{"success":true,"count":9,"items":[{"videoId":"We7BZVKbCVw","chunkIndex":0,"totalChunks":9,"title":"Head of Claude Code: What happens after coding is solved | Boris Cherny — Part 1 of 9","thumbnail":"https://i.ytimg.com/vi/We7BZVKbCVw/maxresdefault.jpg","duration":5265,"uploader":"Lenny's Podcast","youtubeUrl":"https://www.youtube.com/watch?v=We7BZVKbCVw","keywords":["ai-coding","software-engineering","developer-tools","artificial-intelligence","productivity","anthropic","claude","future-of-work","automation"],"normalizedKeywords":["엔지니어링","프로덕트","기술 트렌드"],"targetAudience":[{"who":"엔지니어","why":"AI가 코딩과 도구 사용을 어떻게 대체·확장하는지 직접 들을 수 있음"},{"who":"프로덕트 매니저","why":"코딩이 쉬워질수록 제품 발상과 의사결정의 비중이 커지는 흐름을 이해할 수 있음"},{"who":"창업자","why":"개발 생산성 급증이 팀 구조와 역할 정의를 어떻게 바꾸는지 힌트를 얻을 수 있음"}],"normalizedAudience":["엔지니어·개발자","프로덕트 매니저·기획자","창업자·스타트업"],"summary":"이 영상은 Anthropic의 Claude Code를 만든 Boris Cherny가, 지난 1년 동안 AI 코딩 도구가 소프트웨어 개발을 어떻게 바꿨는지 이야기하는 대담이다. 그는 자신의 코드가 전부 Claude Code로 작성되고, 하루에도 수십 개의 PR을 내는 현재의 작업 방식에서 출발해, 이제는 ‘코드를 쓰는 사람’보다 ‘문제를 정의하고 도구를 조율하는 사람’의 중요성이 커지고 있다고 본다. 또한 AI가 단순히 코드를 생성하는 수준을 넘어 버그 보고, 텔레메트리, Gmail, Slack까지 다루며 실제로 세계에 작동하는 주체가 되고 있다고 강조한다.\n\n대화는 Boris가 Anthropic을 잠깐 떠났다가 다시 돌아온 이유, Claude Code를 처음 만들 때의 시행착오, 그리고 왜 이것이 단순한 생산성 향상이 아니라 안전(safety)과도 연결되는지로 이어진다. 핵심 메시지는 ‘코딩이 해결되면 끝’이 아니라, 누구나 만들 수 있는 시대에는 제품 감각, 도구 활용 능력, 문제 정의 능력이 더 중요해진다는 점이다.","insights":["코딩 자동화는 개발자의 역할을 '작성'에서 '지시·검증'으로 바꾼다.","AI가 좋아질수록 제품 감각과 문제 정의 능력이 더 중요해진다.","도구를 쓰는 AI는 단순 보조가 아니라 실제 동료처럼 작동한다.","생산성 증가는 직무를 없애기보다 직무의 경계를 재편한다.","큰 변화는 기술 성능보다 성장 속도가 증명한다."],"keyClips":[{"clipId":"We7BZVKbCVw:c0:1-11","startSegmentIndex":1,"endSegmentIndex":11,"startTime":1.79,"endTime":58.39,"durationSeconds":56.6,"preview":"코딩이 바뀐 순간","mustSee":true},{"clipId":"We7BZVKbCVw:c0:12-19","startSegmentIndex":12,"endSegmentIndex":19,"startTime":58.39,"endTime":96.52000000000001,"durationSeconds":38.1,"preview":"Claude Code의 충격","mustSee":false},{"clipId":"We7BZVKbCVw:c0:43-57","startSegmentIndex":43,"endSegmentIndex":57,"startTime":235.52,"endTime":335.84000000000003,"durationSeconds":100.3,"preview":"Anthropic으로 돌아간 이유","mustSee":false},{"clipId":"We7BZVKbCVw:c0:61-71","startSegmentIndex":61,"endSegmentIndex":71,"startTime":349.2,"endTime":431.919,"durationSeconds":82.7,"preview":"지표가 말하는 확산","mustSee":true},{"clipId":"We7BZVKbCVw:c0:72-96","startSegmentIndex":72,"endSegmentIndex":96,"startTime":431.919,"endTime":609.4,"durationSeconds":177.5,"preview":"처음엔 작은 해킹","mustSee":false}],"curatedSegments":[{"segmentIndex":6,"text":"In a year or two it's not going [music] to matter. Coding is virtually solved. I imagine a world where everyone is able to program.","startTime":22.72,"endTime":27.88,"durationSeconds":5,"level":"B2","overallScore":8.4,"rationale":"미래 변화에 대한 강한 관점 제시."},{"segmentIndex":83,"text":"It acts in the world. Um, and I think now with Co-work, we're starting to see the transition for non-technical folks also. And I think the craziest thing for me isn't even the number that we're at right now, but the pace at which we're growing because if you look at Claude code's growth rate kind of across any metric, it's continuing to accelerate. And I think honestly, that's just for me as an engineer, I find that to do good work, you really have to understand the layer under the layer at which you work.","startTime":501.88,"endTime":508.8,"durationSeconds":7,"level":"C1","overallScore":9,"rationale":"성장 해석과 일의 원칙이 매우 통찰적."},{"segmentIndex":26,"text":"It's just And when I think about it, kind of part of the reason Quad Code works is this idea of latent demand where we bring the tool to where people are and it makes existing workflows a little bit easier. But also because it's a it's in a terminal, it's like a little surprising.","startTime":732.52,"endTime":745.16,"durationSeconds":13,"level":"C1","overallScore":9,"rationale":"제품 성공 원인을 원리 수준에서 설명함."},{"segmentIndex":48,"text":"Um, so we really just think in exponentials. And if you kind of look at the exponential of the percent of code that was written by Claude at that point, and if you just trace the line, it's pretty obvious we're going to cross 100% by the end of the year, even if it just does not match intuition at all.","startTime":876.76,"endTime":890.92,"durationSeconds":14,"level":"C1","overallScore":9,"rationale":"지수 추세 해석이 통찰·표현 모두 강함."},{"segmentIndex":54,"text":"Um you just have to give people space. You have to give them maybe the word is like safety. So, it's like psychological safety that it's okay to fail.","startTime":926.8,"endTime":931,"durationSeconds":4,"level":"C1","overallScore":8.8,"rationale":"혁신 문화의 조건을 구체적으로 제시함."},{"segmentIndex":24,"text":"So, I think a lot of the people that will be rewarded most over the next few years, they won't just be AI native and they don't just know how to use these tools really well, but also they're curious and they're generalists and they cross over multiple disciplines and can think about the broader problem they're solving rather than just engineering part of it.","startTime":2530.2,"endTime":2538.12,"durationSeconds":8,"level":"C1","overallScore":8.8,"rationale":"미래 인재상 통찰이 매우 응축됨."},{"segmentIndex":31,"text":"I do think that there is a future where I think by the end of the year what we're going to start to see is these start to get even murkier where I think in some places the title software engineer is going to start to go away and it's just going to be replaced by builder or maybe it's just everyone's going to be a product manager and everyone codes or something like this.","startTime":2587.96,"endTime":2606.2,"durationSeconds":18,"level":"C1","overallScore":8.8,"rationale":"직함 변화 전망이 매우 선명함."},{"segmentIndex":66,"text":"It's whatever people are doing if you can make that a little bit easier then that's just going to be a much better product that people enjoy more and this is just this principle of latent demand which I think is just the single most important principle in product.","startTime":2836.2,"endTime":2849.84,"durationSeconds":14,"level":"C1","overallScore":8.8,"rationale":"핵심 제품 원칙을 강하게 일반화함."},{"segmentIndex":68,"text":"Latent demand is this idea that if you build a product in a way that can be hacked or can be kind of mis [clears throat] used by people in a way it wasn't really designed for it to do kind of something that they want to do then this helps you as the product builder learn where to take the product next.","startTime":2857.64,"endTime":2875.08,"durationSeconds":17,"level":"C1","overallScore":9,"rationale":"개념 정의와 시사점이 매우 선명하다."},{"segmentIndex":8,"text":"So when you see people abusing the product in this way, using it in a way that it wasn't designed in order to do something that is useful for them, it's just such a strong indicator that you should just build a product and people are going to like that.","startTime":3036.24,"endTime":3047.2,"durationSeconds":11,"level":"C1","overallScore":9,"rationale":"제품 통찰이 강하고 표현도 풍부함."},{"segmentIndex":44,"text":"But this is kind of the way that we learn, um, both on the product side and on the safety side is we have to release things a little bit earlier than we think so that we can get the feedback, so that we can talk to users.","startTime":3230.04,"endTime":3238.68,"durationSeconds":9,"level":"C1","overallScore":9,"rationale":"학습 원칙이 선명하고 표현도 풍부함."},{"segmentIndex":45,"text":"But actually almost always you get better results if you just give the model tools, you give it a goal, and you let it figure it out.","startTime":3846.28,"endTime":3851.52,"durationSeconds":5,"level":"B2","overallScore":8.8,"rationale":"핵심 원칙과 실용 표현이 함께 강하다."},{"segmentIndex":53,"text":"Uh and it's actually a really simple idea. His idea was that the more general model will always outperform the more specific model.","startTime":3895.4,"endTime":3902.12,"durationSeconds":7,"level":"B2","overallScore":8.8,"rationale":"일반 모델 우위를 압축한 핵심 통찰."},{"segmentIndex":79,"text":"But if you build for the model 6 months out, when that model comes out, you're just going to hit the ground running, and the product is going to click and start to work.","startTime":4026.56,"endTime":4035.84,"durationSeconds":9,"level":"B2","overallScore":9,"rationale":"행동 지침과 표현이 모두 강함."},{"segmentIndex":48,"text":"And they're just following the momentum and not thinking about it. I think the best results that I see are people thinking from first principles and just developing their own common sense.","startTime":5069.56,"endTime":5077.64,"durationSeconds":8,"level":"B2","overallScore":8.8,"rationale":"성과 원인을 원칙 수준에서 설명함."},{"segmentIndex":4,"text":"I have never enjoyed coding as much as I do today because I don't have to deal with all the minutiae.","startTime":13.32,"endTime":18.04,"durationSeconds":5,"level":"B2","overallScore":7.6,"rationale":"코딩 만족도의 이유를 설명해 통찰적."},{"segmentIndex":11,"text":"The title software engineer is going to start [music] to go away. It's just going to be replaced by builder and it's going to be painful for a lot of people.","startTime":47.6,"endTime":58.39,"durationSeconds":11,"level":"B2","overallScore":7.8,"rationale":"직함 변화 전망과 여파를 압축 제시."},{"segmentIndex":55,"text":"Um and so the this kind of like mission drivenness just really, really resonated with me, and I just know personally it's something I need in order to be happy. Um and I that's just the thing that I really missed.","startTime":311.68,"endTime":321,"durationSeconds":9,"level":"C1","overallScore":8,"rationale":"가치관과 행복 조건을 깊이 있게 설명."