the, everyone's on holiday. I'm gonna take all the OpenAI GPUs and g- go do this thing." Yeah. [laughs] That was a good time.

And that was like the, the pre-Typesafe- Yeah, yeah, yeah... moment, right? I-- That might have been... Was that when the coup was happening? I don't really know. Yes, actually. [laughs] Yeah, yeah, yeah. That, that sounds right. Yeah. I remember. Oh, my God. I don't wanna- [laughs] 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- The, the coup was annoying or the, the run was annoying? The, the coup was annoying. The coup. Okay. Yeah, yeah, yeah. Yeah. Uh, it, uh, I will- Safetyists took over the company... Yeah. Anyway. Maybe next time we chat- Okay. All right. All right... I'll, I'll dump tea about- Tea... a, a tea about the coup. Um, yeah. It actually, this problem was one that, like, was in my mind since before ChatGPT even launched.

I was like, "Holy shit, the ChatGPT team is cooking. They are doing the right task. They are doing the thing that AI researchers are bad at, but successful product people are good at, which is giving a lot of fucks about the experience." You know, it's, it's, it's very rare. They, like, there's very few people like that at OpenAI. Um, and those guys were cooking on it really, really well. And to be clear, this is the whole journey from GPT-3 to 3.5, which included AI Dungeon- Yeah...

which you've talked about as like, yeah, well, that's, that's an example of a, a use case that we never predicted. Yes, exactly. That's right. Yeah. Well, uh, oh, yeah, that is, uh... Also, I had fought very, very hard to deploy InstructGPT. Um, like actually, the early versions of it were even trained with, like, an algorithm we didn't publish that I made myself because it was too slow to clean the, the PPO data, and I was like, "Fuck it.

This is so fucking good. We need to get it in the hands of users." Mm. And, like, basically, immediately, it took 50% of the market share of LLMs at the time, and, but-- And we thought it- I, I made-- I went through great effort to make sure everything in our launch video is true. Um, you know, we-- I truly was thinking, like, "Is this AGI because it's superhuman at instruction in, instruction out?"

Y- y- obviously, it's not, but, like, everyone, I think, should have an answer to why that was not AGI, 'cause it looks very smart. And my answer to that ended up, uh, you know, like, ended up only being used for copywriting. You know, Jasper AI, Copy.ai, like, writing, like, you know, what is now called slop on web pages.

Um, and we were worried we made the internet a worse place, right? And I went back to the drawing board, and I was like, "What's missing? We are smart, clearly. Something is missing from it, like, creating value. What is it?" Like, I actually was doing more philosophy at the time of, uh, like, you know, like, "What is going on?" And the answer was, oh, machines.

You know, the question I asked myself is like, let's work backwards from an AI-based economic revolution. When that happens, what will c- be... wh- what'll be calling the AI if AI is an API? Will it be humans, or it'll be code? And I figured it was many nines of code, and, but, like, all the optimization was going into the humans part. And then it clicked for me, and like, holy shit, this is the North Star.

I think, like, I wrote a document. I was, like, talking to Sam about this. Sam was like, "This is so fucking good. You should go work on it." And we're like, "Yeah, yeah, yeah, Sam, I have a job." You know, like, I'm, you know, I, I was working on- Sam just told you to do it. Do, go do it. Like [laughs] Yeah. But, like, my, my guess at the time is, like, this is super obvious. Like, it's so unbelievably obvious.

Anthropic must be working on this already. You know? And, like, we're already cooked, and, like, actually, OpenAI does better at, like, catching up than it does at, like, actually innovating. So, like, uh, uh, ChatGPT was a copy of Claude, right? Um, like, they had an internal thing. They- Yes... just didn't ship it. Yeah. Um- Claude, Claude in Slack.

But, uh, but, you know, reasoning, I, I would say first-ish. Yeah. Yeah. But debatable how good of a product that is. Yeah. Um, great research, though. Super great research. I'm just not sure if people had that product need. Um, and, you know, Claude did the coding agent stuff too. So, Sam says that, and, you know, I just go back to my job for a while.

Eventually, like, you know, the instruction following team just says, "We won. We, we've solved instruction following. We don't need to do stuff anymore." I'm, like, trying to think about what I do next. I was like, "You know, maybe I'll just, like, start playing around with this." Um, I, you know, do more philosophy and design and thinking. I thought it would end up taking a week, uh, when I started training models.

