FOR EDUCATIONAL AND KNOWLEDGE SHARING PURPOSES ONLY. NOT-FOR-PROFIT. SEE COPYRIGHT DISCLAIMER.

“I just see stuff, say stuff, run stuff, and copy-paste stuff, and it mostly works,” Andrej Karpathy—co-founder of OpenAI, former AI director at Tesla

Breaking News:

Andrej Karpathy recently coined the term “vibe coding” to describe how LLMs are getting so good that devs can simply “give in to the vibes, embrace exponentials, and forget that the code even exists.” We dive into this new way of programming and what it means for builders in the age of AI. Apply to Y Combinator: https://ycombinator.com/apply Chapters (Powered by https://bit.ly/chapterme-yc) – 0:00 Intro 0:42 What is vibe coding? 1:00 What founders in the current YC batch are saying 4:35 Debugging and building systems 6:59 The models people are using now 10:01 What percentage of code is being written by LLM’s? 11:58 What changed and what stayed the same? 18:08 How Triplebyte did candidate assessments and how would that change in this era 21:37 Key skills that will remain relevant 23:01 How do you develop taste without classical training? 30:59 Outro

Intro was like somebody dropped some like giant beanock seeds at night we woke up in the morning going on I mean I think our sense right now is this isn’t a fat this isn’t going away this is actually the dominant way to code and if you’re not doing it like you might just be left [Music] behind welcome back to another episode of the light con I’m Gary this is Jared Harge and Diana and we’re Partners at y combinator collectively we funded companies worth hundreds of billions of dollars right when it was just an idea and a few people so today we’re talking about Vibe coding which is from a Andre What is vibe coding? karpathy post that went viral recently there’s a new kind of coding I call Vibe coding where you fully Give In To The Vibes Embrace exponentials and forget that the code even exists y so we we What founders in the current YC batch are saying surveyed the founders in the current YC batch um to get their take on Vibe coding and we know we essentially ask him a bunch of questions we asked them what tools are you using um what work how has your workflows changed and soort of generally where do you think the um future of software engineering is going and how will the role of software engineer change as we get into a world of vibe coding and we got like some pretty interesting responses anyone have any favorite quotes that jumped out from the founders I think one of them that I can read B verbatim is I think the role of software engineer will transition to product engineer human taste is now more important than ever as Coen tools make everyone a 10x engineer that’s from the founder of Outlet I got one um Obby from asra said I don’t write code much I just think and review this is like a super technical founder whose like last company was also a Dev tools company he’s like extremely able to code and so it’s fascinating to have people like that saying things like this there’s another qu from a different RB RB from copycat who said he’s I am far less attached to my code now so my decisions on whether we decide to scrap or refactor code or less biased um since I can code three times as fast it’s easy for me to scrap and rewrite if I need to and then I guess the really cool thing about this stuff is it actually paralyzes really well so yoav from cix he says I write everything with cursor sometimes I even have two windows of cursor open in parallel and I prompt them on two different features which makes sense why not three you know do a lot actually and I think another one that’s great is uh from the founder of a train Loop mentions how coding has changed six to one months ago 10x speed up one month ago to now is 100x speed up exponential acceleration and he says I’m no longer an engineer I’m a product person yeah that’s super interesting I I think like that might be something that’s happening broadly you know um it really ends up being two different roles you need I mean it actually maps to how Engineers sort of self assigned today in that either you’re uh you know front end or back end and then back end ends up being about actually infrastructure and then front end is so much more actually being a PM you’re sort of um almost being like an ethnographer going into the Obscure underserved parts of the pie of GDP and you’re trying to extract out like this is what those people in that GDP Pi actually want and then I’m gonna you know turn that into code and then actually evals are the most important part of that when I was running triple bite this was actually one of the things we noticed it was almost as important as a technical assessment of Engineers when trying to figure out who’s a good match for a specific company is there’s a certain threshold of techn ability you need but beyond that it was do you actually want to talk to users or not like some Engineers are actually just very a lot more motivated by working on things where they know who the users are and they get to like communicate with them and they get live feedback and they can iterate um and essentially being a product engineer and other Engineers really don’t want to do that at all like they find it annoying having to deal with users and they want to just like work on like hard technical problems and refactor back end engineer yeah that’s that’s what we call a back end engineer yeah sure and that’s only a theme that came up in these survey responses right this idea of sort of the llms are maybe going to push people to choose do the actual writing of the code may become less important it’s about are you really do you have taste and you want to to solve product problems or are you an architect and you want to solve