},{"segmentIndex":56,"text":"And I found that, you know, whatever the work might be, no matter how exciting, even if it's building a really cool product, it's just not really a substitute for that.","startTime":321,"endTime":330.6,"durationSeconds":10,"level":"C1","overallScore":7.8,"rationale":"일의 의미에 대한 보편적 통찰이 있음."},{"segmentIndex":74,"text":"And you know, for Anthropic for a long time, we were building the models in this way that kind of fit our mental model of the way that we build safe AGI, where the model starts by being really good at coding, then it gets really good at tool use, then it gets really good at computer use.","startTime":443.919,"endTime":457.8,"durationSeconds":14,"level":"C1","overallScore":8,"rationale":"제품 철학과 발전 경로 통찰이 크다."}],"generatedAt":"2026-06-24T23:59:05.143Z","keyClipsTotalSec":1420},{"videoId":"We7BZVKbCVw","chunkIndex":1,"totalChunks":9,"title":"Head of Claude Code: What happens after coding is solved | Boris Cherny — Part 2 of 9","thumbnail":"https://i.ytimg.com/vi/We7BZVKbCVw/maxresdefault.jpg","duration":5265,"uploader":"Lenny's Podcast","youtubeUrl":"https://www.youtube.com/watch?v=We7BZVKbCVw","keywords":["artificial-intelligence","software-engineering","coding-tools","developer-productivity","product-iteration","startup","workflow-automation","future-of-work","anthropic","ai-agents"],"normalizedKeywords":["엔지니어링","프로덕트","기술 트렌드"],"targetAudience":[{"who":"엔지니어","why":"AI가 코딩과 코드리뷰를 어떻게 바꾸는지 실무 관점에서 볼 수 있음"},{"who":"프로덕트 매니저","why":"사용자 피드백을 제품 아이디어로 바꾸는 방식이 구체적으로 나온다"},{"who":"창업자","why":"초기 제품을 작게 시작하고 빠르게 학습하는 전략을 배울 수 있음"},{"who":"지식노동자","why":"코딩 밖 일반 업무까지 AI로 자동화하는 흐름을 이해하는 데 유용함"}],"normalizedAudience":["엔지니어·개발자","프로덕트 매니저·기획자","창업자·스타트업"],"summary":"이 영상은 Claude Code가 어떻게 탄생했고, 왜 터미널이라는 의외의 형태로 시작했는지, 그리고 그 선택이 어떻게 빠르게 확산되었는지를 설명한다. 핵심 메시지는 AI가 코딩을 빠르게 '해결되는 문제'로 바꾸고 있으며, 제품 설계 역시 거대한 계획보다 작은 실험과 사용자 피드백에서 출발해야 한다는 것이다.\n\n후반부에서는 더 큰 전환점으로, Claude가 단순히 코드를 작성하는 수준을 넘어 버그 수정 아이디어를 내고, 프로젝트 관리나 행정 업무까지 대신하는 흐름을 보여준다. 결국 이 대화는 '코딩 자동화'를 넘어서, 무엇을 만들지 정하고 일상 업무를 처리하는 전반적 지식노동의 자동화로 확장되는 미래를 그린다.","insights":["좋은 AI 제품은 처음부터 완성형일 필요가 없다.","모델이 너무 빨리 변하므로 UI도 그 속도에 맞춰야 한다.","코딩이 자동화되면 다음 경쟁력은 무엇을 만들지 정하는 능력이다.","혁신은 계획보다 실험과 심리적 안전에서 더 자주 나온다.","피드백 루프가 짧을수록 사용자는 더 많이 말하고 제품은 빨리 좋아진다."],"keyClips":[{"clipId":"We7BZVKbCVw:c1:1-18","startSegmentIndex":1,"endSegmentIndex":18,"startTime":601.11,"endTime":694.6,"durationSeconds":93.5,"preview":"터미널에서 시작한 이유","mustSee":false},{"clipId":"We7BZVKbCVw:c1:19-35","startSegmentIndex":19,"endSegmentIndex":35,"startTime":694.6,"endTime":806.08,"durationSeconds":111.5,"preview":"느린 이해와 피드백","mustSee":false},{"clipId":"We7BZVKbCVw:c1:42-49","startSegmentIndex":42,"endSegmentIndex":49,"startTime":837.88,"endTime":901.16,"durationSeconds":63.3,"preview":"지수적 변화 읽기","mustSee":false},{"clipId":"We7BZVKbCVw:c1:53-60","startSegmentIndex":53,"endSegmentIndex":60,"startTime":921.88,"endTime":967.68,"durationSeconds":45.8,"preview":"실험이 혁신을 낳는다","mustSee":false},{"clipId":"We7BZVKbCVw:c1:63-76","startSegmentIndex":63,"endSegmentIndex":76,"startTime":974.68,"endTime":1047.04,"durationSeconds":72.4,"preview":"코드 작성의 자동화","mustSee":true},{"clipId":"We7BZVKbCVw:c1:81-97","startSegmentIndex":81,"endSegmentIndex":97,"startTime":1074.84,"endTime":1206.88,"durationSeconds":132,"preview":"다음 일은 누가 정하나","mustSee":true}],"curatedSegments":[{"segmentIndex":6,"text":"In a year or two it's not going [music] to matter. Coding is virtually solved. I imagine a world where everyone is able to program.","startTime":22.72,"endTime":27.88,"durationSeconds":5,"level":"B2","overallScore":8.4,"rationale":"미래 변화에 대한 강한 관점 제시."},{"segmentIndex":83,"text":"It acts in the world. Um, and I think now with Co-work, we're starting to see the transition for non-technical folks also. And I think the craziest thing for me isn't even the number that we're at right now, but the pace at which we're growing because if you look at Claude code's growth rate kind of across any metric, it's continuing to accelerate. And I think honestly, that's just for me as an engineer, I find that to do good work, you really have to understand the layer under the layer at which you work.","startTime":501.88,"endTime":508.8,"durationSeconds":7,"level":"C1","overallScore":9,"rationale":"성장 해석과 일의 원칙이 매우 통찰적."},{"segmentIndex":26,"text":"It's just And when I think about it, kind of part of the reason Quad Code works is this idea of latent demand where we bring the tool to where people are and it makes existing workflows a little bit easier. But also because it's a it's in a terminal, it's like a little surprising.","startTime":732.52,"endTime":745.16,"durationSeconds":13,"level":"C1","overallScore":9,"rationale":"제품 성공 원인을 원리 수준에서 설명함."},{"segmentIndex":48,"text":"Um, so we really just think in exponentials. And if you kind of look at the exponential of the percent of code that was written by Claude at that point, and if you just trace the line, it's pretty obvious we're going to cross 100% by the end of the year, even if it just does not match intuition at all.","startTime":876.76,"endTime":890.92,"durationSeconds":14,"level":"C1","overallScore":9,"rationale":"지수 추세 해석이 통찰·표현 모두 강함."},{"segmentIndex":54,"text":"Um you just have to give people space. You have to give them maybe the word is like safety. So, it's like psychological safety that it's okay to fail.","startTime":926.8,"endTime":931,"durationSeconds":4,"level":"C1","overallScore":8.8,"rationale":"혁신 문화의 조건을 구체적으로 제시함."},{"segmentIndex":24,"text":"So, I think a lot of the people that will be rewarded most over the next few years, they won't just be AI native and they don't just know how to use these tools really well, but also they're curious and they're generalists and they cross over multiple disciplines and can think about the broader problem they're solving rather than just engineering part of it.","startTime":2530.2,"endTime":2538.12,"durationSeconds":8,"level":"C1","overallScore":8.8,"rationale":"미래 인재상 통찰이 매우 응축됨."},{"segmentIndex":31,"text":"I do think that there is a future where I think by the end of the year what we're going to start to see is these start to get even murkier where I think in some places the title software engineer is going to start to go away and it's just going to be replaced by builder or maybe it's just everyone's going to be a product manager and everyone codes or something like this.","startTime":2587.96,"endTime":2606.2,"durationSeconds":18,"level":"C1","overallScore":8.8,"rationale":"직함 변화 전망이 매우 선명함."},{"segmentIndex":66,"text":"It's whatever people are doing if you can make that a little bit easier then that's just going to be a much better product that people enjoy more and this is just this principle of latent demand which I think is just the single most important principle in product.","startTime":2836.2,"endTime":2849.84,"durationSeconds":14,"level":"C1","overallScore":8.8,"rationale":"핵심 제품 원칙을 강하게 일반화함."},{"segmentIndex":68,"text":"Latent demand is this idea that if you build a product in a way that can be hacked or can be kind of mis [clears throat] used by people in a way it wasn't really designed for it to do kind of something that they want to do then this helps you as the product builder learn where to take the product next.","startTime":2857.64,"endTime":2875.08,"durationSeconds":17,"level":"C1","overallScore":9,"rationale":"개념 정의와 시사점이 매우 선명하다."},{"segmentIndex":8,"text":"So when you see people abusing the product in this way, using it in a way that it wasn't designed in order to do something that is useful for them, it's just such a strong indicator that you should just build a product and people are going to like that.","startTime":3036.24,"endTime":3047.2,"durationSeconds":11,"level":"C1","overallScore":9,"rationale":"제품 통찰이 강하고 표현도 풍부함."},{"segmentIndex":44,"text":"But this is kind of the way that we learn, um, both on the product side and on the safety side is we have to release things a little bit earlier than we think so that we can get the feedback, so that we can talk to users.","startTime":3230.04,"endTime":3238.68,"durationSeconds":9,"level":"C1","overallScore":9,"rationale":"학습 원칙이 선명하고 표현도 풍부함."},{"segmentIndex":45,"text":"But actually almost always you get better results if you just give the model tools, you give it a goal, and you let it figure it out.","startTime":3846.28,"endTime":3851.52,"durationSeconds":5,"level":"B2","overallScore":8.8,"rationale":"핵심 원칙과 실용 표현이 함께 강하다."},{"segmentIndex":53,"text":"Uh and it's actually a really simple idea. His idea was that the more general model will always outperform the more specific model.","startTime":3895.4,"endTime":3902.12,"durationSeconds":7,"level":"B2","overallScore":8.8,"rationale":"일반 모델 우위를 압축한 핵심 통찰."},{"segmentIndex":79,"text":"But if you build for the model 6 months out, when that model comes out, you're just going to hit the ground running, and the product is going to click and start to work.","startTime":4026.56,"endTime":4035.84,"durationSeconds":9,"level":"B2","overallScore":9,"rationale":"행동 지침과 표현이 모두 강함."},{"segmentIndex":48,"text":"And they're just following the momentum and not thinking about it. I think the best results that I see are people thinking from first principles and just developing their own common sense.","startTime":5069.56,"endTime":5077.64,"durationSeconds":8,"level":"B2","overallScore":8.8,"rationale":"성과 원인을 원칙 수준에서 설명함."