It ended up taking, um, many years. At, at some point, I was like, "Holy shit, you know, there's signs of life here. This..." It, it obviously didn't work, right? Otherwise, we would have deployed it. But, like, I wanna explore what it would be like research-wise to go all in on this. You know, like, I wanna really see, like, like, what it would be like if you went, like, absolutely, insanely all in in this direction.

Uh, because of what I said, you know, like, if an AI winter happened, would I-- how would I feel? I would consider myself personally responsible. I talked to other companies at the time, and I was like, "Hey, I want to start a lab on this direction." And, you know, like, there was interest. And I just talked to them like, "How fast-- What would be faster?

This or a startup?" And they're like, "Startup." And I'm like, "Fuck it, man. We ball." Yeah. "I guess we're doing some crazy shit." [laughs] And- And you called Eric and Sacha and- Yeah. Well- Yeah... I, I call Eric first. Um, with Sacha, I actually didn't try to recruit her. I, I tried to be good, and I was just like- Mm... "Hey, am I crazy? Is something missing here?

You know, isn't there, like, uh, l- like, like, am I too much in the OpenAI bubble that I didn't realize there must be a solution to this?" And then Sacha was like, "I'm in." And I'm like, "Sacha, you're working at a startup." And she's like, "I'm folding it right now." And I'm like, "Do you wanna think about that?" She's like, "Oh, yeah, good point. Let me think about it."

[both laughing] And then she joined. Yeah. And then, you know, within two weeks, we had funding. We, we had, like, people move into my apartment. It was the worst 'cause I'm a neat freak. Um, and we just kept on cooking, and eventually, we got the research that, that, that showed the signs of life. Yeah. You know? It was, it was a crazy time.

So the, the, the qu- the question is, that was all long context. Oh, yeah. And then another question is, someone like you- Yeah... 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?

Should they do it? Ooh, that's a fascinating question. Um... Ooh, man, how do I do this without burning bridges? [laughs] 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 because... Like, number one, I don't really value researchers. Um, I value people who, like, like, look at my Bitters lesson, right? The data, the task. I, I, I want... Well, not just that. Yeah. 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, like, 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 a 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 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- Hive mind. Yeah. Yeah, exactly. 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- Solvable...

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 think, like, all of these, like the, like, really unearthing technology I think is, like, just good. 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 Open AI folks I was talking to, SpaceX as well, by the way. Oh.

Uh, is that, uh, it is, this is a political thing much more so than a pure xRisk thing. Yep. Uh, so yeah. Of course. Political positioning is- And, and that, and that's beyond my pay grade. Exactly, yeah. That's well beyond my pay grade. Once they sh- once they told me that, I was like, "I get it. Uh, this is about the 2028, uh, election." [laughs] Oh, no. Yeah. [laughs] Oh, I wish I didn't hear that.

That's such a bad vibe. [laughs] No, no, no, no. And so- This is not the whole company. Yeah. This is just that, that room's discussion. No, no, no, that, that- Yeah... makes sense. Yeah. That, that, that makes me lose faith in humanity a bit, but maybe I'm just a naive technologist. Uh, it's really starting to matter- Yeah...

who's, who's in charge of the governments, uh, that, that will help to regulate, uh, these things as they emerge. And, like, as a lab- I, I totally-... you should probably think that through. No. Uh, no. [laughs] I, I totally agree with that, to be clear. Like, I, I think being opinionated on that matters a lot. I personally am afraid of trying to mislead people because I think that bites people in the ass a lot.

You know, like, um, I think that, like, people trying to be overconfident, like, uh, I, I, I, I, I, I obviously t- I, I'm not actually gonna talk about politics. I think what happened in COVID is, like, people leaned too much in, like, appeals to authority and being overconfident to try to get people to behave in certain ways, and, like, obviously our response was extremely suboptimal.

And that had, like, like, ripples of downstream ramifications that are now, I think, extremely bad for the world. Like, maybe I'm naive. I think that misleading people, even for the greater good or what it- they think is the greater good, is just, uh, it's just, I, I'm not a fan. Yeah. I'd rather, I'd rather not do it.