systems problems oh and interestingly I guess um one thing the survey did indicate is that this stuff is terrible at debugging Debugging and building systems yeah and so you still the humans have to do the debugging still they have to figure out well what is the code actually doing here’s a bug where’s you know spot the bug where’s the code path that you know we have some you know logic error you know just didn’t figure this out right there doesn’t seem to be a way to just tell it debug you were saying that you have to uh be very explicit like as if giving instructions to a firsttime software engineer you have to really spoon feed at the instructions to get it to to debug stuff or you can kind of Embrace The Vibes I’d say Andre karpathy style is just like ignore the bug and just like roll just like just like tell it to try again from scratch like it’s it’s wild how your coding style changes when actually writing the code becomes a THX cheaper like as a human you would never just like Blow Away something that you’d worked on for a long time and rewrite it from scratch because you had a bug You’ you like you’d always fix the bug but like for the llm if you can just like rewrite a thousand lines of code in like six seconds like why not that’s kind of like um writing you know taking the approach of uh you know how people use um Mid Journey or playground when you’re trying to generate images like if there artifacts or things that I don’t like sometimes I don’t even change the prompt I just click roll and I do that five times and you sometimes it just works I’m like oh I can use that now which is very different frame of Building Systems because you’re not building foundationally step by step you’re really doing it from scratch because fundamentally what’s going on is like all these tools today are coming from the world of a generated code that are in this latent space hidden somewhere and you have to do it from scratch to find like a different gradient and not get stuck and then you want to like add a bit of randomness get it to regenerate but I do think maybe I don’t know whatever next generation of 05 maybe we’ll get to the point that actually is able to build upon I mean as of right now I think most of it is you need to reroll and rewrite but it doesn’t build upon it yet but we haven’t seen any of the coding tools right now work well with reasoning I think we have well 03 is infinitely better at debugging than 3.5 onet so like it definitely feels like we’re headed in the direction where this may not be true in you know six months the next time we do this episode Diana do you want to talk about like the models that people are using the idees The models people are using now that the people are using there’s some really interesting Trends there yeah I think as we mentioned a couple episodes ago we already saw this in uh the shift started happening the Vive started to shift back in summer 24 when cursor was being used by a big portion of the batch and now by far is the leader but the other thing that’s happening this is a very fast and moving environment wind serve is a fast follower it’s starting to be a very good product as opposed to to cursor and I think Jared you have some first experience with why wind surf is like better than cursor yeah I think the number one reason that people are switching is that cursor today largely needs to be told what files to look at in your codebase so you have a large code base you can tell what to do but you have to tell it like where to look in the code base wind surf indexes your whole code base and is pretty good at figuring out what files to look at on its own there’s there’s other differences too but I think I think that’s the most important one at the moment notable Devon does get mentioned but the the drawback of Devon not really being used for serious features is that it doesn’t really understand the code base it’s being used mostly for small features and barely it’s like barely mentioned the other one people still use uh chat jbt and the reason they use it is because they want to actually use the reasoning models so it do get posted people post some of the debugging questions to to figure out the use the more powerful models for reasoning because right now cursor with serve are still in the old world I mean old world less than six months ago of pre- reasoning models not in the test time compute so Founders are using that and uh there’s some Founders some of them are self-hosted as well self-hosting models because maybe they have more critical sensitive IP they do that and now talking about the shifts in terms of uh models the thing about Coen the big game in town that we saw uh six months ago was CLA Sonet 3.5 is still actually a big tender most are still using it but 01 01 Pro and 03 meaning all these uh reason models are starting to see a it’s almost like getting neck to neck now close with a Sonet 3.5 the other one is 40 virtually no use for Coan and the other interesting thing is uh deep seek R1 is getting mentioned is uh beenin use it’s like a viable Contender as well and Gemini not really mentioned the the the one thing I’ve heard from Gemini is because it has like the longest context window I’ve heard from a couple of Founders that they do use it and the way that they use it is they put their entire code base into the Gemini context window and they just like tell it to like fix a bug and it doesn’t always work but like sometimes it can just like onshot fixed up because it is the whole thing in the context window it will be interesting to see as people get more adoption on the newly released reasoning models with flashback 2.0 uh I don’t think people have tried it yet but the long context window plus reasoning could be a good Contender