},{"segmentIndex":4,"text":"I have never enjoyed coding as much as I do today because I don't have to deal with all the minutiae.","startTime":13.32,"endTime":18.04,"durationSeconds":5,"level":"B2","overallScore":7.6,"rationale":"코딩 만족도의 이유를 설명해 통찰적."},{"segmentIndex":11,"text":"The title software engineer is going to start [music] to go away. It's just going to be replaced by builder and it's going to be painful for a lot of people.","startTime":47.6,"endTime":58.39,"durationSeconds":11,"level":"B2","overallScore":7.8,"rationale":"직함 변화 전망과 여파를 압축 제시."},{"segmentIndex":55,"text":"Um and so the this kind of like mission drivenness just really, really resonated with me, and I just know personally it's something I need in order to be happy. Um and I that's just the thing that I really missed.","startTime":311.68,"endTime":321,"durationSeconds":9,"level":"C1","overallScore":8,"rationale":"가치관과 행복 조건을 깊이 있게 설명."},{"segmentIndex":56,"text":"And I found that, you know, whatever the work might be, no matter how exciting, even if it's building a really cool product, it's just not really a substitute for that.","startTime":321,"endTime":330.6,"durationSeconds":10,"level":"C1","overallScore":7.8,"rationale":"일의 의미에 대한 보편적 통찰이 있음."},{"segmentIndex":74,"text":"And you know, for Anthropic for a long time, we were building the models in this way that kind of fit our mental model of the way that we build safe AGI, where the model starts by being really good at coding, then it gets really good at tool use, then it gets really good at computer use.","startTime":443.919,"endTime":457.8,"durationSeconds":14,"level":"C1","overallScore":8,"rationale":"제품 철학과 발전 경로 통찰이 크다."}],"generatedAt":"2026-06-24T23:59:33.062Z","keyClipsTotalSec":1420},{"videoId":"We7BZVKbCVw","chunkIndex":2,"totalChunks":9,"title":"Head of Claude Code: What happens after coding is solved | Boris Cherny — Part 3 of 9","thumbnail":"https://i.ytimg.com/vi/We7BZVKbCVw/maxresdefault.jpg","duration":5265,"uploader":"Lenny's Podcast","youtubeUrl":"https://www.youtube.com/watch?v=We7BZVKbCVw","keywords":["ai-coding","developer-productivity","software-engineering","code-review","llm-tools","startup","automation","engineering-management","type-systems","functional-programming"],"normalizedKeywords":["엔지니어링","비즈니스·전략","기술 트렌드"],"targetAudience":[{"who":"엔지니어","why":"Claude를 활용한 개발 생산성, 디버깅, 리뷰 자동화의 실제 원칙을 배울 수 있음"},{"who":"스타트업 창업자","why":"적은 인원과 빠른 실행으로 AI를 제품·조직에 녹이는 운영 철학이 유용함"},{"who":"기술 리더","why":"토큰 사용, 인력 배치, 속도 우선 문화 같은 팀 운영 원칙을 참고할 수 있음"}],"normalizedAudience":["엔지니어·개발자","창업자·스타트업","디자인 리더·CXO"],"summary":"이 영상은 Claude Code가 실제로 개발 조직의 병목을 어떻게 바꾸고 있는지에 대한 Boris Cherny의 실전 경험을 중심으로 전개된다. 단순히 코딩을 더 빨리 하는 수준이 아니라, 코드 작성·PR 생성·버그 디버깅·코드 리뷰까지 AI가 넓게 자동화하면서 엔지니어 생산성이 과거와 비교할 수 없을 만큼 커졌다고 주장한다. 특히 모델이 자주 바뀌므로 과거의 감각에 머무르지 말고, 지금 시점의 능력을 기준으로 다시 사고해야 한다는 점을 강하게 강조한다.\n\n또한 팀 운영 철학으로는 '적당히 underfunding 해서 더 스스로 Claudify하게 만들기', '가능하면 오늘 바로 실행하기', '초기에는 토큰을 아끼지 말고 충분히 써 보라'는 원칙을 제시한다. 결국 이 영상의 핵심은 AI 시대에는 적은 인원, 빠른 실험, 넉넉한 모델 사용이 오히려 더 좋은 결과를 만든다는 것이다. 마지막에는 자신이 코딩을 시작한 계기와 프로그래밍의 미학까지 이야기하며, 코딩은 목적이 아니라 무엇인가를 만들기 위한 도구라는 관점을 다시 확인한다.","insights":["AI 시대의 병목은 코딩이 아니라 리뷰와 판단이다.","최신 모델에 맞춰 사고를 갱신하지 않으면 금세 뒤처진다.","초기엔 비용 최적화보다 충분한 실험이 더 중요하다.","적은 인원과 빠른 실행이 Claude 활용도를 오히려 높인다.","코딩의 가치는 기술 자체보다 무언가를 만들어내는 데 있다."],"keyClips":[{"clipId":"We7BZVKbCVw:c2:1-8","startSegmentIndex":1,"endSegmentIndex":8,"startTime":1201.27,"endTime":1242.28,"durationSeconds":41,"preview":"코딩보다 리뷰 병목","mustSee":false},{"clipId":"We7BZVKbCVw:c2:10-16","startSegmentIndex":10,"endSegmentIndex":16,"startTime":1250.6,"endTime":1295.92,"durationSeconds":45.3,"preview":"생산성 폭증의 현실","mustSee":false},{"clipId":"We7BZVKbCVw:c2:25-38","startSegmentIndex":25,"endSegmentIndex":38,"startTime":1345.28,"endTime":1443.2,"durationSeconds":97.9,"preview":"모델이 바꾸는 사고","mustSee":true},{"clipId":"We7BZVKbCVw:c2:42-66","startSegmentIndex":42,"endSegmentIndex":66,"startTime":1460.16,"endTime":1615.84,"durationSeconds":155.7,"preview":"적게 두고 더 빠르게","mustSee":false},{"clipId":"We7BZVKbCVw:c2:77-98","startSegmentIndex":77,"endSegmentIndex":98,"startTime":1678.28,"endTime":1798.68,"durationSeconds":120.4,"preview":"코딩은 수단이다","mustSee":false}],"curatedSegments":[{"segmentIndex":6,"text":"In a year or two it's not going [music] to matter. Coding is virtually solved. I imagine a world where everyone is able to program.","startTime":22.72,"endTime":27.88,"durationSeconds":5,"level":"B2","overallScore":8.4,"rationale":"미래 변화에 대한 강한 관점 제시."},{"segmentIndex":83,"text":"It acts in the world. Um, and I think now with Co-work, we're starting to see the transition for non-technical folks also. And I think the craziest thing for me isn't even the number that we're at right now, but the pace at which we're growing because if you look at Claude code's growth rate kind of across any metric, it's continuing to accelerate. And I think honestly, that's just for me as an engineer, I find that to do good work, you really have to understand the layer under the layer at which you work.","startTime":501.88,"endTime":508.8,"durationSeconds":7,"level":"C1","overallScore":9,"rationale":"성장 해석과 일의 원칙이 매우 통찰적."},{"segmentIndex":26,"text":"It's just And when I think about it, kind of part of the reason Quad Code works is this idea of latent demand where we bring the tool to where people are and it makes existing workflows a little bit easier. But also because it's a it's in a terminal, it's like a little surprising.","startTime":732.52,"endTime":745.16,"durationSeconds":13,"level":"C1","overallScore":9,"rationale":"제품 성공 원인을 원리 수준에서 설명함."},{"segmentIndex":48,"text":"Um, so we really just think in exponentials. And if you kind of look at the exponential of the percent of code that was written by Claude at that point, and if you just trace the line, it's pretty obvious we're going to cross 100% by the end of the year, even if it just does not match intuition at all.","startTime":876.76,"endTime":890.92,"durationSeconds":14,"level":"C1","overallScore":9,"rationale":"지수 추세 해석이 통찰·표현 모두 강함."},{"segmentIndex":54,"text":"Um you just have to give people space. You have to give them maybe the word is like safety. So, it's like psychological safety that it's okay to fail.","startTime":926.8,"endTime":931,"durationSeconds":4,"level":"C1","overallScore":8.8,"rationale":"혁신 문화의 조건을 구체적으로 제시함."},{"segmentIndex":24,"text":"So, I think a lot of the people that will be rewarded most over the next few years, they won't just be AI native and they don't just know how to use these tools really well, but also they're curious and they're generalists and they cross over multiple disciplines and can think about the broader problem they're solving rather than just engineering part of it.","startTime":2530.2,"endTime":2538.12,"durationSeconds":8,"level":"C1","overallScore":8.8,"rationale":"미래 인재상 통찰이 매우 응축됨."},{"segmentIndex":31,"text":"I do think that there is a future where I think by the end of the year what we're going to start to see is these start to get even murkier where I think in some places the title software engineer is going to start to go away and it's just going to be replaced by builder or maybe it's just everyone's going to be a product manager and everyone codes or something like this.","startTime":2587.96,"endTime":2606.2,"durationSeconds":18,"level":"C1","overallScore":8.8,"rationale":"직함 변화 전망이 매우 선명함."},{"segmentIndex":66,"text":"It's whatever people are doing if you can make that a little bit easier then that's just going to be a much better product that people enjoy more and this is just this principle of latent demand which I think is just the single most important principle in product.","startTime":2836.2,"endTime":2849.84,"durationSeconds":14,"level":"C1","overallScore":8.8,"rationale":"핵심 제품 원칙을 강하게 일반화함."},{"segmentIndex":68,"text":"Latent demand is this idea that if you build a product in a way that can be hacked or can be kind of mis [clears throat] used by people in a way it wasn't really designed for it to do kind of something that they want to do then this helps you as the product builder learn where to take the product next.","startTime":2857.64,"endTime":2875.08,"durationSeconds":17,"level":"C1","overallScore":9,"rationale":"개념 정의와 시사점이 매우 선명하다."},{"segmentIndex":8,"text":"So when you see people abusing the product in this way, using it in a way that it wasn't designed in order to do something that is useful for them, it's just such a strong indicator that you should just build a product and people are going to like that.","startTime":3036.24,"endTime":3047.2,"durationSeconds":11,"level":"C1","overallScore":9,"rationale":"제품 통찰이 강하고 표현도 풍부함."},{"segmentIndex":44,"text":"But this is kind of the way that we learn, um, both on the product side and on the safety side is we have to release things a little bit earlier than we think so that we can get the feedback, so that we can talk to users.","startTime":3230.04,"endTime":3238.68,"durationSeconds":9,"level":"C1","overallScore":9,"rationale":"학습 원칙이 선명하고 표현도 풍부함."},{"segmentIndex":45,"text":"But actually almost always you get better results if you just give the model tools, you give it a goal, and you let it figure it out.","startTime":3846.28,"endTime":3851.52,"durationSeconds":5,"level":"B2","overallScore":8.8,"rationale":"핵심 원칙과 실용 표현이 함께 강하다."