The Secret AI Prompt Tool Silicon Valley Engineers Are Using

Everyone's talking about vive coding, but the reality is, for most things, vibe coding doesn't work right now. And even the guy who coined the term, Andre Karpathy, he recently posted that he's now trying to provide more context to models because he realized that's what you have to do to get good results back. Welcome to the Next Wave podcast. I'm your host, Nathan Lands, and today I'm going to show you the secret weapon that all the top AI coders are using. You know, everyone's talking about Vibe coding this, Vibe code that, but what they're not telling you is that you can't Vibe code most of anything that's actually important right now. For any important AI coding, you want to give it the proper context to know what it's doing, versus just throwing everything at, which is what Cursor and Windsurf and a lot of these other tools that everyone's talking about do today. I've got the founder of Repo Prompt, Eric Provinger, on here, and he's going to show you how you can use Repo Prompt to take your AI coding to the next level. So let's just jump right in.

Using only 20% of your business data is like dating someone who only texts and emojis. First of all, that's annoying, and second, you're missing a lot of context, but that's how most businesses operate today, using only 20% of their data. Unless you have HubSpot, where all the emails, call logs, and chat messages turn into insights to grow your business. Because all that data makes all the difference. Learn more@HubSpot.com.

Thanks for coming on.

Yeah, yeah, it's nice, you know, finally put a face to it. You know, been chatting for a while and it's cool that you've been using Repo Prompt for a few months now.

Yeah, yeah, I've been telling people about Repo Prompt for, like, the last, you know, probably six months or so. Kind of felt like it's been almost like my, like, AI coding secret weapon, you know, it's like I hear everybody talking about Cursor and now Windsurf, and I do find Cursor useful, but I was like, why is no one talking about Repo Prompt? Because, like, for me, every time I'd get into, like, a complicated project, as soon as the project got a little bit complicated, the code from Cursor would just stop working for me. Like, it would just not know what was going on. You could tell it wasn't, like, managing the context properly, and then when O1 Pro came out, that was when I really noticed Repo Prompt and started using it a lot. You had to go to 01 Pro to really get the best out of AI for coding. @ that point, absolutely. Cursor wouldn't even work with the one Pro. And so Repo Prompt was by far the best. And it was just kind of shocking to me. Like, only, like, a few people on X are talking about this. Most people don't know about.

It. Yeah, I mean, like, it's the only tool that I use to work with AI. And, you know, for a long time it was just Sonnet, and I would feel like I was able to get a lot more out of Sonnet than other tools just because, you know, the full context window was there and, you know, I wasn't bleeding through the nose with API costs using the web chat. And just let me get to a place where I was able to get a tool that was able to do, like, not just, like, putting context out, but, like, taking the changes back in and applying them. Yeah, I like to think I'm the number one user, but I actually, like, look at the stats sometimes, and I don't think that's even true.

Anymore. Yeah, man, I really wanted to bring you on after I saw that tweet from Andrej Karpathy the other day. So Andrej Karpathy, he used to be at Tesla AI now. He's like, one of the best educators about how LLMs work and things like that. He had this tweet saying, noticing myself adopting a certain rhythm in AI assisted coding code, I actually and professionally care about contrast to vive code. You know, he coined the term vive code, which everyone's been using. And then he basically goes on to talk about, like, stuffing everything relevant into context. All this. I was like, he literally. He doesn't know about Repo Prompt. I'm like, how did this, like, top AI educator in the world, top expert, everything. Totally has no idea about Repro Proproft. I was like, okay, so I need to get Eric on the podcast. At least try to help with.

That. Yeah, I appreciate that. Yeah. I mean, yeah, looking at that. That tweet, you see exactly like, that flow that, like, got me started. Like, when you start getting serious about coding with AI, like, you start thinking like, it will. How do I get the information to the AI model? And, like, the UX on all these other tools is just not cutting it. You need a tool to just be able to quickly select, search for your files, like, find things and. Yeah, you Know, I recently added the context builder. I don't know if you've tried that out, but maybe, you know, if.

You could explain like try to simplify it. Yeah, and I think we should then just jump into a demo and we can kind of just go from.

There. Sure thing. Sure.

Thing.

