AI Programming Cursor Founder: What is Valuable in the Post-Programming Era is "Taste"

The goal of Cursor is to create a whole new way of programming. This article is based on an article written by Xin and was compiled, compiled and written by TechFlow. (Synopsis: OpenAI's new engineer agent Codex!) AI writable function, fix bugs, run tests: Limited to 3 types of users to use first) (Background supplement: Podcast essence: AI and bots cover the cryptocurrency boom, the next era of micro-entrepreneurship is the strongest) As one of the fastest-growing products of all time, Cursor reached $100 million in ARR just 20 months after its release. Over the next two years, it surpassed $300 million in ARR and continues to revolutionize the way engineers and product teams develop software. As of early 2025, Cursor has over 360,000 paying subscribers. Michael Truell is the co-founder and CEO of Anysphere, Cursor's parent company. Together with three MIT classmates, he founded Anysphere and launched Cursor in three months. Michael Truell rarely gives interviews to podcasts, having only been on Lex Fridman podcasts before. In this issue, he talks about predictions for the "After code" era, counterintuitive experiences in building Cursor, and perspectives on the future of engineers. This content is from Lenny's Podcast, and the following is the full text of the compilation. Cursor's goal is to create a whole new way of programming: people will see virtual code closer to English sentences in the future. People will have strong control over various details of the software, and have the ability to modify and iterate extremely quickly. "Taste" will become more and more valuable: "taste" is at its core of having a clear understanding of what should be built. The users who use AI best are conservative in their use of technology: they're very good at limiting the scope of tasks to AI to a smaller, more explicit one. The centerpiece of the Cursor interview is a two-day assessment: these assessments are mock, but allow candidates to produce real work over two days. This is not only a test of "whether you want to work with them", but also very important for attracting candidates. The only thing that attracts people to join in an early company is often a team that they feel is worth working with. The main problem with chatbot-style programming is the lack of precision Lenny: We've talked about what happens in the post-code era. How do you see Cursor heading in the future? How will technology move from traditional code to other forms? Michael Truell: The goal of Cursor is to create a whole new way of programming, a different way of building software. You simply describe your intentions to the computer in the most concise way possible, and it's up to you to define how the software should work and how it should be presented. As today's technology continues to mature, we believe we can pioneer a whole new way of building software that will be higher, more efficient, and easier to use than it is today. This process will be very different from the way software is written today. I'd like to contrast it with several mainstream views of the future of software form, some of which we don't quite agree with. One is that software construction in the future will still be very similar to today, relying mainly on text editing using formal programming languages such as TypeScript, Go, C, Rust, etc. Another idea is that you just type commands into the chatbot, let it build the software for you, and then let it modify it at any time. This chatbot style is like talking to your engineering department. We think there are problems with both visions. The main problem with chatbot-style programming is the lack of precision. If you want people to have full control over the look and functionality of the software, you need to provide a more precise way to instruct them to make the changes they want, rather than saying to a bot in a chat box, "Change this part of my app," and then removing it entirely. On the other hand, the worldview that thinks that nothing will change is also wrong, because technology will only get stronger. In the "post-code" world we envision, the expression of software logic will be closer to English. You can imagine that it will exist in a more canonical form, moving in the direction of virtual code. You can write the logic of the software, edit it at a higher level, and navigate through it easily. This will not be an obscure code that is difficult to understand, millions of lines. Instead, it will be clearer and easier to understand and locate. We're working to evolve complex symbol and code structures into forms that are easier for people to read and edit. In the post-code era, taste will become more and more valuable Lenny: It's profound, and I want to make sure people understand your point. The shift you imagine is that people don't see code anymore and don't have to think in JavaScript or Python. In its place is a more abstract form of expression, closer to the virtual code of English sentences. Michael Truell: We think it's going to eventually get to that stage. We believe that achieving this stage requires the participation and promotion of existing professional engineers. In the future, people will still dominate in the driver's seat. People will have a strong control over the various details of the software, and they will not easily give up this control. People also have the ability to modify and iterate extremely quickly. The future won't depend on the kind of work that happens in the background, is slow, and takes weeks to complete. Lenny: This begs the question, what skills do you think will become increasingly valuable in the "post-code era" for current engineers, or people who are considering becoming engineers, designers, or product managers? Michael Truell: I think "taste" is going to become more and more valuable. When people talk about taste in the software world, it's easy to think of visuals, smooth animations, color matching, UI, UX, and more. Vision is very important for products. But as mentioned earlier, I think the other half of the important thing is the product logic and the way it works. We have many tools to design visuals, but code is still the best representation of the logic that software executes. You can use Figma to show the effect or sketch it in your notes. But the logic can only be clearly presented when you have a prototype that is actually usable. The engineers of the future will become more and more like "logic designers." They need to express their intentions precisely, moving from the "how" behind the scenes to the high-level "what" and "what", which means that "taste" will be more important in software development. We haven't reached that point in software engineering yet. There are many interesting and thought-provoking jokes circulating on the Internet about people's over-reliance on AI development, and obvious flaws and functional problems in software. But I believe that the software engineers of the future will not have to pay too much attention to detail control as they do today, and we will slowly shift from rigorous and meticulous to more "taste". Lenny: It reminds me of vibe coding. Is this similar to what you described as not having to think too much about the details, but a more natural way of programming? Michael Truell: I think there's a connection. The current people talk about vibe...