},{"segmentIndex":53,"text":"Uh and it's actually a really simple idea. His idea was that the more general model will always outperform the more specific model.","startTime":3895.4,"endTime":3902.12,"durationSeconds":7,"level":"B2","overallScore":8.8,"rationale":"일반 모델 우위를 압축한 핵심 통찰."},{"segmentIndex":79,"text":"But if you build for the model 6 months out, when that model comes out, you're just going to hit the ground running, and the product is going to click and start to work.","startTime":4026.56,"endTime":4035.84,"durationSeconds":9,"level":"B2","overallScore":9,"rationale":"행동 지침과 표현이 모두 강함."},{"segmentIndex":48,"text":"And they're just following the momentum and not thinking about it. I think the best results that I see are people thinking from first principles and just developing their own common sense.","startTime":5069.56,"endTime":5077.64,"durationSeconds":8,"level":"B2","overallScore":8.8,"rationale":"성과 원인을 원칙 수준에서 설명함."},{"segmentIndex":4,"text":"I have never enjoyed coding as much as I do today because I don't have to deal with all the minutiae.","startTime":13.32,"endTime":18.04,"durationSeconds":5,"level":"B2","overallScore":7.6,"rationale":"코딩 만족도의 이유를 설명해 통찰적."},{"segmentIndex":11,"text":"The title software engineer is going to start [music] to go away. It's just going to be replaced by builder and it's going to be painful for a lot of people.","startTime":47.6,"endTime":58.39,"durationSeconds":11,"level":"B2","overallScore":7.8,"rationale":"직함 변화 전망과 여파를 압축 제시."},{"segmentIndex":55,"text":"Um and so the this kind of like mission drivenness just really, really resonated with me, and I just know personally it's something I need in order to be happy. Um and I that's just the thing that I really missed.","startTime":311.68,"endTime":321,"durationSeconds":9,"level":"C1","overallScore":8,"rationale":"가치관과 행복 조건을 깊이 있게 설명."},{"segmentIndex":56,"text":"And I found that, you know, whatever the work might be, no matter how exciting, even if it's building a really cool product, it's just not really a substitute for that.","startTime":321,"endTime":330.6,"durationSeconds":10,"level":"C1","overallScore":7.8,"rationale":"일의 의미에 대한 보편적 통찰이 있음."},{"segmentIndex":74,"text":"And you know, for Anthropic for a long time, we were building the models in this way that kind of fit our mental model of the way that we build safe AGI, where the model starts by being really good at coding, then it gets really good at tool use, then it gets really good at computer use.","startTime":443.919,"endTime":457.8,"durationSeconds":14,"level":"C1","overallScore":8,"rationale":"제품 철학과 발전 경로 통찰이 크다."}],"generatedAt":"2026-06-24T23:59:54.127Z","keyClipsTotalSec":1420},{"videoId":"We7BZVKbCVw","chunkIndex":3,"totalChunks":9,"title":"Head of Claude Code: What happens after coding is solved | Boris Cherny — Part 4 of 9","thumbnail":"https://i.ytimg.com/vi/We7BZVKbCVw/maxresdefault.jpg","duration":5265,"uploader":"Lenny's Podcast","youtubeUrl":"https://www.youtube.com/watch?v=We7BZVKbCVw","keywords":["ai","software-engineering","agents","automation","future-of-work","coding","productivity","education","history","anthropic"],"normalizedKeywords":["엔지니어링","기술 트렌드","커리어·성장"],"targetAudience":[{"who":"엔지니어","why":"AI 에이전트가 개발자의 역할과 숙련도 개념을 어떻게 바꾸는지 다룸"},{"who":"프로덕트 매니저","why":"엔지니어링 인접 직무까지 AI가 확장될 미래를 이해하는 데 유용함"},{"who":"학생","why":"앞으로 코딩을 배워야 하는지, 무엇을 배워야 하는지 고민할 때 참고됨"}],"normalizedAudience":["엔지니어·개발자","프로덕트 매니저·기획자","학생·주니어"],"summary":"이 영상은 AI 코딩 에이전트가 개발자의 일을 어떻게 바꾸고 있는지, 그리고 그 변화가 앞으로 다른 지식노동까지 어떻게 확장될지를 다룬다. 화자는 코딩을 더 이상 목적이 아니라 도구로 보며, AI가 반복적이고 지루한 구현 작업을 덜어 주는 만큼 엔지니어는 문제 정의, 사용자 이해, 시스템 설계에 더 집중하게 된다고 말한다.\n\n또한 이런 변화가 스크라이브와 인쇄술의 관계처럼, 소수의 전문 기술이 더 넓게 민주화되는 흐름이라고 설명한다. 단기적으로는 아직 하위 계층을 이해하는 것이 중요하지만, 가까운 미래에는 그 필요가 크게 줄어들 수 있다고 보고, 제품·디자인·데이터사이언스 같은 인접 직무까지 영향이 퍼질 것이라 전망한다.","insights":["코딩은 목적이 아니라 문제를 푸는 도구다.","AI가 지루한 구현을 대체할수록 상위 사고가 더 중요해진다.","새 도구 시대엔 레이어를 아는 능력보다 활용이 더 중요해진다.","기술 변화는 숙련의 소멸이 아니라 역할의 재배치다.","AI 확산은 엔지니어링 밖의 지식노동까지 밀어 올린다."],"keyClips":[{"clipId":"We7BZVKbCVw:c3:1-5","startSegmentIndex":1,"endSegmentIndex":5,"startTime":1800.95,"endTime":1834.56,"durationSeconds":33.6,"preview":"코딩은 도구일 뿐","mustSee":false},{"clipId":"We7BZVKbCVw:c3:6-18","startSegmentIndex":6,"endSegmentIndex":18,"startTime":1834.56,"endTime":1896.08,"durationSeconds":61.5,"preview":"숙련 저하의 재해석","mustSee":false},{"clipId":"We7BZVKbCVw:c3:19-25","startSegmentIndex":19,"endSegmentIndex":25,"startTime":1896.08,"endTime":1935.76,"durationSeconds":39.7,"preview":"에이전트의 의미","mustSee":false},{"clipId":"We7BZVKbCVw:c3:32-52","startSegmentIndex":32,"endSegmentIndex":52,"startTime":1974.6,"endTime":2122.36,"durationSeconds":147.8,"preview":"인쇄술 같은 전환","mustSee":true},{"clipId":"We7BZVKbCVw:c3:53-60","startSegmentIndex":53,"endSegmentIndex":60,"startTime":2122.36,"endTime":2162.48,"durationSeconds":40.1,"preview":"막힘을 푸는 속도","mustSee":false},{"clipId":"We7BZVKbCVw:c3:63-81","startSegmentIndex":63,"endSegmentIndex":81,"startTime":2172,"endTime":2299.16,"durationSeconds":127.2,"preview":"확산되는 직무 변화","mustSee":true},{"clipId":"We7BZVKbCVw:c3:82-97","startSegmentIndex":82,"endSegmentIndex":97,"startTime":2299.16,"endTime":2407.44,"durationSeconds":108.3,"preview":"일이 더 늘어나는 미래","mustSee":false}],"curatedSegments":[{"segmentIndex":6,"text":"In a year or two it's not going [music] to matter. Coding is virtually solved. I imagine a world where everyone is able to program.","startTime":22.72,"endTime":27.88,"durationSeconds":5,"level":"B2","overallScore":8.4,"rationale":"미래 변화에 대한 강한 관점 제시."},{"segmentIndex":83,"text":"It acts in the world. Um, and I think now with Co-work, we're starting to see the transition for non-technical folks also. And I think the craziest thing for me isn't even the number that we're at right now, but the pace at which we're growing because if you look at Claude code's growth rate kind of across any metric, it's continuing to accelerate. And I think honestly, that's just for me as an engineer, I find that to do good work, you really have to understand the layer under the layer at which you work.","startTime":501.88,"endTime":508.8,"durationSeconds":7,"level":"C1","overallScore":9,"rationale":"성장 해석과 일의 원칙이 매우 통찰적."},{"segmentIndex":26,"text":"It's just And when I think about it, kind of part of the reason Quad Code works is this idea of latent demand where we bring the tool to where people are and it makes existing workflows a little bit easier. But also because it's a it's in a terminal, it's like a little surprising.","startTime":732.52,"endTime":745.16,"durationSeconds":13,"level":"C1","overallScore":9,"rationale":"제품 성공 원인을 원리 수준에서 설명함."},{"segmentIndex":48,"text":"Um, so we really just think in exponentials. And if you kind of look at the exponential of the percent of code that was written by Claude at that point, and if you just trace the line, it's pretty obvious we're going to cross 100% by the end of the year, even if it just does not match intuition at all.","startTime":876.76,"endTime":890.92,"durationSeconds":14,"level":"C1","overallScore":9,"rationale":"지수 추세 해석이 통찰·표현 모두 강함."},{"segmentIndex":54,"text":"Um you just have to give people space. You have to give them maybe the word is like safety. So, it's like psychological safety that it's okay to fail.","startTime":926.8,"endTime":931,"durationSeconds":4,"level":"C1","overallScore":8.8,"rationale":"혁신 문화의 조건을 구체적으로 제시함."},{"segmentIndex":24,"text":"So, I think a lot of the people that will be rewarded most over the next few years, they won't just be AI native and they don't just know how to use these tools really well, but also they're curious and they're generalists and they cross over multiple disciplines and can think about the broader problem they're solving rather than just engineering part of it.","startTime":2530.2,"endTime":2538.12,"durationSeconds":8,"level":"C1","overallScore":8.8,"rationale":"미래 인재상 통찰이 매우 응축됨."},{"segmentIndex":31,"text":"I do think that there is a future where I think by the end of the year what we're going to start to see is these start to get even murkier where I think in some places the title software engineer is going to start to go away and it's just going to be replaced by builder or maybe it's just everyone's going to be a product manager and everyone codes or something like this.","startTime":2587.96,"endTime":2606.2,"durationSeconds":18,"level":"C1","overallScore":8.8,"rationale":"직함 변화 전망이 매우 선명함."},{"segmentIndex":66,"text":"It's whatever people are doing if you can make that a little bit easier then that's just going to be a much better product that people enjoy more and this is just this principle of latent demand which I think is just the single most important principle in product.","startTime":2836.2,"endTime":2849.84,"durationSeconds":14,"level":"C1","overallScore":8.8,"rationale":"핵심 제품 원칙을 강하게 일반화함."},{"segmentIndex":68,"text":"Latent demand is this idea that if you build a product in a way that can be hacked or can be kind of mis [clears throat] used by people in a way it wasn't really designed for it to do kind of something that they want to do then this helps you as the product builder learn where to take the product next.","startTime":2857.64,"endTime":2875.08,"durationSeconds":17,"level":"C1","overallScore":9,"rationale":"개념 정의와 시사점이 매우 선명하다."},{"segmentIndex":8,"text":"So when you see people abusing the product in this way, using it in a way that it wasn't designed in order to do something that is useful for them, it's just such a strong indicator that you should just build a product and people are going to like that.","startTime":3036.24,"endTime":3047.2,"durationSeconds":11,"level":"C1","overallScore":9,"rationale":"제품 통찰이 강하고 표현도 풍부함."