Yeah. I mean the first thing you're going to do when you're going to open up repo prompt is pick. So I can either open a folder manually or just go to the last ones used. But generally when you're working with some code base like this in Flutter, like this has a lot of like different build targets and things that are not like relevant to working with Flutter. If you're not familiar with Flutter, it's a way of working to build multi platform apps. And so you can see it's got like Linux, macOS, web and all that stuff. But yeah, like when you're working in a repo like this, you want to think through like what are the files that are going through and if you're using a coding agent like with cursor or whatever, the first thing they're going to do when you ask a question is, okay, well let me go find what the user's trying to do. Let me search for files and pick those out. And if you know what you're doing with your code base, you tend to know like, okay, well I'm working on this button toolbar. Great. So I'll just like clear the selection out and I'm just working on these views here. Great. So I've selected those and that's it. So then I can see token use for those files. It's pretty small, so I'm able to just get to work, type my prompt and paste that in here. Help me update all the docs pages. So if I do that and then I just do Gemini Flash quickly to show what that looks like. So the context builder, the way that works and is it will actually search for files using an LLM based on the prompt that you've typed out. A big part of using repo prompt is that you have to know what it is that you're trying to select here. What I noticed a lot of users, they were just putting everything in, they would just say, okay, just select all and that would be it. And you'd be like, we still get the first. Yeah, that's the easy thing to do. You're like, okay, well there's the code base, perfect. But there's plenty of tools that can just zip up your code base and that's Easy. But the power of repro prompt is you can be selective, you don't have to select everything. So I can just hit replace here and then. Okay, well, what did that do? Okay, well, that actually found all these files here that are related to my query. Put them in order of priority of like importance based on what the LLM's judgment is. And of course, if you use Gemini Flash, you're not going to get the best results compared to like using, you know, like a bigger model like Gemini 2.5 Pro. But it'll pick those out. It'll use something called codemaps to help with that. And you can see the actual token file selection queries is just 6k tokens. Working with a code base, if you've spent some time, you know, programming in the past, I know a lot of folks, they're not super familiar with all the technicals there, but like vibe coding. Yeah, exactly, exactly. So revamprump has this codemap feature and what this will do is it will basically, as you add files, it'll index them and extract what's called, like, it's a map. But if you've used C before, there's like a header file and a CPP file. And what that is is basically you're explaining to the compiler, like, what is all the definitions in this file? Like you've got your functions, you've got your variables and all that stuff. And so it's like a high.

Level extracted kind of like an index, like an index of your code.

Base. Exactly, yeah. The context builder uses that data to help you find what the relevant files are based on your query. So it has like a kind of peek inside the files without having all of the details. And it's able to kind of surface that relevant information for you so that you can use that in a.

Prompt. One thing I love about Repo prompt, so when I first started using it, I had been like using just like a custom script I had created to like take my code base and like, and then like put, you know, the relevant, you know, context in there, which a lot of times I was just doing all of it. I was literally putting all into a single file and I'd, I'd copy and paste that into ChatGPT. Yeah, I think I tweet about this and someone told me like, oh, you got to try repo prompt. When I tried repo prompt, the fact that I could like see how much context I was sharing with the model was amazing. And it seems like that's super relevant too, because you Know, at least from the benchmarks I've seen, you know, everyone's talking about how much context you can put into their LLM. I think on the benchmarks for llama 4, as soon as you went over like 128k context, like nowhere near the 10 million, like the quality just like dropped like, like a.

Rock. Until Gemini 2.5 came out, pretty much all the models you would really want to stay below 32k tokens. In general, I find like over that, you're just losing a lot of intelligence. So there's this concept of effective context, you know, the effective context window. Like at what point does the intelligence stop being like as relevant for that model? And for a lot of smaller models and local models it's a lot lower and you probably want to stay around 8k tokens. But like for bigger models, 32k is a good number. And it's only now with Gemini that you're able to kind of use the full package, the full context window. But yeah, so you're using this context, you've picked out your files, say you want to use as many as you want, 100k, like what do you do with that? So you have a question like, help me change how links are handled with my docs. And so I have a question here. I'm just going to paste it to O3 and you'll see what is O3 getting out of this? So it's getting basically this file tree. So it's getting a directory structure of this project, it's getting basically the high level code maps of the files that I haven't selected. So basically when it's set to complete, everything that I haven't selected gets kind of shipped in and then you have the files that I did select. And so then the context is able to go ahead and it's able to do that. And so this is like a great way to kind of just get this information into O3, get the most out of this model. And O3 is an expensive model if you're trying to use it a lot. Like this is a great way to kind of get more value out of it, move fast and get good.

