当我们什么都能做，还知道自己想做什么吗？When We Can Do Anything, Do We Still Know What We Want? — Ryo Lu, ex-Cursor

Next up is Ryo. He was a founding engineer at Cursor. Please welcome him. Designer, sorry, designer. Welcome, Ryo. You'll want this one. The handheld mic is more fun. I'm a little nervous. Go easy on me. OK. Hi, I'm Ryo. My talk today is called “When we can do anything, do we still know what we want?” Let's get started. To start: our tools are stronger. Why do we feel more uneasy? I won't tell you how impressive AI is. You've seen it. You've used it. A website from one sentence. A feature in minutes.

You wake up to a whole night's work done for you, including things you hadn't even thought of. I was excited about that too. Things I could only imagine, I could finally make. Maybe you've felt something else too. Someone makes something with a new model. Your first reaction is, “Wow, that's amazing!” Then comes the second one: “Am I falling behind again?” Everyone seems smarter. Everyone's getting more done. A skill you spent years developing now looks like a sentence anyone can type into a box. And you think, I should do more too. Learn a few more tools. Spin up a few more agents. But if we really can keep this up forever, when is it enough? I've always believed in making tools.

I designed collaboration tools so people could build their own systems. Then I designed tools to help anyone make software. I wanted to close the gap between imagination and reality. I still believe in that. But as these tools let me do more, something else matters to me. Do they help me see what I want to make? Or am I just trying to keep up? I'm not here to debate whether to use AI, or tell you that slow is always better. When we can do anything, do we still know what we want? And with tools this powerful, how do we keep developing our own judgment, figuring out what we care about, and living our own lives? Part one. Before making things became proof of who we are. A lot of that anxiety comes from a habit.

We use what we can do to prove who we are. I'm a designer because I know design. I'm an engineer because I can write code by hand. So when a machine can do those things too, my job is only part of the worry. There's a more personal question. If what I do isn't special anymore, what's special about me? I don't have a neat answer. But I remember how I felt about computers before any of this became my job. When I moved to Canada as a kid, I saw that Apple computer with a screen that seemed to float. How could something be that beautiful? I couldn't afford one, so I made my own computer look like it. Changed the icons and fonts. Figured out how to make those buttons you wanted to lick. I wasn't trying to become a designer. I liked it. I was curious.

I wanted a closer look, to see if I could make something feel that way too. Eventually that became my job. I was lucky. I could make a living doing something I cared about and meet people who cared about it too. Then curiosity turned into planning. Play turned into output. Taste turned into decisions I had to make on other people's behalf. You get good at answering: What's it worth? When can we ship? But do you remember why you wanted to make it before anyone asked you to? I don't take comfort in saying, “A machine could never do that.” That line keeps moving. What I want to hold on to is my own relationship with making things. I shouldn't have to prove I'm better than anyone else to keep doing what I love.

A machine can make ten versions in no time. I can still stop at one detail and think, “That feels off. I want to fix it myself.” I can't always put that feeling into words right away. But it's closer to what I really want to make than thinking, “I should be doing more.” So my question is, are these tools skipping over that whole process too quickly? Part two. A complete answer doesn't mean I've thought it through. Say I ask, “Make me a tool to record my day.” Two minutes later, it says, “Done. Take a look.” Here's the journal it made. It even explains why I need each feature. Mood scores help me spot trends. Daily reminders help me build a habit. Weekly reports help me reflect. Sounds reasonable. I'd probably say, “Sure. Let's go with that.” But hang on. Have I really thought about why my life needs a score?

Did I decide that keeping a journal means writing every day? Or was its explanation of what others do so complete that I mistook it for my own understanding? That's what worries me. Sometimes I don't form an opinion of my own before I ask the tool for help. I see what it says first. I can't think of an objection, so I go with it. The answer might be right. It might even be better than anything I'd imagined. But if I do that every time, I may get good at judging whether something sounds reasonable, while getting less and less practice thinking a problem through from my own experience. Having an answer doesn't mean I've finished thinking. An obvious mistake is easy to spot. What's harder to see are the questions a complete answer covers up before we've even had a chance to ask them. Let's try something. Look at this journal. For a moment, set aside what features it might be missing.