},{"segmentIndex":44,"text":"But this is kind of the way that we learn, um, both on the product side and on the safety side is we have to release things a little bit earlier than we think so that we can get the feedback, so that we can talk to users.","startTime":3230.04,"endTime":3238.68,"durationSeconds":9,"level":"C1","overallScore":9,"rationale":"학습 원칙이 선명하고 표현도 풍부함."},{"segmentIndex":45,"text":"But actually almost always you get better results if you just give the model tools, you give it a goal, and you let it figure it out.","startTime":3846.28,"endTime":3851.52,"durationSeconds":5,"level":"B2","overallScore":8.8,"rationale":"핵심 원칙과 실용 표현이 함께 강하다."},{"segmentIndex":53,"text":"Uh and it's actually a really simple idea. His idea was that the more general model will always outperform the more specific model.","startTime":3895.4,"endTime":3902.12,"durationSeconds":7,"level":"B2","overallScore":8.8,"rationale":"일반 모델 우위를 압축한 핵심 통찰."},{"segmentIndex":79,"text":"But if you build for the model 6 months out, when that model comes out, you're just going to hit the ground running, and the product is going to click and start to work.","startTime":4026.56,"endTime":4035.84,"durationSeconds":9,"level":"B2","overallScore":9,"rationale":"행동 지침과 표현이 모두 강함."},{"segmentIndex":48,"text":"And they're just following the momentum and not thinking about it. I think the best results that I see are people thinking from first principles and just developing their own common sense.","startTime":5069.56,"endTime":5077.64,"durationSeconds":8,"level":"B2","overallScore":8.8,"rationale":"성과 원인을 원칙 수준에서 설명함."},{"segmentIndex":4,"text":"I have never enjoyed coding as much as I do today because I don't have to deal with all the minutiae.","startTime":13.32,"endTime":18.04,"durationSeconds":5,"level":"B2","overallScore":7.6,"rationale":"코딩 만족도의 이유를 설명해 통찰적."},{"segmentIndex":11,"text":"The title software engineer is going to start [music] to go away. It's just going to be replaced by builder and it's going to be painful for a lot of people.","startTime":47.6,"endTime":58.39,"durationSeconds":11,"level":"B2","overallScore":7.8,"rationale":"직함 변화 전망과 여파를 압축 제시."},{"segmentIndex":55,"text":"Um and so the this kind of like mission drivenness just really, really resonated with me, and I just know personally it's something I need in order to be happy. Um and I that's just the thing that I really missed.","startTime":311.68,"endTime":321,"durationSeconds":9,"level":"C1","overallScore":8,"rationale":"가치관과 행복 조건을 깊이 있게 설명."},{"segmentIndex":56,"text":"And I found that, you know, whatever the work might be, no matter how exciting, even if it's building a really cool product, it's just not really a substitute for that.","startTime":321,"endTime":330.6,"durationSeconds":10,"level":"C1","overallScore":7.8,"rationale":"일의 의미에 대한 보편적 통찰이 있음."},{"segmentIndex":74,"text":"And you know, for Anthropic for a long time, we were building the models in this way that kind of fit our mental model of the way that we build safe AGI, where the model starts by being really good at coding, then it gets really good at tool use, then it gets really good at computer use.","startTime":443.919,"endTime":457.8,"durationSeconds":14,"level":"C1","overallScore":8,"rationale":"제품 철학과 발전 경로 통찰이 크다."}],"generatedAt":"2026-06-25T00:00:25.127Z","keyClipsTotalSec":1420},{"videoId":"We7BZVKbCVw","chunkIndex":4,"totalChunks":9,"title":"Head of Claude Code: What happens after coding is solved | Boris Cherny — Part 5 of 9","thumbnail":"https://i.ytimg.com/vi/We7BZVKbCVw/maxresdefault.jpg","duration":5265,"uploader":"Lenny's Podcast","youtubeUrl":"https://www.youtube.com/watch?v=We7BZVKbCVw","keywords":["ai","software-development","product-management","future-of-work","generalist","latent-demand","saas","career-growth","technology-trends"],"normalizedKeywords":["프로덕트","엔지니어링","기술 트렌드"],"targetAudience":[{"who":"프로덕트 매니저","why":"AI 도구로 역할 경계가 흐려질 때 PM이 어떤 역량을 키워야 하는지 보여줌"},{"who":"엔지니어","why":"AI 시대에 더 가치 있는 개발자 역량이 무엇인지 구체적으로 제시함"},{"who":"디자이너","why":"코딩 가능한 디자이너와 새 워크플로가 어떤 변화를 만드는지 참고할 수 있음"},{"who":"초기 단계 창업자","why":"latent demand를 발견해 새 제품 기회를 찾는 관점을 얻을 수 있음"}],"normalizedAudience":["프로덕트 매니저·기획자","엔지니어·개발자","창업자·스타트업"],"summary":"이 구간은 AI가 코딩을 '더 쉽게' 만드는 수준을 넘어, 소프트웨어를 만드는 사람 자체를 바꾸고 있다는 관점을 다룬다. 화자는 앞으로는 누구나 프로그램을 다루는 세상이 오고, 그래서 엔지니어·PM·디자이너의 경계가 점점 흐려지며 'builder' 같은 더 넓은 역할로 재편될 수 있다고 본다. 그 변화 속에서 살아남고 성장하려면 최신 AI 도구를 적극적으로 실험하고, 한 분야만 깊게 파는 전문가보다 여러 분야를 넘나드는 일반가형 인재가 되어야 한다고 조언한다.\n\n또한 사람들은 이미 Cloud Code를 코딩 외 용도로도 쓰고 있으며, 이런 'latent demand'가 새 제품 기회로 이어진다고 설명한다. Facebook Marketplace와 Dating 사례를 들어, 사용자가 원래 의도와 다르게 제품을 쓰는 패턴 속에 다음 제품 방향이 숨어 있다고 말한다. 마지막으로 AI 도입 후 PM과 엔지니어는 일의 만족도가 높아졌지만 디자이너는 상대적으로 덜 그렇다는 조사 결과를 언급하며, 새 도구가 직군별 경험을 다르게 바꾸고 있음을 보여준다.","insights":["AI는 직업을 대체하기보다 역할 경계를 먼저 녹인다.","미래에 강한 사람은 한 분야 전문가보다 교차형 일반가다.","도구를 잘 쓰는 것보다 도구를 통해 문제를 넓게 보는 게 중요하다.","제품 기회는 사용자의 우회 사용과 변형 사용에서 드러난다.","latent demand를 보면 '원래 안 만든 문제'가 다음 시장이 된다."],"keyClips":[{"clipId":"We7BZVKbCVw:c4:4-8","startSegmentIndex":4,"endSegmentIndex":8,"startTime":2415.56,"endTime":2440.24,"durationSeconds":24.7,"preview":"모두가 코딩하는 미래","mustSee":false},{"clipId":"We7BZVKbCVw:c4:11-24","startSegmentIndex":11,"endSegmentIndex":24,"startTime":2458.44,"endTime":2538.12,"durationSeconds":79.7,"preview":"일반가가 더 강해진다","mustSee":false},{"clipId":"We7BZVKbCVw:c4:28-31","startSegmentIndex":28,"endSegmentIndex":31,"startTime":2565.96,"endTime":2606.2,"durationSeconds":40.2,"preview":"역할 경계의 붕괴","mustSee":false},{"clipId":"We7BZVKbCVw:c4:60-66","startSegmentIndex":60,"endSegmentIndex":66,"startTime":2792.64,"endTime":2849.84,"durationSeconds":57.2,"preview":"제품은 사람에게 맞춰라","mustSee":true},{"clipId":"We7BZVKbCVw:c4:68-87","startSegmentIndex":68,"endSegmentIndex":87,"startTime":2857.64,"endTime":2997.16,"durationSeconds":139.5,"preview":"우회사용이 시장이다","mustSee":false}],"curatedSegments":[{"segmentIndex":6,"text":"In a year or two it's not going [music] to matter. Coding is virtually solved. I imagine a world where everyone is able to program.","startTime":22.72,"endTime":27.88,"durationSeconds":5,"level":"B2","overallScore":8.4,"rationale":"미래 변화에 대한 강한 관점 제시."},{"segmentIndex":83,"text":"It acts in the world. Um, and I think now with Co-work, we're starting to see the transition for non-technical folks also. And I think the craziest thing for me isn't even the number that we're at right now, but the pace at which we're growing because if you look at Claude code's growth rate kind of across any metric, it's continuing to accelerate. And I think honestly, that's just for me as an engineer, I find that to do good work, you really have to understand the layer under the layer at which you work.","startTime":501.88,"endTime":508.8,"durationSeconds":7,"level":"C1","overallScore":9,"rationale":"성장 해석과 일의 원칙이 매우 통찰적."},{"segmentIndex":26,"text":"It's just And when I think about it, kind of part of the reason Quad Code works is this idea of latent demand where we bring the tool to where people are and it makes existing workflows a little bit easier. But also because it's a it's in a terminal, it's like a little surprising.","startTime":732.52,"endTime":745.16,"durationSeconds":13,"level":"C1","overallScore":9,"rationale":"제품 성공 원인을 원리 수준에서 설명함."},{"segmentIndex":48,"text":"Um, so we really just think in exponentials. And if you kind of look at the exponential of the percent of code that was written by Claude at that point, and if you just trace the line, it's pretty obvious we're going to cross 100% by the end of the year, even if it just does not match intuition at all.","startTime":876.76,"endTime":890.92,"durationSeconds":14,"level":"C1","overallScore":9,"rationale":"지수 추세 해석이 통찰·표현 모두 강함."},{"segmentIndex":54,"text":"Um you just have to give people space. You have to give them maybe the word is like safety. So, it's like psychological safety that it's okay to fail.","startTime":926.8,"endTime":931,"durationSeconds":4,"level":"C1","overallScore":8.8,"rationale":"혁신 문화의 조건을 구체적으로 제시함."},{"segmentIndex":24,"text":"So, I think a lot of the people that will be rewarded most over the next few years, they won't just be AI native and they don't just know how to use these tools really well, but also they're curious and they're generalists and they cross over multiple disciplines and can think about the broader problem they're solving rather than just engineering part of it.","startTime":2530.2,"endTime":2538.12,"durationSeconds":8,"level":"C1","overallScore":8.8,"rationale":"미래 인재상 통찰이 매우 응축됨."},{"segmentIndex":31,"text":"I do think that there is a future where I think by the end of the year what we're going to start to see is these start to get even murkier where I think in some places the title software engineer is going to start to go away and it's just going to be replaced by builder or maybe it's just everyone's going to be a product manager and everyone codes or something like this.","startTime":2587.96,"endTime":2606.2,"durationSeconds":18,"level":"C1","overallScore":8.8,"rationale":"직함 변화 전망이 매우 선명함."},{"segmentIndex":66,"text":"It's whatever people are doing if you can make that a little bit easier then that's just going to be a much better product that people enjoy more and this is just this principle of latent demand which I think is just the single most important principle in product.","startTime":2836.2,"endTime":2849.84,"durationSeconds":14,"level":"C1","overallScore":8.8,"rationale":"핵심 제품 원칙을 강하게 일반화함."