Responses. I think the average person, like people who are just using ChatGPT or even people who are coding with cursor, they don't realize that you can do that, that you can literally just copy and paste all of that context in there and that the LM gets that and it understands what to.

Do. Yes. You know, in contrast to ChatGPT, Claude is very good at following instructions like it's the best model at following instructions, I find. And I think this is another thing that repo prompt does quite well is so it's got like tools to kind of send information into the LLM, but it's also got tools to go ahead. So it's now it's going to go ahead and write an XML plan and it's going to create this theme selector and it's going to add these files and change files for me. And what's cool with this is that I can just go ahead and use claude with my subscription and then have it modify all these files. So it's basically creating all these files and it. You can search and replace parts of files too, so I don't have to re update and re upload the whole thing, have it output the complete code. So a lot of models struggle with people notice that they're like, oh, this model's really lazy. It's not giving me the whole code. But this kind of circumvents that issue because it lets the AI just kind of get an escape hatch and just do what it needs to do here. Sometimes when I'm coding like this, I'll iterate. So I pasted this question with O3 and often what I'll do is I'll read through the answer and then I'll change my prompt and then paste again into a new chat and try and see where the result is different. Because basically I look at here's the output. Okay, I actually don't care maybe about this copy link button. Okay, then I'll specifically put a mention in my prompt to say let's kind of just focus on this part of the question and kind of reorient it. And that's the nice thing with this, is that I can just hit copy as many times as I want. If you're paying for a pro sub, there's no cost to trying things. There's no cost to hitting paste again. And you just try again. You just paste again, let the model think again and try things. And I think that's like a really important way of working with these models is to experiment and try things and, and see how does changing the context, what files you have selected your prompt. I use these stored prompts that come built in the app. So there's the architect and engineer and these kind of help focus the model. They give them roles. So if I'm working on something complicated, the architect prompt will focus the model on just the design and have it kind of not think about the code itself, whereas the engineer is just the code. Like, don't worry about the design, just kind of give me the code. But just the things that.

Change. Maybe you should explain when you say engineer prompt, it's literally you're just adding stuff that you copy and paste into the LLM, saying, like, you're an expert engineer, and this is what I expect from you. I expect for you to give me xml. That's your job, do it. And that's literally how the LMS work. Like, okay, I'll do.

It. Absolutely, yeah. Giving them roles is crucial. Telling them who they are, what their job description, you know, what do I look for? Like, giving them a performance review evaluation, all that stuff. Like, I find, like, the more detailed you are with your prompts, the more you can help. Like, they kind of color the responses in an interesting way. So just adding the engineer prompt, you see it, spent more time thinking about it. So here, this time it kind of said, okay, this is the file tailwind, here's the change, and this is the change that I'm going to do in a code block. So for the longest time before I had any of these XML features, I was just kind of using repo prompt and getting these outputs and then just copying them back into my code base manually and kind of reviewing them. That was really the antithesis of vibe coding, where everything's kind of.

Automated.

Yeah. So I showed you a lot of stuff, like pasting back, seeing this xml and then kind of putting it back in. What's really nice with repo prompts, like chat flow, is that all of that is automated. So if you want to vibe code and kind of think about it, like, just not think about anything while being kind of cost effective too, you can do that kind of work here. And basically the way this works here is I had GPT 4.1 as my main model. This is all the context I gave it. And then my Pro Edit mode, what it'll do is it'll actually ask a second model to apply the edits. So I have Gemini Flash that will go ahead and rewrite the file for me and just kind of do that work so I don't have to manually kind of incorporate these. So if I was looking at here like, okay, this is the tailwind file, I'd have to open that up and then go ahead and introduce it in. But having it kind of just go in the chat, having different models kind of do that work, you know, it makes a big difference. Working on repo prompt, it's really like there's building your context, that's like the biggest thing, just picking what you want. You want to front load that work. And you know, in contrast to using agents, you're going to have those agents kind of run off, do a lot of work, call a bunch of tool calls. You see like O3 kind of thought for 15 seconds, thought through some tools to call. It didn't really make sense, it just kind of kept going and ended up doing this. And if you've used cursor a lot, you'll see often using O3, it'll call tools, it'll read this file, read that file, read this file. But if you just give it the files up front and you just kind of send it off to work with your prompt, right away you get a response and you're like, okay, well does this make sense to me? Am I able to use this instead of letting it kind of serve for an.

Hour? Yeah, it's a little bit more work at least right now, but I think you get a lot better.

