Lessons from Ryo Liu of Cursor

Ryo Liu (often credited as Ryo Lu), the Head of Design at Cursor, has shared numerous insights into his design philosophy, workflow with AI, and vision for the future of software development.

On Working with AI in Cursor: The 12 Golden Rules

Ryo Liu posted what he calls the "12 Golden Rules" for using Cursor effectively to avoid creating "AI spaghetti code." These have been widely shared and interpreted. [1]

Set clear project rules upfront. "Set 5–10 clear project rules upfront so Cursor knows your structure and constraints. Try /generate rules for existing codebases." [1]
Be specific in your prompts. "Spell out tech stack, behavior, and constraints like a mini spec." [1] A vague prompt will result in vague code. [2]
Work file by file. "Generate, test, and review in small, focused chunks." [1] This methodical approach is more efficient. [2]
Write tests first. "Write tests first, lock them, and generate code until all tests pass." [1] This is a form of test-driven development (TDD) for AI. [2]
Always review and fix AI output. "Always review AI output and hard-fix anything that breaks, then tell Cursor to use them as examples." [1]
Scope the AI's attention. "Use @ file, @ folders, @ git to scope Cursor's attention to the right parts of your codebase." [1][2]
Provide context with design docs. "Keep design docs and checklists in .cursor/ so the agent has full context on next steps." [1][2]
If the code is wrong, write it yourself. "Cursor learns faster from edits than explanations." [1]
Iterate using chat history. "Use chat history to iterate on old prompts without starting over." [1]
Choose your models intentionally. "Gemini for precision, Claude for breadth." [1]
Provide documentation for new stacks. "In new or unfamiliar stacks, paste documentation links. Make Cursor explain all errors and fixes line by line." [1][2]
Be patient with large projects. "Let big projects index overnight and limit context scope to keep performance snappy." [1][2]

On Design Philosophy and Systems Thinking

Ryo emphasizes a systems-first approach to design, especially in the age of AI.

Think in systems, not features. He stresses the importance of understanding the entire system rather than focusing on individual features. [3]
Unify concepts to simplify. "Instead of like five concepts you have just one. and it's almost like you build layers of the same thing for most people." [4]
Design the container, not just the UI. "The things we design as designers. they just kind of go up one level instead of you are designing. exactly how this piece of UI will look you are actually designing a container coming up with the right set of blocks for AI to wield for each user dynamically." [4][5]
The ideal interface is different for every single person. Personalization is the end goal. [3][4]
Throw away titles and work on the thing together. "What I find that worked the best is you kind of throw. away all these titles. and stuff and you just work on the thing. together." [6]
Good design has craftsmanship, playfulness, and mass appeal. He references making software with "the craftsmanship of German cameras, the playfulness of Japanese toys, and the mass appeal of Coca-Cola." [6]
Simplicity needs to exist on a conceptual level. "We need to kind of make every level of that simple both like on the conceptual. level and all the things that tie things around." [6]
AI programming's ultimate form is humans on strategy, AI on tactics. This summarizes his view on the human-AI collaboration. [2]
Explore the map to understand the terrain. "If you play like Age of Empires. you start the game with... you only see like your your little town... and then you need to kind of explore the map... you need to kind of absorb. information like become the sponge." [7]

On the Future of Software and AI

Ryo has a clear vision for how AI will transform the creation of software and the roles of developers and designers.

The future of interfaces will match how people think. [4]
Specialization will matter less in an AI-enabled future. [4]
The IDE is just one form of Cursor. He envisions Cursor being accessible from anywhere, not just a traditional code editor. [4]
You might not even see code in the future. [4]
AI doesn't discriminate between forms of input. Whether you chat, draw, or write a document, the AI can interpret it. [8]
The roles of designers, engineers, and PMs will blur. [8]
We are building tools for creation. "People building these things they built tools for creation they built things for people to make more things to make their ideas become reality." [8]
With tools like Cursor, you can do whatever you want. "If you have the curiosity or a little bit of time. you can actually do whatever you want it might not be perfect the first shot." [8]
You can put a lot of soul into AI-generated work. "You can just vibe code any idea and it doesn't have to be slop. it can be something really polished." [8]
Don't build too many abstractions now. "Especially for AI. if you build too many abstractions. now you're going to get fucked. because a lot of them will not make sense." [3]
Strategically putting humans in the loop is key. It's not about humans writing all the code anymore. [4]

On His Personal Process and Projects (like ryOS)

Ryo is a builder at heart, and his personal projects heavily influence his work at Cursor.

Build to understand. "I try to just do projects on my own... that's how I kind of get intuition of like where the the limits of the AI. are where where our problems are." [4]
Poke at the boundaries of what's possible. "I'm trying to poke at the boundaries. of is this thing possible... because we're building software. and the code is the material. you have no limitations." [4][5]
ryOS is a playground. "I actually like I don't really want to define what it is it's like a playground i want to show people like oh a designer can make all of this by by one person." [7]
Clone how you think into code. His operating system, ryOS, is his brain externalized. [7]
You don't need full fidelity anymore. "You just need to kind of maybe take a picture put your ideas in some bullets. send see what happens tweak." [7]
Build in the way that you prefer. "I don't have to be forced to use. any tool or input. and I can build in the way that. I prefer." [7]
Stop asking for permission. A lesson he learned that fuels his creative freedom. [7]
Your weird side project might be the most important thing you build. [7]
I build baby cursor with cursor. This is how new features can end up in the main product. [4]
Play with different models to build intuition. "You need to know how different models behave... the only way to do it is you play with it you build it." [8]

More Learnings and Quotes

On joining Cursor: He built a prototype of the "future of Notion" in three days using Cursor and was hooked. [3]
On user feedback: He gathers a lot of information from everywhere to keep an awareness of how people perceive Cursor. [4]
On making software before AI: "It was like a really long. process... you need to first assemble a team." [5]
On the power of building blocks: When users figure out how to use the building blocks you provide, "they fall in love with the tool because they it gives them so much power." [6]
On his background: He started making websites when he was 11. [7][8]
On inspiration from Apple: The first iMac "taught me like you can make technology feel more human." [8]
On "vibe coding": He uses this term to describe a more intuitive and less rigidly structured way of building software with AI. [8]
On the core of his work: "I make a world where anyone can make software." [4]
On handling creative friction: When discussing disagreements at Notion, he said, "I was not wrong either," highlighting the importance of different perspectives. [9]
On the new generation of builders: "16-year-olds will out-build senior engineers." [9]
On identity: "We're just builders and makers." [9]

Learn more:

Lessons from Ryo Liu of Cursor

On Working with AI in Cursor: The 12 Golden Rules

On Design Philosophy and Systems Thinking

On the Future of Software and AI

On His Personal Process and Projects (like ryOS)

More Learnings and Quotes

Written by Antoine Buteau

Lessons from Dave Asprey

Lessons from Dr. Rhonda Patrick