},{"segmentIndex":68,"text":"Latent demand is this idea that if you build a product in a way that can be hacked or can be kind of mis [clears throat] used by people in a way it wasn't really designed for it to do kind of something that they want to do then this helps you as the product builder learn where to take the product next.","startTime":2857.64,"endTime":2875.08,"durationSeconds":17,"level":"C1","overallScore":9,"rationale":"개념 정의와 시사점이 매우 선명하다."},{"segmentIndex":8,"text":"So when you see people abusing the product in this way, using it in a way that it wasn't designed in order to do something that is useful for them, it's just such a strong indicator that you should just build a product and people are going to like that.","startTime":3036.24,"endTime":3047.2,"durationSeconds":11,"level":"C1","overallScore":9,"rationale":"제품 통찰이 강하고 표현도 풍부함."},{"segmentIndex":44,"text":"But this is kind of the way that we learn, um, both on the product side and on the safety side is we have to release things a little bit earlier than we think so that we can get the feedback, so that we can talk to users.","startTime":3230.04,"endTime":3238.68,"durationSeconds":9,"level":"C1","overallScore":9,"rationale":"학습 원칙이 선명하고 표현도 풍부함."},{"segmentIndex":45,"text":"But actually almost always you get better results if you just give the model tools, you give it a goal, and you let it figure it out.","startTime":3846.28,"endTime":3851.52,"durationSeconds":5,"level":"B2","overallScore":8.8,"rationale":"핵심 원칙과 실용 표현이 함께 강하다."},{"segmentIndex":53,"text":"Uh and it's actually a really simple idea. His idea was that the more general model will always outperform the more specific model.","startTime":3895.4,"endTime":3902.12,"durationSeconds":7,"level":"B2","overallScore":8.8,"rationale":"일반 모델 우위를 압축한 핵심 통찰."},{"segmentIndex":79,"text":"But if you build for the model 6 months out, when that model comes out, you're just going to hit the ground running, and the product is going to click and start to work.","startTime":4026.56,"endTime":4035.84,"durationSeconds":9,"level":"B2","overallScore":9,"rationale":"행동 지침과 표현이 모두 강함."},{"segmentIndex":48,"text":"And they're just following the momentum and not thinking about it. I think the best results that I see are people thinking from first principles and just developing their own common sense.","startTime":5069.56,"endTime":5077.64,"durationSeconds":8,"level":"B2","overallScore":8.8,"rationale":"성과 원인을 원칙 수준에서 설명함."},{"segmentIndex":4,"text":"I have never enjoyed coding as much as I do today because I don't have to deal with all the minutiae.","startTime":13.32,"endTime":18.04,"durationSeconds":5,"level":"B2","overallScore":7.6,"rationale":"코딩 만족도의 이유를 설명해 통찰적."},{"segmentIndex":11,"text":"The title software engineer is going to start [music] to go away. It's just going to be replaced by builder and it's going to be painful for a lot of people.","startTime":47.6,"endTime":58.39,"durationSeconds":11,"level":"B2","overallScore":7.8,"rationale":"직함 변화 전망과 여파를 압축 제시."},{"segmentIndex":55,"text":"Um and so the this kind of like mission drivenness just really, really resonated with me, and I just know personally it's something I need in order to be happy. Um and I that's just the thing that I really missed.","startTime":311.68,"endTime":321,"durationSeconds":9,"level":"C1","overallScore":8,"rationale":"가치관과 행복 조건을 깊이 있게 설명."},{"segmentIndex":56,"text":"And I found that, you know, whatever the work might be, no matter how exciting, even if it's building a really cool product, it's just not really a substitute for that.","startTime":321,"endTime":330.6,"durationSeconds":10,"level":"C1","overallScore":7.8,"rationale":"일의 의미에 대한 보편적 통찰이 있음."},{"segmentIndex":74,"text":"And you know, for Anthropic for a long time, we were building the models in this way that kind of fit our mental model of the way that we build safe AGI, where the model starts by being really good at coding, then it gets really good at tool use, then it gets really good at computer use.","startTime":443.919,"endTime":457.8,"durationSeconds":14,"level":"C1","overallScore":8,"rationale":"제품 철학과 발전 경로 통찰이 크다."}],"generatedAt":"2026-06-25T00:00:46.945Z","keyClipsTotalSec":1420},{"videoId":"We7BZVKbCVw","chunkIndex":5,"totalChunks":9,"title":"Head of Claude Code: What happens after coding is solved | Boris Cherny — Part 6 of 9","thumbnail":"https://i.ytimg.com/vi/We7BZVKbCVw/maxresdefault.jpg","duration":5265,"uploader":"Lenny's Podcast","youtubeUrl":"https://www.youtube.com/watch?v=We7BZVKbCVw","keywords":["artificial-intelligence","llm","agents","product-design","ai-safety","interpretability","startup","engineering","open-source","research"],"normalizedKeywords":["프로덕트","엔지니어링","기술 트렌드"],"targetAudience":[{"who":"AI 제품팀","why":"모델이 실제로 뭘 하려는지 보고 제품을 설계하는 관점을 배울 수 있음"},{"who":"안전·정책 담당자","why":"출시 전후의 안전 검증과 현장 학습 방식이 직접적으로 연결됨"},{"who":"엔지니어 리더","why":"에이전트 제품의 구조, 가드레일, 샌드박스 설계가 핵심임"}],"normalizedAudience":["엔지니어·개발자","프로덕트 매니저·기획자","창업자·스타트업"],"summary":"이 영상은 Claude Code와 Claude Work를 만든 경험을 바탕으로, AI 제품을 설계할 때는 '사람이 무엇을 하는지'뿐 아니라 '모델이 무엇을 하려는지'를 봐야 한다는 점을 강조한다. 특히 사용자가 제품을 엉뚱한 방식으로 활용하는 현상과 모델의 행동을 그대로 노출했을 때 드러나는 latent demand가, 새로운 제품 기회를 찾는 중요한 신호라고 설명한다.\n\n동시에 Anthropic이 왜 제품을 비교적 일찍 공개하는지, 그리고 그 배경에 안전 연구가 어떻게 연결되는지도 다룬다. alignment, evals, real-world behavior라는 세 층위의 안전 검증과 mechanistic interpretability, 오픈소스 sandbox 같은 접근을 통해, 빠른 출시와 안전 확보를 함께 밀어붙이는 전략을 보여준다.","insights":["사용자의 우회 사용은 새로운 제품 수요의 가장 강한 신호다.","최신 AI 제품은 '사람'보다 '모델의 의도'를 더 잘 봐야 한다.","모델에 최소한의 도구만 주고 자율성을 높일수록 제품성이 드러난다.","안전은 실험실 평가만으로 부족하고, 실제 사용 데이터가 필수다.","빠른 공개는 품질 저하가 아니라 학습 속도를 높이는 전략이다."],"keyClips":[{"clipId":"We7BZVKbCVw:c5:1-20","startSegmentIndex":1,"endSegmentIndex":20,"startTime":3000.79,"endTime":3115.2,"durationSeconds":114.4,"preview":"우회 사용이 신호다","mustSee":true},{"clipId":"We7BZVKbCVw:c5:13-20","startSegmentIndex":13,"endSegmentIndex":20,"startTime":3070.52,"endTime":3115.2,"durationSeconds":44.7,"preview":"모델 중심 설계","mustSee":false},{"clipId":"We7BZVKbCVw:c5:25-45","startSegmentIndex":25,"endSegmentIndex":45,"startTime":3135.52,"endTime":3245.52,"durationSeconds":110,"preview":"조기출시는 학습이다","mustSee":true},{"clipId":"We7BZVKbCVw:c5:49-70","startSegmentIndex":49,"endSegmentIndex":70,"startTime":3267.64,"endTime":3391.32,"durationSeconds":123.7,"preview":"안전의 세 겹","mustSee":false},{"clipId":"We7BZVKbCVw:c5:79-104","startSegmentIndex":79,"endSegmentIndex":104,"startTime":3435.92,"endTime":3576.88,"durationSeconds":141,"preview":"모델 속을 들여다보기","mustSee":true}],"curatedSegments":[{"segmentIndex":6,"text":"In a year or two it's not going [music] to matter. Coding is virtually solved. I imagine a world where everyone is able to program.","startTime":22.72,"endTime":27.88,"durationSeconds":5,"level":"B2","overallScore":8.4,"rationale":"미래 변화에 대한 강한 관점 제시."},{"segmentIndex":83,"text":"It acts in the world. Um, and I think now with Co-work, we're starting to see the transition for non-technical folks also. And I think the craziest thing for me isn't even the number that we're at right now, but the pace at which we're growing because if you look at Claude code's growth rate kind of across any metric, it's continuing to accelerate. And I think honestly, that's just for me as an engineer, I find that to do good work, you really have to understand the layer under the layer at which you work.","startTime":501.88,"endTime":508.8,"durationSeconds":7,"level":"C1","overallScore":9,"rationale":"성장 해석과 일의 원칙이 매우 통찰적."},{"segmentIndex":26,"text":"It's just And when I think about it, kind of part of the reason Quad Code works is this idea of latent demand where we bring the tool to where people are and it makes existing workflows a little bit easier. But also because it's a it's in a terminal, it's like a little surprising.","startTime":732.52,"endTime":745.16,"durationSeconds":13,"level":"C1","overallScore":9,"rationale":"제품 성공 원인을 원리 수준에서 설명함."},{"segmentIndex":48,"text":"Um, so we really just think in exponentials. And if you kind of look at the exponential of the percent of code that was written by Claude at that point, and if you just trace the line, it's pretty obvious we're going to cross 100% by the end of the year, even if it just does not match intuition at all.","startTime":876.76,"endTime":890.92,"durationSeconds":14,"level":"C1","overallScore":9,"rationale":"지수 추세 해석이 통찰·표현 모두 강함."},{"segmentIndex":54,"text":"Um you just have to give people space. You have to give them maybe the word is like safety. So, it's like psychological safety that it's okay to fail.","startTime":926.8,"endTime":931,"durationSeconds":4,"level":"C1","overallScore":8.8,"rationale":"혁신 문화의 조건을 구체적으로 제시함."},{"segmentIndex":24,"text":"So, I think a lot of the people that will be rewarded most over the next few years, they won't just be AI native and they don't just know how to use these tools really well, but also they're curious and they're generalists and they cross over multiple disciplines and can think about the broader problem they're solving rather than just engineering part of it.","startTime":2530.2,"endTime":2538.12,"durationSeconds":8,"level":"C1","overallScore":8.8,"rationale":"미래 인재상 통찰이 매우 응축됨."