what is the estimated code that’s being written by llms in the current Badger this is pretty crazy so we we explicitly What percentage of code is being written by LLM’s? asked this question what percent of your codebase do you estimate is AI generated the way I interpret the question is like like of the actual characters in your code base not including any libraries that you imported like how percentage of like the characters were like typed by human hands versus like emitted by an llm and the crazy thing is one quarter of the founders said that more than 95% % of their code base was AI generated wow which is like an insane statistic and it’s not like we funded a bunch of non-technical Founders like every one of these people is like highly tactical completely capable of building their own product from scratch a year ago they would have built their own product from scratch but now 95% of it is built by an AI except for you know maybe it sounds like we have one or two examples of people who um they’re so young that they learned to code in the last two years so they actually don’t know a world where cursor didn’t exist yeah this is um one of my best companies this batch actually is exactly this the founders are extremely techical minds but they’re not classically trained in computer science and programming and they are incredibly productive and able to produce just a ton of like really amazing product and AI is writing almost the entire thing it kind of makes me think a lot of uh the discourse around sort of J these are the first digital native that grew up with the internet this is like the generation that grew up with Native AI coding tools that they skip the classical training of a software engineer and they just do it with with the Vibes but they are actually very technical minded I mean they have degrees in math and physics physics yeah so they they they have that raw let’s call it more like system thinking type of mind that you still need maybe we should talk a bit about that it’s like what’s still the same and what has changed I think this Vibe coding will enable people who have those What changed and what stayed the same? kinds of tactical Minds who come from other tactical disciplines like math and physics to become highly productive as programmers much faster than it was in the past like I remember they were like coding boot camps like back in the day they would try to like retrain physics people into programmers and like it didn’t work that well because it just takes too long to learn all of the syntax and all of the libraries and all the stuff that you have to know to be really productive but like now now it’s a new world the boot camps are also very specifically focused on getting you hired at companies and then I think there was it was during this around like 2015 era where just companies themselves were rethinking how to evaluate software engineers in their hiring processes and it was moving there was a real shift away from like we want to hire classically trained computer scientists um whiteboard algorithmic problems towards we actually want people who are just really productive and write code quickly and some some of these arguments are like Evergreen at right like I remember when rails first came out there was just like a real sense of oh like I don’t know like active record as a way to like interact with your database was seen as a great abstraction but like there was still this same flavor of argument right like oh know like if you don’t really understand the internals like you’re just going to write like crappy low performing web software how do you feel those arguments have aged if you look back on it now my feeling is that many of the most successful companies I would say stripe gusto so are just two that really spring to my mind as ones that really heavily leaned into the actually we just want people who are really productive with the tools and we’re going to change our whole hiring process to just select for people who are good at like the interview shifted from teach us how you think to you’ve got three hours on a laptop and you need to build a to-do list app and build it as quickly as you can and those companies have had a tremendous amount of success it is it does seem like at some point as they grew and they scaled then the bottleneck did actually become having people who were classically train and systems thinkers to sort of scale up and architect things it does seem like uh how people are hiring Engineers is changing but maybe not changing fast enough yet the results of the survey are relatively surprising to the four of us here probably pretty shocking out there even it’s just like this thing that popped up in our backyard only in the last 6 to nine months my guess would be engineering hiring period has not actually caught up to this people are still standing at whiteboards and doing that kind of thing as opposed to what can you get done and so it sounds like the stripes of the world they were ahead of the game and everyone has to hire Engineers this way now I mean I I wonder if actually even that’s going to be sort of the old metod I mean something that stood out from the survey responses was this idea of um two themes we talked about right like one is how okay we’re all just product people now like actually the thing that you need is really great taste and understand what to build and the second was of um actually now what’s really valuable is to be like a a systems thinker and an architect and to really understand the bigger picture in which case actually like maybe being a really productive coder because that’s definitely something that always fit my definition of when you’re talking about who are great Engineers you know and like that one of the