Closer to the Material: How AI Changes How We Build, Ryo Lu | Compile 26

⬅️ Back to Videos

🎥 Closer to the Material: How AI Changes How We Build and What It Must Not Erase, Ryo Lu | Compile 26

Ryo Lu (Cursor). Duration: 21 min

Timestamps

  • 0:00 Building ryOS
  • 1:21 What should exist?
  • 2:11 The loop AI changes
  • 4:17 The black box risk
  • 6:43 Output vs. material
  • 8:31 The Glass interface
  • 11:14 Prototyping Glass with Cursor
  • 14:03 When software felt alive
  • 17:36 Where craft moves
  • 20:25 A more human future

Ryo Lu designs at Cursor and gave probably the most philosophical talk at Compile 26. He built a personal computing environment called ryOS as a way of thinking through what AI should and should not take away.

  1. Ryo built ryOS because he missed when computers felt alive. Not siloed apps, not command prompts. A place he could tinker and think in. Before Cursor, that feeling would have stayed a note in his head. Instead he built a piece of it, played with it, reshaped it, and eventually it stopped being a vibe-coded prototype and became a place he could think inside. “It has my soul.”

  2. The loop is the thing AI actually changes. Before, making software was expensive. You planned before touching the material because being wrong cost a lot. Now the loop is minutes or seconds: idea, ask for a version, inspect, change, run it, keep going. That changes how judgment forms because the first version is almost never right, and you only notice after it exists.

  3. That small correction cycle is where taste comes from. You try something. The world answers. You notice what feels wrong. You adjust. That is where craft develops. The promise of AI is not making more things faster; it is letting more people into that loop as deep as they want.

  4. The black box is the risk. When the loop gets hidden, people stop learning from it. You make a wish, you get an output, you accept or reject. You never know why it worked or where it failed. Ryo calls this “pulling a slot machine.” If AI makes execution cheaper but judgment weaker, we do not get a renaissance. We get slop with working buttons.

  5. More people can build and more mediocre things can be built, and both are true. When things cost almost nothing to make, mediocre things do not look bad. They look fine. They ship. They fill the space. Ryo worries about software that functions but means nothing, interfaces that are correct but feel dead, products made by agents and cared for by no one.

  6. The danger is confusing output with material. Output ends the loop. Material invites you back in. Output says “Here’s the answer.” Material says “Touch it, shape it, make it yours.” Making is not just producing an artifact. It is a way of thinking. If AI hides too much of the process, it takes away the struggles where you become stronger.

  7. Glass is Cursor’s answer: a design principle, not a visual style. The system should let you see through its work. You see the plan, the thoughts, the tools streaming, the changes, the commands. You can stop it, shape it, inspect deeper, take over. You do not have to read every line, but you always can. Black boxes optimize for clean output. Glass optimizes for human agency.

  8. They prototyped Glass by building Cursor with Cursor. The prototype started as Baby Cursor 3, an Electron app on top of the Cursor CLI. The point was to feel the questions, not answer them in Figma: what should the interface feel like with one-to-N agents? How much state should be visible? Speed was not the interesting part. Faster prototyping helped them arrive at conviction.

  9. Software used to have texture. Then we optimized it away. Ryo remembers when the Mac dock bounced, the Genie effect swooped, Exposé scattered windows like cards. None of it was necessary, but all of it felt like someone cared. The quirks got removed because they did not test well. The warmth got cut because it was not measurable. We optimized our way into a world of things that work perfectly but feel like nothing.

  10. Taste is not a prompt. Caring is not a parameter. When making things costs almost nothing, slop becomes free too. Ryo’s argument: humans matter more, not less. The weird, specific, personal thing you put into something cannot be averaged into existence.

  11. Craft does not disappear. It moves upstream and downstream. When execution was expensive, craft lived in the code. When generation becomes cheap, craft moves upstream to judgment: what to ask, what to keep, what to refuse, where to slow down, what not to make. And downstream to responsibility: what did we release, what did it change, who did it serve? Making is a bet on what kind of world should exist.

  12. The point of AI is not to make humans disappear. Ryo ends where he started: feeling that something should exist, and making it real enough for someone else to see and touch. That part has not changed and will not change. The tools will change, the models will change, the economics will change. But that part stays.

Related TMFNK Content

Crepi il lupo! 🐺