Results. Yeah, yeah, Just front loading that context, being able to think through and iterate on that, and that's the whole philosophy around it is just like thinking through, like making this easy. The context builder helps you find that context. You know, eventually I'm going to add MCP support so you can query documentation, find things related to your query as well, and just spend time as an engineer sitting through what do I want the LLM to know and then what do I want it to do and then make that flow as quick and as painless as possible. And like that's kind of everything. And I think, you know, going forward and as you get serious coding with AI, like that's what the human's job is in this loop. This engineer's job is figuring out the context. I think that's the new software engineering.

Job. Like I said before, I'm so surprised a lot of people haven't talked about this because like for me, like right now cursor is good for like something very simple, like okay, change some buttons or change some links or change whatever, you know. But anything complicated repo prompt, I got like way, way better results. So I'm curious, like, you know, have you ever thought about like this being used for things outside of coding and do you think it would be useful for anything outside of.

Coding? Yeah, I mean I've gotten academics reach out to me telling me they're using it for their work. There's folks in different fields for sure. I think some of the UX has to probably improve A little bit. But in general, like, you know, if you're working with plain text files, you know, Repo Prompt can service those use cases for sure. It's all set up to read any kind of file and then apply edits to any kind of file too. Like, I don't differentiate. If I can read it, then I'll apply edits for you. And I think a whole bunch of work is around just, like, gathering context and kind of iterating on stuff. Like, even, you know, in doing legal work, I do think, you know, a flow that is still missing from this app. It's just that, like, kind of collaborative nature. I think there's still some work that needs to kind of be done to kind of make this a more collaborative tool, make this a tool that kind of syncs a little bit better with different things. Like, for now, like, developers use Git, and that's that kind of collaboration bedrock. But I think, like, lawyers need other things. Yeah.

Yeah. That's something I think too, is like, yeah, Repo Prompt's super useful, but you have to be a little bit more advanced, like, than the average Vibe coder or the average person using an LLM. And, you know, no offense, you can kind of tell one person has built this. You know, it's amazing, but you can.

Tell. Yeah, yeah, yeah. No, it's all.

Good. I'm kind of curious, like, why did you not go the VC route? Where's Repo Prompt at right now? Like, where is it now and what's your plan for.

It? You know, I've had a lot of folks, you know, bring that up to me, and they're kind of thinking through, like, you know, why not VC or whatever. And I think it's not something that the door's closed on forever. It's just, I think right now it's working. I'm able to build and I'm able to kind of listen to my users and pay attention to what they need. And I think it's just not super clear to me where this all goes. This is an app that is super useful and it's helping me and I'm able to build it. But is it something that necessarily makes Sense to have $100 million invested into it to grow a huge team to.

Build?

Maybe. I don't know. But I want to kind of take things as they go as well. And right now I'm able to monetize it a bit. It's got some passionate users. It's working well this way. But again, it's all new to Me, I've not gone through this whole VC story myself. I've had friends who kind of shy me away from it, but I try to listen to the folks around me too, and see where it.

Goes. Yeah, there's pluses and minuses to VC you hear on Twitter and things like that. People who are like, oh, VC is horrible, or oh, it's amazing. There's good and bad to all of it. Yeah, I feel like everything with AI right now is like, who knows what's going to happen? Like, yeah, in a year, everything could be different. In five years, who the hell knows, right? Yeah, like right now, because AI is such a big wave. That's why we call the show the Next Wave. Right. It's such a large wave of transformation happening that you are going to see the largest investments ever, I think, in history, as well as the largest acquisitions ever. And I think these have yet to come. We're like in the early part of this transition, I think the best two routes for you, in my opinion would be either to go really big and go the VC route or to go more like, hey, who knows what's going to happen with it? I just want to get my name out there and I can leverage my name for something else in the future. And like open source. That's my kind of thought on strategically what I would do. It's like either go really big or open source it and make it free and just put it out there and say, yeah, you know, and get some reputation, benefit from.