dimensions that they’re just really like they can write code really fast like maybe that’s outdated like maybe that’s not if the llms are actually really good at writing code quickly your roll andite from scatch debug theills might be completely different problem is there’s like two different stages there’s zero to one which in which case speed is the only thing that matters and then to your point about active record and rails that battle was actually fought to a standstill because of course using active record or rails allowed you go to go from zero to one very quickly but then what happened to Twitter it became the fail whale right like basically once get to one like one you know that architecture will not get you to a billion or 10 billion or hundred billion dollar in valuation or users or whatever like it’s just not going to actually work so I think you’re going to see a the same thing and then that there’s Nuance in what you just said right like getting zero to one quickly and then being able to scale to a billion users are to two totally different sets and then I think that that might be Irreplaceable for now around people and one of the things I discovered as we were scaling one of the biggest rail sites I mean you scaled one of the ra biggest rail sites too is that um there aren’t that many people who have to do it getting to one is so rare yeah that’s true that uh did you reach this point where how you got to zero to one very quickly was like you use lots and lots of Open Source and then at some point like maybe two years into this maybe a year and a half into the startup even we could not use random gems anymore gems anymore cuz they they were just never designed for companies at or scale and so we have to and it it just fall over right this very good examples of what you’re saying Gary I think maybe to summarize a bit more zero to one will be great for Vibe coding where Founders can ship features very quickly but once they hit product Market fit they’re still going to have a lot of really hardcore systems engineering where you need to get from the one to n and you need to hire very different kinds of people and to that I think there’s very good historical example like Facebook as well I mean they got yeah they got away with PHP which personally is a terrible language yeah maybe I get flame but I think it’s a bad language I’m with you I never liked it sorry guys it was very bad but you could ship things very quickly but at some point it became such a big bottleneck for them to ship features that they had to hire the hardcore system people to build a custom compiler hip hop yeah that that it would run fast on bare metal because just too expensive to replace all the code yeah and that kind of people that did that are not the vibe code people are sort of this hardcore systems people that based on our survey current tools are not good at that lowlevel systems engineering har I’m not How Triplebyte did candidate assessments and how would that change in this era sure everybody who’s listening knows what triple bite is but it’s actually very relevant you do you want to just describe for everybody but yeah trip is the company I started in 2015 and our we were essentially building a technical assessment for engineers like our goal was how can you use software to automate evaluating software engineers and the way we did it was pre- all these Cen models which is we buil all of our own custom software to interview Engineers have humans interview engineers and then essentially just like label the data like in and and interview them by like asking them to write code they highly techn interview yeah it was high yes like it was aing to write code um we did actually include algorithmic problems and is it true that you and your co-founders have done more technical interviews than any other people on the planet I think so like in terms of just like pure hours cuz it was that was just like the early days of it were like all every day just thousand thousands and thousands of them right um and then we scaled up and we had like a team of about like 100 Engineers contracted just that that we would pay per interview completed and so you’re exactly the right person to ask this question you’ve literally spent more time thinking about this than anyone else on the planet um if you were starting triple bite again today and you had to design like new Tactical assessments for engineers what would you have them do the big takeway I had with triple bite and this screen in that in particular is just people want different things and so you kind of need to know up front like what exactly it is that you’re evaluating for and then design your technical screen um around that it’s kind of what I’m getting at with where stripe and Gusto and these companies just knew that they didn’t care if someone had fundamental CS knowledge so it didn’t make sense to screen them on that like they they wanted to screen for the thing that they were actually going to do in their job and then our product was more trying to screen for everything company trying to get a taste of everything companies might want and then figure out what someone’s Max skill was um and then send the people with the max skill to the companies that would value that Max skill and in today’s world I think I would actually have a screen that at least accounted for just how well people knew how to use these tools like and so again it’s maybe contradicting what I’m saying earlier but it might be the case of maybe how product how quickly you can code and can build product is actually something to explicitly screen on um just at the bar much higher you probably have to ask different questions cuz I’ll bet if you go back to the original triple bite assessment