14 Lessons From Cursor’s Ryo Lu on Iterative Creativity, AI, and Soulful Design

1. It's all the same thing. Everything—apps, organisms, ideas—is built from simple parts that recombine into complexity. In many ways, design is pattern recognition plus taste in how you remix the fundamentals.

2. Get close to the material. You learn by making, not planning. For software makers, that’s code, at the most basic level. For creatives, it's testing whatever medium gives feedback: clay, prototypes, rough cuts. Mocks and renders show possibilities. Prototypes show reality.

3. Greatness is emergent. It only looks inevitable in retrospect. You always start with shit. But because it exists, it can be improved. For Ryo, design feels like shaping clay or finding David in the marble. Poke, prod, ship, and quality sharpens into shape.

4.  Simplicity is earned, not given. “Simple” design is achieved by wrestling with complexity and compressing it into a digestible form. The swan glides elegantly because it's paddling like hell underneath. Easy is what comes after hard.

5. Design is making things true. Not polish. Not more details. It’s asking: what are the essential qualities of this thing? And then making them better, continually.

6. Learn to communicate at different altitudes. A prototype and a spec doc are the same thing, just at different levels of abstraction and in different flavors. The fundamental goal is translating ideas into forms people can absorb, and eventually use.

7. Great systems flirt with chaos. Push right up to the edge of collapse, then rein it back. You want weirdness, divergence, slack; just shy of too much. Don’t snuff out the new too soon.

8. AI is raw material, not finished goods. The purple gradient the AI tools give you is just the beginning. The difference between slop and soulful products is iteration and taste. Most people stop at the first output, rather than using it to begin.

9. Don't dumb things down. Give users pro-grade tools with beginner-friendly packaging. Discovery beats prescription every time. Get out of the way and let users surprise you. Treat their growth as journey you accompany them on, and build affordances along the way.

10. Build bricks, not readymades. AI is really good at composing parts, so build great bricks. Agents without guidance reinvent the wheel, so give them consistent, reliable materials to work with. Remember: There's no such thing as "the final design."

11. Tenderness is care made visible. Understand what people need and are frustrated with. Build tools that amplify their strengths while erasing drudgery. Good design is generous.

12. A band of pilgrims, not conscripts. Teams that build greatness don't hide behind convoluted processes or bloated hierarchies. They're there because they truly love the work. And you can tell: they talk about ideas and impact, not valuations and vesting cliffs.

13. Become irreplaceable. AI is trained on all public knowledge. You're trained on what you’ve experienced and how you respond: your taste, judgment, how you feel. That gap is what makes you distinct, and thus valuable.

14. Start anywhere. Compound everywhere. Small ideas stack up. You don't need permission or a grand vision. Make a choice, observe, then make it better.