Mood scores, writing streaks, daily reminders, weekly reports. Which one would you take out? Hands up if you'd drop mood scores. Writing streaks? Daily reminders? And weekly reports? I'd probably drop the streaks, the daily reminders, and the weekly reports. Miss one day, and I'd feel like I owed somebody. I want to record my life. I don't want homework. Maybe you're the opposite. That reminder is exactly what you need. Maybe keeping a streak matters to you. Or maybe it's all wrong. You want an infinite canvas to arrange memories and bits of your life however you like. No right answer here.

All we did was pause. But something happened in that pause. We picked out one decision, noticed we felt something about it, and started to say why. That vague idea, “I want to record my life,” finally started taking on a shape of its own. You don't always have to know what you want before you start. Sometimes you find out by making it. Part three. A service counter and a workbench. So how can our tools help? I'd hate to say you have to puzzle through everything first or you're being lazy. Some jobs I genuinely want to hand off. Move the data. Convert the file. I don't need a lesson every time. Some things I'm still figuring out. I want a tool that lets me keep exploring, rather than one that closes the question for me.

Compare these two screens. The one on the left is like a service counter. It processes my request and hands me a new version. If it happens to be good, great. I'll use it. But if it removes a reminder I like or changes the design, I have to explain again what to keep and what to drop. If all I see is the finished product, my choices boil down to accept, reject, or explain it again. The one on the right is more like a workbench. Same request, but I can see what it's planning to change, which features it'll remove, or what it wants to suggest. Seeing that makes me think: wait. Maybe I want to keep that reminder. I can step into the plan before it starts. Once it's done, I can fine-tune the result in a live preview. There's more to the difference than a few extra buttons. Here, I can put my hands on the thing I'm changing.

I can step inside instead of waiting for an answer. Sometimes I only realize I want that reminder when I see it's about to disappear. Sometimes I try my own design and realize the layout won't work. Or I can think it through before I run the plan. The tool helps me discover what I think. I don't have to explain the idea perfectly at the start. That's why, when I design software, I care about files, plans, edits, what changed between versions, and where a person can step in. Nobody should have to read every line of code. Showing every detail isn't the only way to be transparent. What matters is that when I want to understand something, there's a way in. When I want to change it, I can. Chat can be part of a workbench too.

I wouldn't put one interface or one brand permanently on either side. The question is: for this particular task, can I change just one part? Can I see why a decision was made? Can I take back control? Does the interface give me room to think and explore? I want tools that involve me more. I don't need proof they can do without me. If you build tools too, here are three small places to start. First, make important assumptions visible. Make each change small enough to try. Let people take back a task they've handed off. Second, you don't need a confirmation button at every step. Show people they have more choices than accepting or rejecting the whole answer. Third, remove the obstacles.

Leave people the chance to form their own judgment along the way. Part four. It's always there. Do I always need it? Even at a great workbench, you don't have to spend your life sitting there. We're still figuring out what kind of relationship we want with AI. It can help me think. Or I can stop thinking for myself. It can save me time, then give me more reasons to fill it right back up. A better tool still can't tell me how to use it. Take email. Having it doesn't mean there's an email to send about everything, any time of day. Some things are worth writing down. Some are better face to face. Some can wait for a reply. Sometimes nobody needs one more email. The tool is always there.

When to use it and what to use it for are still our choices. AI is the same. Sometimes I want speed. Help me sort through a complicated job. Sometimes I want to write a few lines myself first. They might be terrible. I want to see what's in my head. Sometimes I've had enough analysis. What I need is to go have tea with that person and actually talk. None of these choices is more virtuous. I want to keep my reasons for choosing differently. Leaving San Francisco and moving to Taipei was partly a way of practicing that. I still wanted to make things. I just didn't want to explain everything in my life in terms of whether it helped my work. Stepping out of my old routine didn't immediately make me sure of myself. With an empty calendar, I had to ask myself every day: what do I want to do?