I bet a lot of those questions you could literally just copy and paste the question into chat GPT and it would spit out a perfect answer in which case you’re not really proving that much competence if you’re just copy like the the questions probably have to be like 100 times hard I mean this gets you the deepest stuff right not necessarily because if you have someone else mon it depends on like what conditions you’re going to put on the screen which are things interesting so I know classic question was like build tic tac toe um yes of course if you do that unsupervised and you just let someone like come back with their Tic Tac Toe Solution that’s going to take like two seconds right um um if you want to watch them coded and force them to not use an llm it I guess that’s a question do you force them to code it without an llm with like the old questions or do you let them use an llm and now you need new questions because the other the old ones became trivial that I think is what everyone hiring software Engineers right now should be thinking about and trying to figure out yeah I’m not sure I know what the correct answer is to that yeah I Key skills that will remain relevant think there going to be probably you’re going to test for different things because I also did a lot of engineering hiring I think one key skill that’s going to I think remain that’s constant I do think skills of reading code and debugging are maximum is like you have to have the taste and enough training to know that the llm is spitting bad stuff or good stuff so like bad code or good and I think you can see it clearly sometimes if a candidate is using the tools and there’s actually a reasonable solution that the llm outputs and then the candid is like oh this is actually bad that is a sign so I think knowing kind of more the high level thinking to know what is good versus bad in order to do good VI coding you still need to have the taste and you still need that kind of classical train maybe not necessarily classical train but enough knowledge to judge what’s good versus bad and you only become good with enough practice I think that will be one that will be constant that would be my opinion yeah that’s interesting just coder it’s like more like code review as the interview versus like actually like producing code yeah I I mean you could have a some form of system design you want to know how good they can put a product out there so then is testing for taste so we’re going to test for debugging and then taste but then how do you get to I guess this is a question going to these kits that you have that we call it AI coding natives yeah how do you develop taste when you don’t come from a How do you develop taste without classical training? classically train World which would be interesting for next gener well you have to because if you don’t the startup dies right so let’s say this founder they go off they have 95% written by AI the proof is in a year out two years out um they you know have a 100 million users on that thing you know does it fall over or not and then one of the things that’s pretty clear is these systems uh you know in the first realm the first versions of reasoning models they’re not that good at debugging so you actually would need to descend down into the depths of what’s actually happening yeah and if you can’t then you got you I mean let’s hope that they can go find another architect they’re going to have to hire someone who can I think there going to be a generation of softw Engineers that are like good enough because it’s so easy to retool there with with all these coach tools like the barrier ENT is so low you’re going to be good enough Engineers there’s going to be tons of those but to be exceptional like the top 1% I think you’re going to need to get into deliberate practice I mean the analogy we’re talking about is uh Mal and Gladwell popularize this concept of 10,000 hours of practice to become an expert which came from this research from uh what was his name the uh Anders Ericson Anderson Ericson right which it wasn’t just the research was very specific it was about uh how do you find World Class violinist and it wasn’t about just putting the time but delivered practice it’s like hours that actually plan and thought and is hard work you could become an expert with less hours so I think what’s happening with now Coen tools is that is very cheap to put in the hours because the output is just so quickly you can get to good enough but to become the best in the world and the best founder you’re going to need that delivered practice to go into the details and you’re G to have to peel the onion and understand the systems and get to again to some extent being classically trained I mean a good example is maybe we go back to history is like Picasso one of the greatest painters it was amazing at drawing lifelike pictures which is not what he’s famous for of course when when you imagine a Picasso you imagine the opposite of that yeah there’s this famous sequence of drawings on how he got to a abstract Bowl it starts from being lifelike to iterations until he gets to the essent to kind of the abstract art that he’s very well known for but he could could only get to be the best in the world because he was actually a very good painter and classically trained and could draw super well but that’s not what he’s known for so I do think we’ll see these two classes of Engineers you’ll still have like a very fat class of like good enough you need Engineers for those but the best in the world the founders that become outliers are going to need to put in the deliberate practice yes and no I mean I think uh there are lots of really amazing examples of um great systems level worldclass Engineers who ended up