For what it's worth, I, I, it's not a- Yeah... I don't think it's misleading. It is just, like, this is why now. Yeah. Uh, why, yeah. Like, you know. That, I, I think that if it, if it- Like, Dario said in May, like, how come we're doing it now? I think if that is why now- Yeah... that is a little bit of misleading about, like, the risks versus, like, the, the objective. It, there is, like- Yeah... some level of, like, sneakiness latent in it- Yeah, yeah...

that, um, is worth calling out and I think owning up to. Well, ob- obviously they want to, if, if they want to manipulate, then they shouldn't own up to that. That seems like a bad strategy. No. But, like, that to me is just, just sad for the world. Yeah. Hopefully, I'm never... I've- [laughs] Yeah. Uh, hopefully, like, we are never involved in anything- Yeah... like that. It might be inevitable as we get big.

Um, but I want to, I wanna stay, like, pure technologist to my roots as much as I can. I mean, Jeff for president, why not? Uh, yeah. I, I, I can, I, you know, I would trust Jeff's decisions over, um, my, my own. Um, okay, so, so, uh, less shitposting, uh, more about- [laughs] That's shitposting. No, no, no, no. You're just cr- For, for me. I'm sh- I'm shitposting. Oh, okay.

You're just crushing my hopes about, uh- No, no, no, I'm not calling you shitposting... like America and the world right now. [laughs] Oh, my Lord. Yeah. I mean, like, there's, uh, I, I think I watch too much TV about, like, conspiracies to think about the presidency. Oh, no. Um, the, uh, you have chosen your North Star.

You have chosen reliability, y- and programmable and composable AI. And cheap. Uh, and cheap. Yeah. 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? Ooh. Like, uh, just basically give people tasks.

Give people tasks? Yeah, like that your tasks- Oh, there's so many I want. Oh, what? You have picked your tasks, right? You know what I mean? What? Wait, I-- that's such a good question. Holy crap. Oh man, I'm so excited by that. 'Cause you're, you're gonna be, you're b-- f-- so the next, like fifty years you're gonna be busy doing your thing. Hell yeah.

Okay, 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 Allie's Doom demo wh-where like you can like get NPCs to control stuff, like, you know, l-like that was just really like the- Yeah... 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 you could make like such a compelling world. Oh, man. Mm. Um, and man, a little sad that I can't work on these types of things. Yeah. My, my life path is a little bit set right now, and I'm, um, yeah. Yeah, but you can call someone else to work on it. Yeah, yeah. That's cool. And then, and then you can like- Yeah... feedback on it.

And, um, the thing that I would really, really like to explore is like coding agents free from the tyranny of the KV cache. L-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. Tha-th-that's why I wrote the article, "KV Cache Rules Everything Around Me." Mm.

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- Okay... 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- Right... works and everything. And, um, I think, uh, oh, um, yeah, um, like, uh, it explains a lot of stuff, like why routing is really hard, why sub-agents don't seem to work, like why compaction is such a hard problem. And I, I'm going to try to release a document.

My team might veto me because believe it or not, I'm not in charge. [laughs] Um, you know, but, uh, I, I wish. Um, but, um, I want to release a document of like here are my thoughts. Please play with it, and please figure out all the ways that we can do things with coding agents, like once you're freed from that, you know, that, that KV cache tyranny.

Which is it locks you in, and- Well, not-- it lo- it, it locks you in into one model, right? And in order to do it efficiently, you need to like keep on appending to it. Mm-hmm. So now you're not doing best software practices like state management, abstraction, decomposition. Why can't you give an easier task some... Y-y-- why, why can't you give a sub-agent an easier task?

Because of the state that you're passing around. Oh, I touched this. Uh, because of the state you're passing around. You nee- would need intelligence that is way cheaper than the intelligence using to read this in order to pass this state around. Why can't you be smart about it, right? And I think there's just like tons of really cool, fun research to be had there on like different programming patterns. You know, kind of like how people are playing around like with like recursive language models.

Like I feel like there's like just lots of cool stuff in here when you think about like, oh, I want to explicitly label the state of everything. Or imagine you have like a sub-task. Like coding agents, I, I think it's fair to say they work on sub-tasks at a time as from a decomposition perspective. Why do you need to pass all of that state back into the parent task? Yeah.

Why couldn't you do smart things about it? A-and also, if you had a hierarchy of labeled sub-tasks, why can't you do a search through that sub-task tree for the relevant context when you need it in, right? And then, you know, another thing that you can do... Oh, man, I forgot to write something about this. Um, uh, the, I have like some cooks in here that are really, really cool. Um, hope to publish it.

Jev: System One models for Prod, not God — with Diogo Almeida, CEO, TypeSafe AI

Kapitelklicken springt direkt hin

Shownotes

Transkriptvom Podcast bereitgestellt · Zeitmarken springen hin