},{"segmentIndex":31,"text":"I do think that there is a future where I think by the end of the year what we're going to start to see is these start to get even murkier where I think in some places the title software engineer is going to start to go away and it's just going to be replaced by builder or maybe it's just everyone's going to be a product manager and everyone codes or something like this.","startTime":2587.96,"endTime":2606.2,"durationSeconds":18,"level":"C1","overallScore":8.8,"rationale":"직함 변화 전망이 매우 선명함."},{"segmentIndex":66,"text":"It's whatever people are doing if you can make that a little bit easier then that's just going to be a much better product that people enjoy more and this is just this principle of latent demand which I think is just the single most important principle in product.","startTime":2836.2,"endTime":2849.84,"durationSeconds":14,"level":"C1","overallScore":8.8,"rationale":"핵심 제품 원칙을 강하게 일반화함."},{"segmentIndex":68,"text":"Latent demand is this idea that if you build a product in a way that can be hacked or can be kind of mis [clears throat] used by people in a way it wasn't really designed for it to do kind of something that they want to do then this helps you as the product builder learn where to take the product next.","startTime":2857.64,"endTime":2875.08,"durationSeconds":17,"level":"C1","overallScore":9,"rationale":"개념 정의와 시사점이 매우 선명하다."},{"segmentIndex":8,"text":"So when you see people abusing the product in this way, using it in a way that it wasn't designed in order to do something that is useful for them, it's just such a strong indicator that you should just build a product and people are going to like that.","startTime":3036.24,"endTime":3047.2,"durationSeconds":11,"level":"C1","overallScore":9,"rationale":"제품 통찰이 강하고 표현도 풍부함."},{"segmentIndex":44,"text":"But this is kind of the way that we learn, um, both on the product side and on the safety side is we have to release things a little bit earlier than we think so that we can get the feedback, so that we can talk to users.","startTime":3230.04,"endTime":3238.68,"durationSeconds":9,"level":"C1","overallScore":9,"rationale":"학습 원칙이 선명하고 표현도 풍부함."},{"segmentIndex":45,"text":"But actually almost always you get better results if you just give the model tools, you give it a goal, and you let it figure it out.","startTime":3846.28,"endTime":3851.52,"durationSeconds":5,"level":"B2","overallScore":8.8,"rationale":"핵심 원칙과 실용 표현이 함께 강하다."},{"segmentIndex":53,"text":"Uh and it's actually a really simple idea. His idea was that the more general model will always outperform the more specific model.","startTime":3895.4,"endTime":3902.12,"durationSeconds":7,"level":"B2","overallScore":8.8,"rationale":"일반 모델 우위를 압축한 핵심 통찰."},{"segmentIndex":79,"text":"But if you build for the model 6 months out, when that model comes out, you're just going to hit the ground running, and the product is going to click and start to work.","startTime":4026.56,"endTime":4035.84,"durationSeconds":9,"level":"B2","overallScore":9,"rationale":"행동 지침과 표현이 모두 강함."},{"segmentIndex":48,"text":"And they're just following the momentum and not thinking about it. I think the best results that I see are people thinking from first principles and just developing their own common sense.","startTime":5069.56,"endTime":5077.64,"durationSeconds":8,"level":"B2","overallScore":8.8,"rationale":"성과 원인을 원칙 수준에서 설명함."},{"segmentIndex":4,"text":"I have never enjoyed coding as much as I do today because I don't have to deal with all the minutiae.","startTime":13.32,"endTime":18.04,"durationSeconds":5,"level":"B2","overallScore":7.6,"rationale":"코딩 만족도의 이유를 설명해 통찰적."},{"segmentIndex":11,"text":"The title software engineer is going to start [music] to go away. It's just going to be replaced by builder and it's going to be painful for a lot of people.","startTime":47.6,"endTime":58.39,"durationSeconds":11,"level":"B2","overallScore":7.8,"rationale":"직함 변화 전망과 여파를 압축 제시."},{"segmentIndex":55,"text":"Um and so the this kind of like mission drivenness just really, really resonated with me, and I just know personally it's something I need in order to be happy. Um and I that's just the thing that I really missed.","startTime":311.68,"endTime":321,"durationSeconds":9,"level":"C1","overallScore":8,"rationale":"가치관과 행복 조건을 깊이 있게 설명."},{"segmentIndex":56,"text":"And I found that, you know, whatever the work might be, no matter how exciting, even if it's building a really cool product, it's just not really a substitute for that.","startTime":321,"endTime":330.6,"durationSeconds":10,"level":"C1","overallScore":7.8,"rationale":"일의 의미에 대한 보편적 통찰이 있음."},{"segmentIndex":74,"text":"And you know, for Anthropic for a long time, we were building the models in this way that kind of fit our mental model of the way that we build safe AGI, where the model starts by being really good at coding, then it gets really good at tool use, then it gets really good at computer use.","startTime":443.919,"endTime":457.8,"durationSeconds":14,"level":"C1","overallScore":8,"rationale":"제품 철학과 발전 경로 통찰이 크다."}],"generatedAt":"2026-06-25T00:01:01.396Z","keyClipsTotalSec":1420},{"videoId":"We7BZVKbCVw","chunkIndex":6,"totalChunks":9,"title":"Head of Claude Code: What happens after coding is solved | Boris Cherny — Part 7 of 9","thumbnail":"https://i.ytimg.com/vi/We7BZVKbCVw/maxresdefault.jpg","duration":5265,"uploader":"Lenny's Podcast","youtubeUrl":"https://www.youtube.com/watch?v=We7BZVKbCVw","keywords":["ai","machine-learning","developer-tools","coding","llm","startup","product-strategy","automation","agentic-ai"],"normalizedKeywords":["프로덕트","엔지니어링","기술 트렌드"],"targetAudience":[{"who":"AI 제품 창업자","why":"모델이 빨리 좋아지는 환경에서 제품 전략을 어떻게 짜야 하는지 배울 수 있음"},{"who":"엔지니어","why":"Claude Code 같은 에이전트형 도구를 더 효율적으로 쓰는 실전 팁이 많음"},{"who":"프로덕트 매니저","why":"모델 중심 제품 설계에서 스캐폴딩보다 도구 제공이 중요하다는 관점을 얻을 수 있음"}],"normalizedAudience":["창업자·스타트업","프로덕트 매니저·기획자","엔지니어·개발자"],"summary":"이 구간은 Claude Code의 사용 방식과 AI 제품을 만드는 원칙을 중심으로, 지금의 코딩이 '코드를 직접 치는 일'에서 '모델에게 원하는 결과를 설명하는 일'로 바뀌고 있다는 관점을 보여준다. 화자는 에이전트가 오래 자율적으로 돌고, 터미널·데스크톱·iOS까지 여러 환경에서 작업하는 것이 이미 일상이라고 말하며, AI 도구 사용이 점점 더 자연스러워지고 있음을 강조한다.\n\n핵심 조언은 모델을 과도하게 박스에 넣지 말고, 도구와 목표를 주고 스스로 풀게 하라는 것이다. 더 일반적인 모델이 더 특화된 모델이나 복잡한 워크플로우를 장기적으로 이긴다는 'bitter lesson'을 반복해서 강조하며, 제품도 현재 모델이 아니라 6개월 뒤 모델에 맞춰 설계해야 한다고 주장한다. Claude Code 사용 팁으로는 가장 강력한 모델을 쓰고, 계획 모드로 시작하는 습관이 특히 중요하다고 말한다.","insights":["AI 제품은 현재 모델이 아니라 다음 모델을 기준으로 설계해야 한다.","복잡한 워크플로우보다 도구와 목표를 주는 편이 대체로 낫다.","더 일반적인 모델이 특화된 스캐폴딩의 이점을 결국 잠식한다.","좋은 모델은 오래 자율적으로 돌수록 가치가 커진다.","비용 절감은 약한 모델보다 강한 모델이 더 유리할 때가 많다."],"keyClips":[{"clipId":"We7BZVKbCVw:c6:3-14","startSegmentIndex":3,"endSegmentIndex":14,"startTime":3611.48,"endTime":3676.4,"durationSeconds":64.9,"preview":"코딩의 정의 변화","mustSee":false},{"clipId":"We7BZVKbCVw:c6:42-50","startSegmentIndex":42,"endSegmentIndex":50,"startTime":3822.96,"endTime":3880.51,"durationSeconds":57.6,"preview":"모델을 박스에 넣지 말라","mustSee":false},{"clipId":"We7BZVKbCVw:c6:51-63","startSegmentIndex":51,"endSegmentIndex":63,"startTime":3880.51,"endTime":3950.16,"durationSeconds":69.6,"preview":"비터 레슨의 교훈","mustSee":false},{"clipId":"We7BZVKbCVw:c6:64-79","startSegmentIndex":64,"endSegmentIndex":79,"startTime":3950.16,"endTime":4035.84,"durationSeconds":85.7,"preview":"6개월 뒤를 설계하라","mustSee":true},{"clipId":"We7BZVKbCVw:c6:84-89","startSegmentIndex":84,"endSegmentIndex":89,"startTime":4060.76,"endTime":4119.44,"durationSeconds":58.7,"preview":"장시간 자율 실행","mustSee":false},{"clipId":"We7BZVKbCVw:c6:97-105","startSegmentIndex":97,"endSegmentIndex":105,"startTime":4156.96,"endTime":4208.64,"durationSeconds":51.7,"preview":"실전 Claude Code 팁","mustSee":false}],"curatedSegments":[{"segmentIndex":6,"text":"In a year or two it's not going [music] to matter. Coding is virtually solved. I imagine a world where everyone is able to program.","startTime":22.72,"endTime":27.88,"durationSeconds":5,"level":"B2","overallScore":8.4,"rationale":"미래 변화에 대한 강한 관점 제시."},{"segmentIndex":83,"text":"It acts in the world. Um, and I think now with Co-work, we're starting to see the transition for non-technical folks also. And I think the craziest thing for me isn't even the number that we're at right now, but the pace at which we're growing because if you look at Claude code's growth rate kind of across any metric, it's continuing to accelerate. And I think honestly, that's just for me as an engineer, I find that to do good work, you really have to understand the layer under the layer at which you work.","startTime":501.88,"endTime":508.8,"durationSeconds":7,"level":"C1","overallScore":9,"rationale":"성장 해석과 일의 원칙이 매우 통찰적."},{"segmentIndex":26,"text":"It's just And when I think about it, kind of part of the reason Quad Code works is this idea of latent demand where we bring the tool to where people are and it makes existing workflows a little bit easier. But also because it's a it's in a terminal, it's like a little surprising.","startTime":732.52,"endTime":745.16,"durationSeconds":13,"level":"C1","overallScore":9,"rationale":"제품 성공 원인을 원리 수준에서 설명함."},{"segmentIndex":48,"text":"Um, so we really just think in exponentials. And if you kind of look at the exponential of the percent of code that was written by Claude at that point, and if you just trace the line, it's pretty obvious we're going to cross 100% by the end of the year, even if it just does not match intuition at all.","startTime":876.76,"endTime":890.92,"durationSeconds":14,"level":"C1","overallScore":9,"rationale":"지수 추세 해석이 통찰·표현 모두 강함."