being CEO and CEO of the biggest public companies in the world I think of Max Levin I think of Toby luky from Shopify I mean these are people who just like actually that great um and the thing is there are lots of other people who were not that great but also still CEO or co-founders of companies and then it kind of goes back to to link up what we were saying earlier it goes back to hiring I mean I I keep thinking about the Twitter analogy that you brought up CU I think it’s a really interesting one like if you if you compare Facebook and Twitter in both cases they went very quickly from zero to one in sort of Scrappy moast break things way um Facebook was able to solve the scaling technical challenges in a pretty impressive way I think most people would would agree I mean Mark Zuckerberg was by far way more technical way more in the weeds probably maybe but I I don’t know like I think Twitter scalability challenges were also harder based on the usage patterns like the thing about the usage of Facebook is that it grows like it’s pretty smooth throughout the day like people just use it all the time the problem with Twitter is that the usage is incredibly spiky you get like a super bowl or like a you know like like a world event and all of a sudden you have like 10 times as much usage the way the fan out of the feed works is I think like fundamentally a very difficult computer science problem okay that’s fair though I also think that they were like really hamstrung by their tools do you remember using this terrible Q system called Starling AB I used it because I thought oh Twitter’s so much bigger than us they’re so smart they wouldn’t use something that’s crap no they totally use crap and then I use crap and I couldn’t make it work it was like it was dropping jobs on the floor like it’s like all these crazy bugs happened and then I was like finally I was like I’m not using that anymore I have to switch to rabbit mq or whatever the heck the the the actually correct thing to use was yeah and and like Ruby is an incredibly slow language even like 10x slower than PHP which was already too slow so I don’t know I mean basically you should be so lucky to get to one yeah is there an advantage for a technical founder to be classically trained and be a really deep systems thinker well I mean you just I mean a Toby or a Max levchin is not going to get bullshitted by people from stripe is the same I mean I’ll tell you a crazy story uh when I was uh at palent here you know I sort of burnt out there after a couple years after I design the logo and then I actually between that and going to start my YC startup I spent 6 months as a interaction designer and I was it was this like terrible Venture backed company that ended up going in the ground and it was like credit card software it was the worst I spent 6 months building uh like basically just interaction designs which was really fun that’s what allowed me to work on my startup in my spare time because I had a lot of spare time but I remember designing um you know this faceted search thing for like rental cars or something like that and then I go into my meeting with my Dev manager and Engineers who were going to implement it and they were like oh yeah it can’t be done we can’t do it that way oh like and I was like what are you talking about just make the indexes like this and they were like whoa what do you mean like and then they looked up my resum hear that going that from the interaction designer yeah basically they’re like how did you know that and I’m like you lied to me and that’s the thing like what Founders and like you know when when you’re hiring people what like that that was like the wildest thing to me but it was like sort of the ultimate lesson where it’s like when you’re in the workplace you sort of assume that the people you bring on like they’re not going to lie they’re not lazy like we’re all for the goal and the mission right and I like no like people who hire you hire like they totally will lie to you if you cannot tell that they’re lying and then the worst part is like you kind of have to call them on it like you know sometimes you have workplace cultures that are so polite that people are like oh like I’m going to let that pass and then I’m going to talk about them behind their back and it’s like you should fire them the AI agents incidentally will do exactly the same thing they will absolutely like the AI agents will you just like a human employee will if you don’t like if you’re not techical enough to like call them out on their and be like like no like you didn’t make the change that I it goes back to your to your point about like why being classically trained is still helpful like you you have to be able to call out all the people working for you whether they’re human or not being technical enough to be able to do that is is a superpower so just to wrap up basically what’s going on with all these tools giving superpowers to the best engineers and making the bad Engineers also worse is this is a quote from the founder of a train Loop how coding has changed 6 to one month ago 10 XP up now one month ago 100 XP up now it’s exponential acceleration sort of crept up on us actually yeah it was like somebody dropped some like uh giant beanock seeds at night we woke up in the morning what’s going on so I mean I think our sense right now is this isn’t Outro a fad this isn’t going away this is actually the dominant way to code and if you’re not doing it like you might just be left behind this is just here to stay and you know Vibe coding is not a fad it’s time to accelerate so with that we’ll see you guys for the next light cone [Music]

FOR EDUCATIONAL AND KNOWLEDGE SHARING PURPOSES ONLY. NOT-FOR-PROFIT. SEE COPYRIGHT DISCLAIMER.