It. There is a free tier. It's not open source, but there is a free tier. You know, the thing about open source actually is something I've thought about a lot. And the big issue with it right now, especially as people are building AI tools, is that like, it's never been easier to fork a project and kind of go off and just build it as a competitor. If you've looked at Klein like Klein is a big tool, you know, that came around, actually started around a similar time as me working on Repo Prompt. And if you're not familiar, the Klein is an AI agent that sits in VS code and it's pretty cool, but the thing that is not so cool about it is that it eats your tokens for lunch. Like that thing will churn through your wallet like faster than any other tool that exists just because it goes off and reads files, stuffs the context as big as possible. So a lot of people really enjoy using it because it does good results for certain things. But yeah, that cost is very high. But the thing that I was trying to bring up with this is that so Klein was actually forked a few months ago by another team of developers. And it's called Roux, the alternative. And if you look at open router in some stats, like Roux is actually surpassing Klein. And so that fork is now overtaken the original. And that's the kind of space that we're in where different teams will kind of take your code, take it in their direction, and then all of a sudden they've overtaken you and you kind of lose track of, you know, where things are going there. So, like, it's a crazy space. It's never been easier to open, pull requests with AI, you don't need to understand the code. You're like, oh, I have this open source project, I'm just going to fork it. Yeah, add my features and kind of go. And, and it's a tricky thing, but like, you know, having a free version and kind of trying to ship and grow a community of users who are passionate, who like, can talk back to you and, you know, I mean, that's kind of the route I've taken right now. And it's kind of been working so far. I was in beta for a long time. Yeah, you know, it's still new. Figuring out where to go next with.

It. And it's Mac only right now, is that.

Correct? Yeah, that's true. It's Mac only. And I think a part of that is that I started off just kind of trying to think about how do I build this in a good way. And the problem is I immediately ran into issues trying to build for different platforms. And I spent a bunch of time debugging, just getting svg, icon rendering, all these little things that are just rabbit holes. And you're like, okay, well, you're so abstracted from the base of what's happening and you spend a lot of time just solving build issues that it's like, I'm just going to go ahead and build native and just run with it and have better performance doing so. You know, if you open an IDE like VS code, you open up like a huge repo. What actually happens is that it'll load the file tree and it will just kind of lazy load everything. Like not everything needs to load. Because if you're opening an IDE as a coder, traditionally you only have a couple files open at a time. Maybe you have a.

Dozen.

Right. You're not going to be processing 50,000 files at the same time. But an AI model can, you know, if you give it to Gemini, Gemini will want all those files. It'll want as much as you can give it because they can read all of it. And so you need a tool that is built different, that is kind of organized in a way where it's kind of thinking first through that performance of mass data processing that you need to kind of do. It's a whole different way of working. That's why it's native, because, like, I want that performance. Processing all these files, there's all this concurrency happening where you're like, in parallel editing these files, like processing them and doing all this stuff. Like, it's very hard to do if you're just using JavaScript or.

TypeScript. When I use repo prompt, it seems like you've done a really great job of building it. It works really well. It is all just you, like right.

Now. Yeah, it is just me. Yeah, I've been working on it a lot.

Yeah. That's.

Crazy. Yeah, it's come a long way. I iterated a lot on it, you know. But that's the power of dogfooding too. Like, if you're not familiar, like folks listening, dogfooding is when you, like, kind of use your own product to iterate on it and build with it. And you kind of make it a habit of making sure that you're a number one user of your app, you know, your own product, to make sure that you see all the stuff that sucks about it, and for the longest time, it really sucked. And just that struggle and that pain of using it and forcing yourself to feel that pain, that's what makes it good. That's where you're able to feel those things that the user's using in the app will feel. And that's when you end up with something that is great in the.

End. So where do you think repo prompt is going long term? Which maybe long term now means one year. Where's repo prompt going next year? That's long.

Term. It's hard to say, honestly. It's weird. In December, OpenAI announces 03 and they're like, oh, it beats all the ARC AGI tests. And you're like, well, is this AGI? What is this? And then it shifts and it's like, okay. I mean, it's a better model. It lies to you. It's not like the Messiah, you know, so it's hard to say, like, I don't know, like, where we go. Like, I have ideas on where the future is one year from now. I think I'll have to adapt this product and keep iterating on it to kind of stay relevant. So it's going to keep changing. But like, I think that the flow I'm kind of pushing towards of that context building, I think that remains relevant for a while longer. And what improves is like the layers of automation around that.

Work.

Yeah. So I think, like, long term, I still think that is kind of the vibe that I want to go towards though. I think just like integrating mcp, just embracing that, like, universality of all of these different tools. So for folks listening, if they're not sure what is, MCP is another.

Acronym we got lots of in.

