Andrej Karpathy on software 3.0 & the future of programming
Karpathy frames programming in eras: hand-written code (1.0), neural-network weights learned from data (2.0), and now LLMs programmed with English prompts (3.0), where much of what apps and code used to do is absorbed by the model or an agent. His stance on agents moved fast, from keeping AI on a leash in mid-2025 to delegating nearly all his coding by early 2026, and he now calls the professional version agentic engineering: humans own the spec and the quality bar while agents do the work.
15 points · 6 appearances · 2022–2026
How the view evolved
Software 2.0: neural nets are taking over software → Software 3.0: prompts are programs, written in English
In 2022 he described Software 2.0, programming by curating datasets and training nets; by 2025 he named a new era, Software 3.0, where LLMs are programmed with prompts written in English.
Software 3.0: prompts are programs, written in English → The new program is text you paste to your agent
In 2025 prompts were programs for an LLM; by 2026 the program is text handed to an agent that brings its own intelligence, inspects your environment and debugs in a loop.
Keep AI on the leash; the human is the bottleneck → Remove yourself as the bottleneck to maximize leverage
In mid-2025 he wanted AI kept on a leash, with the human as the necessary bottleneck reviewing diffs; by March 2026 he said the goal is to remove yourself as the bottleneck and arrange things to run fully autonomously.
Industry oversells coding agents; autocomplete is the sweet spot → In December, coding flipped to agents writing most code
In October 2025 he said coding agents were oversold slop and autocomplete was his sweet spot; by March 2026 he said that in December he flipped to delegating most code and had barely typed a line since.
Vibe coding: hand control to the agent and hope → Vibe coding raises the floor; agentic engineering keeps the quality bar
In February 2025 he coined vibe coding as giving in to the agent and hoping it works; by 2026 he separated it from agentic engineering, where vibe coding raises the floor but professionals still keep the quality bar.
Up next
2022–2026 · changed view
“Neural networks have done language modeling for a long time. So really what's new or interesting or exciting is just realizing that when you scale it up with a powerful enough neural net, Transformer, you have all these emergent properties, where basically what happens is if you have a large enough data set of text, you are in the task of predicting the next word, you are multitasking a huge amount of different kinds of problems. You are multitasking understanding of, you know, chemistry, physics, human nature. Lots of things are sort of clustered in that objective. It's a very simple objective, but actually you have to understand a lot about the world to make that prediction.”
2022–2026 · changed view
“One more thing that I would say is that many times, when people think about education, they think more about what I would say is a softer component of diffusing knowledge. I have something very hard and technical in mind. In my mind, education is the very difficult technical process of building ramps to knowledge. In my mind, nanochat is a ramp to knowledge because it's very simple. It's the super simplified full-stack thing. If you give this artifact to someone and they look through it, they're learning a ton of stuff. It's giving you a lot of what I call eurekas per second, which is understanding per second. That's what I want, lots of eurekas per second. So to me, this is a technical problem of how do we build these ramps to knowledge. So I almost think of Eureka as maybe not that different from some of the frontier labs or some of the work that's going on there.”
2022–2026 · changed view
“The consistent thing I've been always interested in, for me at least, is AI, and that's probably where I'm spending my rest of my life on, because I just care about a lot. And I actually care about like many other problems as well, like say aging, which I basically view as disease, and I care about that as well, but I don't think it's a good idea to go after it specifically. I don't actually think that humans will be able to come up with the answer. I think the correct thing to do is to ignore those problems and you solve AI and then use that to solve everything else. And I think there's a chance that this will work; I think it's a very high chance. And that's kind of like the way I'm betting.”
2022–2026
“Yeah, the data engine is what I call the almost biological feeling like process by which you perfect the training sets for these neural networks. So because most of the programming now is in the level of these data sets, and make sure they're large, diverse and clean. Oh, basically you have a data set that you think is good, you train your neural net, you deploy it, and then you observe how well it's performing, and you're trying to always increase the quality of your data set. So you're trying to catch scenarios basically there are basically rare, and it is in these scenarios that the neural nets will typically struggle in, because they weren't told what to do in those rare cases in the data set. But now you can close the loop, because if you can now collect all those at scale, you can then feed them back into the reconstruction process I described and reconstruct the truth in those cases and add it to the data set. And so the whole thing ends up being like a staircase of improvement of perfecting your training set. And you have to go through deployments so that you can mine the parts that are not yet represented well in the data set. So your data set is basically imperfect, it needs to be diverse, it has pockets there are missing, and you need to pad out the pockets. You can sort of think of it that way.”