[Highlight] The Future of Coding and Work in the Age of AI with Henry Shi

0:00Hey, Ben Kesinoka here, co-founder and general partner of Village Global, a network native venture firm. What you're about to hear is a clip from a longer conversation. If you'd like to listen to the full-length version, find the link in the description. Of course, subscribe so you don't miss the next one. Enjoy.

0:20I think of Anthropic as really leading in coding, which is with Claude. What is the future of coding, do you think? I think Anthopic is certainly leading in coding in many ways, and in many areas outside of coding as well, like knowledge work and things like that. One thing, and this is all public, which is if you think about how coding has developed, right? Like a year ago, 2004, we barely had tab autocomplete, right? You could tab autocomplete some functions, and that was like pretty incredible. Then 2005 came around, and you had agentic coding. Like you can chat with the agent, vibe coding.

0:56you had clock code, windsurf, things like that. And now by the end of 2025, you have pretty good agents that are kind of junior intermediate level engineers, right? And this is a tweet from one of the, Adam Wolf, which is a former engineering manager on clock code. He said something to the effect of like software engineering, as we know it, may be done by 2026, right? And this is, you can look at them on Twitter. So this is something that, again, no one knows the future, but if you think about just the fact of writing code and the pace at which we scale from 24 to 25 to 26, it's a real shot that the way we write code is going to change forever and no one's going to, and I think the analogy used is it's kind of like no one really thinks about a compiler, right?

1:43You just write the code and it spits out some binary bytecode and it just works. So because you have that and you write English and out comes Python, which then gets translated to binary, like maybe. Yeah. And so for the foreseeable future, how much do you think a coding background matters? I think having problem-solving critical reasoning matters a ton still because coding is just a way to express formal logic. And a lot of that is grounded in problem-solving, critical reasoning, first principle thinking, and things like that. So I think those foundational building blocks still matter. But yeah, writing Python, I'm not sure.

2:21just like writing COBOL and 4chan doesn't really matter right now. Or like writing C++ in many cases may not even matter. Like speaking Latin? Yeah, exactly. And so, yeah, yeah. And I think in many ways maybe English is the Python of the future. And how do you think about in this age where maybe expertise is getting free or something, what are humans good at? I'm not sure. I'm really not sure because like you can argue even the AI today is superhuman in many ways, right? Like it can solve open math problems. It can do coding competitions. It can do therapy. It can write poetry. It can do all these things that the average person probably can't do.

3:05And if you take any one slice of that, it's probably better than most people. Now, there's obviously experts who are still going to be better. But the average of that is pretty powerful. So I'm really not sure. I mean, this is a non-answer. I don't know if it's a good answer. And what about, so if you fast forward a couple years, what will work look like? I don't think anybody can predict it's a few years out, to be honest. If you look at AI 2027, the timeline towards that, I'm really not sure. I mean, I think an interesting thought experiment is whether it's going to be humans with a bunch of AI employees or is it going to be AI bosses with a bunch of human employees.

3:46Right. Right. Because if you think about AI, what it's good or not good at, it's probably really good at synthesizing all this information and making a decision. Right. But to actually do stuff, it kind of has to interact with the physical world. So if the AI boss says, hey, I need you to go to this place and pick up this item and come back and do this thing, you might not understand why, but it's probably calculated all the permutations and decision tree there. But the AI can't really do that because it doesn't have many times physical interactions. Versus humans, if you're humans with a boss and delegate into AI, you guys can do a lot too.

4:23But is the human boss going to make the most optimal decisions? And I'm not sure. And maybe it's a hybrid because you always need a human to be accountable or something or hold them accountable or something. Right. So, yeah, I don't know. I don't know which way it's going to fold. Okay. So if we can't predict the future, at least we can prepare. So how are you staying current on the latest developments and where do you learn? A couple of things. So one is just really playing with these models and staying very close to the frontier, playing with these models. like really experimenting and just seeing it for yourself because sometimes you have to see the power of these models to believe and to see kind of their capabilities because it's different between when someone tells you something and then you're actually seeing it for yourself and getting your mind blown.

5:13My mind is blown like once every couple of weeks. And so sometimes you just have to really play with it and experience it yourself. And two is probably consider joining Frontier Labs, especially for all the founders well seriously especially for all the founders and builders out there i think there's so much happening at these labs that um it's really incredible opportunity and also it's an opportunity to help shape the future all right for for a good cause on a good mission so i think definitely you get to see a lot of things firsthand and there's so many talented people with so many different ideas there's no shortage of different things to do or to innovate But definitely there's a lot that you kind of experience that kind of help shape your thought process about the development of AI.

6:01And is there anything specifically that you've coded up or something that helps you with personal productivity or anything like that? You know, I used to ask the question a lot, which is kind of interesting, the answers, which there is there doesn't seem to be this like magical AI tool that somehow is so amazing yet no one's heard about. Yeah, it's usually the usual suspects that you hear about, like clock code or cursor and things like that. So that's one thing I've learned. There's no magic tool that's somehow really good. And certainly if it's really good, you probably know about it. But I think part of it is just like continuously experimenting with these tools.

6:42and maybe one thing is also think about like what can you build even as a non-engineer right like as the model capabilities get better and better and you have from tab autocomplete to junior engineers to maybe eventually senior engineers what could you create that maybe it doesn't exist that but tell it for you that fits your workflow and just continuously trying that because at some point it's going to work. Right. It reminds me of homebrew computer clubs years and years ago. It's like people would get together in small groups and say, what can we do with this thing, which is a computer, right?

7:16Because it feels like it's still a little brittle and you generally often need somebody who's an expert in, you know, like who's coded before to help you, even if it's vibe coding. Andre Karpathy and I don't vibe code exactly the same way. Right. Okay, so let's talk about 2026. and beyond. Do you have any predictions that are? Yeah, 2026 is going to be an incredible year. I think we're going to see a ton of changes. And if we think 2025 is pretty crazy, where these companies are going from zero to 100, 200 plus million revenue, 2026 might be even crazier. I certainly think it's going to be one of the, I mean, every year is the biggest year, but maybe even more so.

8:00And if you, again, if you look at AI 2027 and you look at the sort of meter trend lines of where we're progressing, how many hours of autonomous tasks AIs can do, right? Certainly it's trending exponential. And one of the things I always tell people is, and again, no one knows this, but it's like, if you can stay on the exponential, right, then you'll get to AGI. But if you fall off the exponential, it's going to take you a long time to catch up because even if you fall off, like, even if you fall for a little bit, you probably lost like 10 years, right? Even if you catch up six months. So if you can stay on it, then you'll get to some form of economic AGI.

8:35But if you start to tape them off, maybe not. Thanks so much for listening to the Village Global podcast. You can check us out online at villageglobal.vc. We'd love to hear from you, your feedback, your ideas, your inspirations. You can email us at hello at villageglobal.vc.

8:57I