AI. So the idea there is traditionally, if you use like Claude or OpenAI, they have tools and those tools, you know, one of them could be like search the web, or one of them could be like read the files on your thing or look up documentation or these kinds of things. And there's this protocol MCP that creates an abstraction layer so that any client app can implement this protocol and then users can bring their own tools. So if a user comes in and says like, oh, I want to use. And there's this new one that's really cool, it's called Context seven, where basically they've gone ahead, built a server that fetches the latest documentation for whatever programming language you're using, and we'll kind of pull that in as context, so you can say, okay, great, fetch the latest Angular docs or whatever docs you care about and then you can bring that in. So that kind of work where you're like doing that context retrieval, that's super important. Or like Stripe has one too, where basically all the docs for their tool is set up and you just plug in the Stripe mcp. And then all of a sudden, if you're trying to vibe code your way through integrating Stripe, that's super easy. The work is kind of handled. You can plug in your API keys onto it so it can even talk to the backend for you. That whole work is kind of automated. So it's all about having tools for folks using these models to kind of automate connecting to different services in this universe of all these different, you know, services that exist in the.

World. Yeah, I kind of think of it most. I mean, it's different than xml, but for me, I think of it as almost more just kind of XML is like the information language that AI can understand. The MCP is like the same thing with any service you want to use or tool. It knows how for the AI to know how to work with those.

Things. Yeah. And funny enough, you mentioned xml, because that's actually one of the things that I do a lot with Rewrom is parsing xml. And I think one strength there that I have that like a lot of other tools are kind of ignoring. So traditionally when you're working with these language models, a developer and you can see this, if you use ChatGPT, you'd be like, hey, search the web. It's going to use the search tool and you'll see it say, called tool search and it'll go through. But what happens when it's doing that is that basically it calls that tool, it stops, waits for the result and then continues. I think of it like the robot is kind of being reboot as a new session with that new context, because basically every tool call is a new query. So you're giving back the old information, but you're not necessarily talking to that same instance. It's like a different AI instance that is answering your question from the new checkpoint. So that's a weird thing. So as you're making all of these tool calls, if you use cursor, it'll make 100 tool calls, but by the end of it, you've gone through 25 different instances of these models and then you get a result at the end and you're like, well, it's weird what actually happened. There's some data loss. Weird stuff. We don't know how all of this.

Is. Yeah, it does seem like that could create reliability issues because, you know, the LLMs, sometimes they give you amazing results and other times it's like, what is this? And so every time you're doing a new tool, it sounds like you're almost recreating the chance of it going wrong in a.

Way.

Exactly.

Yeah. You're aggregating these issues, but you don't even know where that inference. It could be different servers that are actually processing all these different tool calls. And it's weird, sometimes you'll have like, oh, that server has some chip issue on its memory and that actually causes some weird issues where Claude is actually really dumb today. But on the other one, it's a lot smarter because their chips, the memory chip is working fine, you know, you don't know. So that kind of thing. So just to close that back in on what I'm doing, the way that I've kind of gone about this is the way I call tools is you have your XML and the AI will just answer in one instance and it'll just give you the Whole thing. And it can call a bunch of tools in there. It can be like, hey, like I want to call this, this, do this and this, and then I just parse that and then bulk call the tools and then get the results and then we go another instance with the results and you can kind of go back and forth like that. So like not have to wait on each single one. You're actually just bulk sending them out, getting that data and it's a lot more efficient. You're able to process say like 25 queries, you know, get read 25 files, bring them all in, you know, let's work from there and see how it goes. And so that kind of thinking. So I think there's a lot to kind of play with in terms of, you know, how you're even getting this data back and forth from the LLMs. Because at the end of the day it's all text, you know, text and images, maybe some video in some cases, but like really text just for your coding, like that's the thing that you're working with and you can do a lot with text, manipulating it and playing with it to kind of get good.

Output. So what do you think? I've heard, you know, YC and others, I think Gary Tan said that. I can't remember if it was 80%. I think he said like 80% of the code for the startups going through YC right now is AI generated. That number could be wrong. Do you think in three years from now, do we still have like normal engineers who don't use AI at all? Is that a real thing? Do you still have the.

Holdouts? Well, first of all, like I think saying a percent like that of how much of it is AI generated, it's a bit.

Misleading. Can be hyped.