If I can't introduce myself by my job, who am I? I started to feel that time could be worthwhile even if it produced nothing right away. Talking to another person. A street I've never walked down. A song I keep replaying. Any of them can remind me I still care about something. That feeling is mine. Life and creative work aren't competing. Squeezing life out won't leave you with more creativity. Life gives me things I want to make. So what do I hope these tools give us in the end? More than maximizing output, I want more time to care for other people and the world that we live in. I don't want to turn caring into another score to maximize. I mean giving time and attention to the person in front of you.

Having room to understand something you haven't figured out yet. Being able to do something about what you care about. Part five. Beyond getting better on our own: playing together. Caring about the world also means letting someone else into my thinking. Back to our journal. We changed one decision. It's closer to the journal I want now, but why would it be the one you want? Here's a simple page for each day, with all the extras stripped out. Or it could be a scrapbook: an open canvas with little fragments of memory. Which of these two would you rather use? Daily pages make a particular day easy to revisit. A freeform canvas brings memories from different times together. We don't have to pick a winner. We can keep both versions. We could even put them together.

Find someone who picked the other one. Ask them why. Now we have more than a finished product. We have something that helps us understand each other. We both say, “I want to record my life,” and think we mean the same thing. Then we see these two versions, and realize we're trying to hold on to different things. That's how a lot of my judgment at work took shape. Sitting alone didn't get me there. Someone tried what I'd made and said it felt odd. Someone insists on a detail I think is trivial. Someone takes apart an idea I was sure about. That's how I see what I missed. That can be uncomfortable. Working together means misunderstandings and disagreements. You apologize and try again. It's also where a lot of relationships I treasure began. People have seen me confused, wrong, exhausted.

They've also seen me genuinely excited. Over time you learn who spots the things you miss, and who can take an idea apart without killing it. It's like a jazz band. Nobody's trying to play the most notes. Nobody needs to prove they're the best player. One person carries the melody. Somebody else listens, leaves a little space, then comes in. Lose the beat? You listen to each other and find it again. What comes out is something none of you had in mind beforehand, and none of you could have made alone. AI is convenient. It's patient. Often it's easier than talking to another person. Sometimes that really does help me. I want to acknowledge that. If I always choose whoever responds to me most easily, I worry about what I'm missing: the chance to get in sync with someone different from me.

I hope our tools don't lock each of us inside a more efficient little world. I don't need a world that just responds better to me. I want a world that can change me. Tools let me do things I couldn't do alone. Other people make things possible that I never would have thought of. The final act. You can start making things before you know who you are. So back to today's question. When we can do anything, do we still know what we want? My answer is still: not necessarily. But I don't think that means there's something missing in us. And we don't need to find some grand mission before we can begin. Sometimes finding yourself has little to do with suddenly knowing how to introduce yourself. It happens in little choices. You find yourself thinking: this is what I care about.

This part I won't take as it comes. That part? I'd like to spend more time on it. Like just now. We didn't write a product spec first. We saw a version, felt something was off, changed it, and found out what we wanted. Making things gives you a way to express who you are. It also lets you meet parts of yourself you didn't know. To me, finding yourself and going beyond yourself go together. Making things shows you what you care about. And with tools and other people, you can reach places you couldn't reach before. That doesn't mean outproducing yesterday's self every day, or turning yourself into a product that always needs an upgrade. Maybe you can now change something you used to have to accept. An idea you never dared show anyone can now be a rough prototype you give someone to try.

Maybe you can finally do that one thing you always wanted to do and never thought you could. If you take one thing away today, I hope it goes beyond trying one more tool. Next time you see a polished answer, give yourself a moment. Ask: which decision in here haven't I thought through for myself? Pick a part you care about. Try one small change. If you get a chance, show it to someone else. You needn't change everything or prove you're better than the machine. As long as you bring what you feel, what you think, and what you care about to that process, you're doing more than standing by and approving a result. You're shaping your own life. I still believe in tools. I hope they give people who couldn't participate before a chance to express themselves. I hope we can make things once beyond us.

And we shouldn't have to hand over our whole lives just to keep the system running. I don't want a tool that lives my life for me. I want one that helps me feel things more clearly, explore freely, and care for the world around me, make things with others that we couldn't make before, and slowly find my way back to myself. Thank you. Thank you, everyone.