Too. Yeah, like I can go ahead and like every line of code, I could basically like type it in pseudocode to the AI model and like have it paste it back in as a fleshed out JavaScript function and say 100% of my code is written by AI. It really depends on how your workflow is, what your pipeline looks like. I do think fundamentally the job of an engineer has changed. It's already done, it's already completely different. You can't work the same way. But it depends on what point in the stack you're working on. I work for some folks who do some really low level graphics work and I talked to someone about how they can't really use AI models because the AI models just hallucinates everything. It's just not trained on anything that they work on, so it's just useless for them. But then if you work up someone who's a Web developer, well, 100% of the code, 98% of the training code is web code and web framework code. And so it's like, okay, well, yeah, 100% of that work can be done by AI. It's really easy. So it really depends on where you are in the stack, what kind of tool you're working with, and how well the AI models can help you in that. But I think as we move forward, more and more, you're going to want to have AI models thinking through hard problems for you, because it just happens much faster as they get better at math, at solving connectivity. Architecture. Architecture is something that these O3 and 01 Pro and hopefully O3 Pro just excel at. They're very good at finding good ways of organizing your code and helping you plan how you connect things together. And I think that's a big part of software engineering in general is just organizing your code, because the actual process of writing it, that's not the fun part or even the interesting part. It's that part of organizing. And I think a human job with this is to iterate on those plans, iterate on these ideas, because that's the kernel of what an AI will generate code with. So I think that's where the work is. You know, I still open the editor, like when I'm working on repo prompt. I don't write a ton of code by hand. Like, most of it is done by AI. But, like, I spend a lot of time thinking about architecture. I spend a lot of time thinking about problems and debugging and thinking through, like, I won't just like, hit the button and say, like, solve my problems, fix my bug. Like, that's just not helpful. But, like, if I read through the code, I'll be like, okay. Like, I think there's like a data race going on over here, this part connecting to this part. Like, there's some concurrency issue. I'll add some logs. Okay, great. I've got some idea of what's going on here. Perfect. Then you can kind of feed that data into the AI model and have it kind of think through a resolution. And often once you've done those steps of troubleshooting, the AI model can solve your problems, but you have to sit down and think through how things are connected and understand what is actually happening. So I think that's kind of where that work.

Changes. Yeah, it's a great Engineering answer. I'm looking for the thing that goes viral on X.

Right? Yeah.

Yeah. All engineers will be gone next year. This kind of.

Thing. Listen, the job is fully changed, I think from today on. If you're not using these tools, you're not learning how they work. I think that's an issue because I don't think a traditional engineer who spends his whole career just typing code up, that doesn't exist anymore. But what does exist is someone who understands code and who can read it and who understands what questions to ask. And if you're prompting about the code, if you understand the connections, that's where you're going to get the best results. And that's why a tool like repo prompt is so helpful, because you're able to do that and feed the right context in. But if you're just saying make the button blue or move it over here, I mean, that works to some extent. As long as your instructions are simple enough and you know what you want, you can get there. But at a certain point, you fall off, and that's when it stops working. And maybe that point where you fall off gets further and further as the models improve, but I don't think that in the next 10 years, we get to a point where that point stops.

Existing. One thing that we didn't talk about, that I was kind of curious to talk about, was what do you do at.

Unity? So what I do there is I've been doing kind of XR research and XR engineering. And so I work on a toolkit called the XR Interaction Toolkit. And basically it's a framework for developers to build interactions with xr. So if you're putting on an Oculus Quest or you're a Hololens or Apple Vision Pro, you want to basically interact with objects in your scene, in your world, in ar. If you're walking up and you want to pick up a virtual cube, how do you process that interaction of you grabbing the cube and picking it up and looking at it? So I've done a lot of research on that, that interaction of input. I written specs that are adopted for the industry in terms of hand interaction. So, like, you know, just tracking your hands, how do you grasp something? What should you be doing? If you want to poke a button that's like, not there, like, what does that look like? So that kind of stuff, that's what I do.

There. That's amazing. That's some really complicated engineering work. How are you doing that? Doing a repo prompt and then you have A baby. Like, how are you doing all.

This? I mean, I don't have a lot of free time.

Yeah, yeah. But I'm passionate about what I do at work too. And then reaper prompts. You know, this is my other baby. And I just think a big part of it is just, you know, when folks come to me and there's like something that's bugging them about the app. I just get an itch and I have to fix it for them. So I just keep tricking on it. But I try to get some sleep in so I don't cut through that too.

Much. One thing I was thinking about too is I have a son. 11 year old, you've got a baby. This actually one of the reasons I even helped start this podcast was I'm constantly thinking about where AI is going and wanting to stay ahead and also think about what does it mean for me and my family. Quite honestly, a selfish level. And people used to ask me when my son was born, because he was born in San Francisco, around tons of founders, VCs, all the kind of people who'd be around, like the birthday parties, right? It was all people from like YC and people like that. And they'd be asking me, like, you know, what do you think your son should do in the future? What will his job be? You know, this is like 11 years ago and I was talking about drones. Like, he probably needs to be like a drone defense engineer. Like building anti drone systems or something would be my common line that I would say at parties. But now with AI, like at that point, we did not know AI would advance as fast as it.

Has. No, it's happened so fast.

Right? It was all just like some stuff out of a book. It was like, oh, yeah, sure, they're talking about at Stanford and they got some cool demos, but like, nothing's working now. It's working. So with your child, have you thought about that.

Yet? Oh, of.

Course. What do you think they should learn? I have no idea.

It's everyone, right? What do you even teach your children? Is it important to learn to code? Do we teach them logic, morals? Probably all of this and more and being flexible and super.

Fluid? I think so. But it is funny on that topic. I look at engineers coming out and learning to code with AI around and I think they're at a disadvantage. It's unfortunate that if you're starting to code today and you have AI to lean on, you just don't have that struggle. You just don't have the pain that I had to go through when I started to code, when engineers who've been in the field for so long that had to struggle and not get the dopamine hit of a fixed problem right away to study it and understand how it works, that just doesn't exist anymore because the AI just solves it for you. And I think that's true in the code, but it's going to be more and more true in every field. And so I think there's going to be a need for people to have the restraint to kind of put aside these tools to struggle a little bit. I think there's a ton of value in kind of using them to learn and grow, but there's also that restraint that you need to form to kind of have the struggle, because that's where the learning is. And it's really tricky. And I don't know how you solve that now, because it's too easy not to struggle now, which is a big.

Problem. Yeah, I've heard Jonathan Blow, I don't know if you know of him, of course. Yeah, the game designer, he talks about exactly what you're saying, that it's in the future. Like, yeah, sure, AI could get amazing at coding in the future, but it's also going to create an issue where, just like you said, people are not going to learn to properly code. He was already complaining of that. Before AI, the code was shitty. And then now with AI, it's like, okay, now we're kind of screwed, I guess, because we're going to have a situation where no one knows what's going on and you're entirely dependent on the AI for.

Everything. Yeah, it's a crutch. So easy to reach.

For. Then what do humans.

Do? But that's the thing. I think maybe that's the middle part where. Where we're at this point where it's like the AI is just not quite good enough to kind of solve all the problems. And you still have problems to solve and you still have people that need to kind of work with the machines to kind of figure out how to go. Maybe at some point in the future, all of it is moot. I know some folks think that, and maybe it doesn't matter, but I think, you know, there's going to be some discomfort in the middle where, you know, the machines are not quite good enough to solve every problem. We lean on them as if they are, and then, you know, we're kind of atrophying a lot of skills that we've heard. You know, I haven't driven in A Tesla with fsd. But I've heard folks say the same thing there where, like, if they're using it all the time, they actually, like, suck at driving without it. And it's like, more and more, that's going to kind of be a thing.

Where. That is the thing where we start to go to almost living in one of those sci fi novels of everything being super, super safe. I live in Japan. Everything used to live in San Francisco. Everything's super safe in Japan. And there's one reason I like it, but you do lose some freedom in that. But do I really want my son driving now? If I really think about it, there's an alternative. Not necessarily. You know, I.

Agree. I mean, I have that same debate with my wife, you know, saying like, I don't think our daughter is going to ever have a driver's license. And she's like, I don't think so. You know, like, we'll see. But I don't know, like, there is the safety part for sure. And I think that's, like, really interesting. And hopefully, like, that is the case that, like, AI just does make it safe for.

Us. Yeah. Right. So, Eric, it's been awesome, and maybe we should tell people where they can find you and where they can find repo.

Prompt. Yeah. So I'm puncher with a V on X. So it's like P V N C H E R on X and most socialists my handle all over. So you can reach out there. My DMs are open if you have questions and repoprot repoprompt.com so you can just head over there and find the app free to download and nice Discord community too. If you want to hop over there and send me some messages and tell me what you think, please.

Do.

Yeah. Thanks for having me on, Nathan. It's been great chatting with you. Yeah, appreciate.

It. It's been.

Great. Yeah, yeah.

Yeah. Had a lot of.

Fun. Cheers.

Likewise. Take.

Care. All.

Right.

Sam.