},{"segmentIndex":54,"text":"Um you just have to give people space. You have to give them maybe the word is like safety. So, it's like psychological safety that it's okay to fail.","startTime":926.8,"endTime":931,"durationSeconds":4,"level":"C1","overallScore":8.8,"rationale":"혁신 문화의 조건을 구체적으로 제시함."},{"segmentIndex":24,"text":"So, I think a lot of the people that will be rewarded most over the next few years, they won't just be AI native and they don't just know how to use these tools really well, but also they're curious and they're generalists and they cross over multiple disciplines and can think about the broader problem they're solving rather than just engineering part of it.","startTime":2530.2,"endTime":2538.12,"durationSeconds":8,"level":"C1","overallScore":8.8,"rationale":"미래 인재상 통찰이 매우 응축됨."},{"segmentIndex":31,"text":"I do think that there is a future where I think by the end of the year what we're going to start to see is these start to get even murkier where I think in some places the title software engineer is going to start to go away and it's just going to be replaced by builder or maybe it's just everyone's going to be a product manager and everyone codes or something like this.","startTime":2587.96,"endTime":2606.2,"durationSeconds":18,"level":"C1","overallScore":8.8,"rationale":"직함 변화 전망이 매우 선명함."},{"segmentIndex":66,"text":"It's whatever people are doing if you can make that a little bit easier then that's just going to be a much better product that people enjoy more and this is just this principle of latent demand which I think is just the single most important principle in product.","startTime":2836.2,"endTime":2849.84,"durationSeconds":14,"level":"C1","overallScore":8.8,"rationale":"핵심 제품 원칙을 강하게 일반화함."},{"segmentIndex":68,"text":"Latent demand is this idea that if you build a product in a way that can be hacked or can be kind of mis [clears throat] used by people in a way it wasn't really designed for it to do kind of something that they want to do then this helps you as the product builder learn where to take the product next.","startTime":2857.64,"endTime":2875.08,"durationSeconds":17,"level":"C1","overallScore":9,"rationale":"개념 정의와 시사점이 매우 선명하다."},{"segmentIndex":8,"text":"So when you see people abusing the product in this way, using it in a way that it wasn't designed in order to do something that is useful for them, it's just such a strong indicator that you should just build a product and people are going to like that.","startTime":3036.24,"endTime":3047.2,"durationSeconds":11,"level":"C1","overallScore":9,"rationale":"제품 통찰이 강하고 표현도 풍부함."},{"segmentIndex":44,"text":"But this is kind of the way that we learn, um, both on the product side and on the safety side is we have to release things a little bit earlier than we think so that we can get the feedback, so that we can talk to users.","startTime":3230.04,"endTime":3238.68,"durationSeconds":9,"level":"C1","overallScore":9,"rationale":"학습 원칙이 선명하고 표현도 풍부함."},{"segmentIndex":45,"text":"But actually almost always you get better results if you just give the model tools, you give it a goal, and you let it figure it out.","startTime":3846.28,"endTime":3851.52,"durationSeconds":5,"level":"B2","overallScore":8.8,"rationale":"핵심 원칙과 실용 표현이 함께 강하다."},{"segmentIndex":53,"text":"Uh and it's actually a really simple idea. His idea was that the more general model will always outperform the more specific model.","startTime":3895.4,"endTime":3902.12,"durationSeconds":7,"level":"B2","overallScore":8.8,"rationale":"일반 모델 우위를 압축한 핵심 통찰."},{"segmentIndex":79,"text":"But if you build for the model 6 months out, when that model comes out, you're just going to hit the ground running, and the product is going to click and start to work.","startTime":4026.56,"endTime":4035.84,"durationSeconds":9,"level":"B2","overallScore":9,"rationale":"행동 지침과 표현이 모두 강함."},{"segmentIndex":48,"text":"And they're just following the momentum and not thinking about it. I think the best results that I see are people thinking from first principles and just developing their own common sense.","startTime":5069.56,"endTime":5077.64,"durationSeconds":8,"level":"B2","overallScore":8.8,"rationale":"성과 원인을 원칙 수준에서 설명함."},{"segmentIndex":4,"text":"I have never enjoyed coding as much as I do today because I don't have to deal with all the minutiae.","startTime":13.32,"endTime":18.04,"durationSeconds":5,"level":"B2","overallScore":7.6,"rationale":"코딩 만족도의 이유를 설명해 통찰적."},{"segmentIndex":11,"text":"The title software engineer is going to start [music] to go away. It's just going to be replaced by builder and it's going to be painful for a lot of people.","startTime":47.6,"endTime":58.39,"durationSeconds":11,"level":"B2","overallScore":7.8,"rationale":"직함 변화 전망과 여파를 압축 제시."},{"segmentIndex":55,"text":"Um and so the this kind of like mission drivenness just really, really resonated with me, and I just know personally it's something I need in order to be happy. Um and I that's just the thing that I really missed.","startTime":311.68,"endTime":321,"durationSeconds":9,"level":"C1","overallScore":8,"rationale":"가치관과 행복 조건을 깊이 있게 설명."},{"segmentIndex":56,"text":"And I found that, you know, whatever the work might be, no matter how exciting, even if it's building a really cool product, it's just not really a substitute for that.","startTime":321,"endTime":330.6,"durationSeconds":10,"level":"C1","overallScore":7.8,"rationale":"일의 의미에 대한 보편적 통찰이 있음."},{"segmentIndex":74,"text":"And you know, for Anthropic for a long time, we were building the models in this way that kind of fit our mental model of the way that we build safe AGI, where the model starts by being really good at coding, then it gets really good at tool use, then it gets really good at computer use.","startTime":443.919,"endTime":457.8,"durationSeconds":14,"level":"C1","overallScore":8,"rationale":"제품 철학과 발전 경로 통찰이 크다."}],"generatedAt":"2026-06-25T00:01:34.592Z","keyClipsTotalSec":1420},{"videoId":"We7BZVKbCVw","chunkIndex":7,"totalChunks":9,"title":"Head of Claude Code: What happens after coding is solved | Boris Cherny — Part 8 of 9","thumbnail":"https://i.ytimg.com/vi/We7BZVKbCVw/maxresdefault.jpg","duration":5265,"uploader":"Lenny's Podcast","youtubeUrl":"https://www.youtube.com/watch?v=We7BZVKbCVw","keywords":["coding-agents","claude-code","product-strategy","ai-tools","user-feedback","competition","agi","sci-fi","japanese-lifestyle","technical-books"],"normalizedKeywords":["엔지니어링","프로덕트","기술 트렌드"],"targetAudience":[{"who":"엔지니어","why":"코딩 에이전트의 사용법, 인터페이스 선택, 제품 철학을 직접 들을 수 있음"},{"who":"프로덕트 매니저","why":"사용자 피드백 중심으로 제품을 개선하는 관점을 배울 수 있음"},{"who":"창업자","why":"경쟁보다 사용자 문제에 집중하는 제품 성장 원리를 참고할 수 있음"}],"normalizedAudience":["엔지니어·개발자","프로덕트 매니저·기획자","창업자·스타트업"],"summary":"이 영상은 Claude Code의 제품 철학과 사용 팁을 중심으로, 코딩 에이전트를 어떻게 더 잘 쓰고 더 잘 만드는지에 대한 대화를 담고 있다. 핵심은 계획 모드처럼 매우 단순한 UX 실험이 실제 사용성을 크게 높일 수 있고, 터미널만이 아니라 데스크톱·웹·슬랙 등 다양한 인터페이스에서 사람마다 맞는 방식을 찾아야 한다는 점이다. 또한 경쟁 제품보다 사용자 피드백에 집중하는 태도, 그리고 여전히 시장과 제품이 '1%만 완성된' 초기 단계라는 인식이 반복된다.\n\n후반부에서는 AGI 이후 무엇을 할지에 대한 질문에, 화자가 일본 시골에서의 생활과 미소 만들기 경험을 연결해 장기적 시간감각을 이야기한다. 엔지니어링의 속도와 대비되는 발효의 느린 리듬이 그의 사고방식을 만들었고, 그 감각이 Anthropic에 합류한 이유와도 이어진다고 말한다. 마지막으로 그는 기능형 프로그래밍 책, Accelerando, The Wandering Earth 같은 책을 추천하며, 지금의 기술 변화를 이해하는 데 SF가 큰 역할을 한다고 강조한다.","insights":["좋은 제품은 복잡한 기술보다 단순한 사용 규칙에서 나온다.","인터페이스는 하나가 정답이 아니라 사용자별 적합성이 핵심이다.","경쟁 분석보다 사용자 문제 해결이 제품 개선을 더 빠르게 만든다.","AI 제품은 아직 초기라서, 성장의 엔진은 결국 피드백 루프다.","장기적 시간감각은 엔지니어링 판단에도 큰 영향을 준다."],"keyClips":[{"clipId":"We7BZVKbCVw:c7:1-15","startSegmentIndex":1,"endSegmentIndex":15,"startTime":4200.27,"endTime":4276.6,"durationSeconds":76.3,"preview":"플랜 모드의 본질","mustSee":false},{"clipId":"We7BZVKbCVw:c7:16-26","startSegmentIndex":16,"endSegmentIndex":26,"startTime":4276.6,"endTime":4333.32,"durationSeconds":56.7,"preview":"경쟁보다 사용자","mustSee":false},{"clipId":"We7BZVKbCVw:c7:27-44","startSegmentIndex":27,"endSegmentIndex":44,"startTime":4333.32,"endTime":4426.8,"durationSeconds":93.5,"preview":"미소와 장기시야","mustSee":false},{"clipId":"We7BZVKbCVw:c7:52-68","startSegmentIndex":52,"endSegmentIndex":68,"startTime":4470.24,"endTime":4567.84,"durationSeconds":97.6,"preview":"아직도 초기 단계","mustSee":false},{"clipId":"We7BZVKbCVw:c7:72-102","startSegmentIndex":72,"endSegmentIndex":102,"startTime":4583.32,"endTime":4756.92,"durationSeconds":173.6,"preview":"책과 SF의 역할","mustSee":false}],"curatedSegments":[{"segmentIndex":6,"text":"In a year or two it's not going [music] to matter. Coding is virtually solved. I imagine a world where everyone is able to program.","startTime":22.72,"endTime":27.88,"durationSeconds":5,"level":"B2","overallScore":8.4,"rationale":"미래 변화에 대한 강한 관점 제시."},{"segmentIndex":83,"text":"It acts in the world. Um, and I think now with Co-work, we're starting to see the transition for non-technical folks also. And I think the craziest thing for me isn't even the number that we're at right now, but the pace at which we're growing because if you look at Claude code's growth rate kind of across any metric, it's continuing to accelerate. And I think honestly, that's just for me as an engineer, I find that to do good work, you really have to understand the layer under the layer at which you work.","startTime":501.88,"endTime":508.8,"durationSeconds":7,"level":"C1","overallScore":9,"rationale":"성장 해석과 일의 원칙이 매우 통찰적."},{"segmentIndex":26,"text":"It's just And when I think about it, kind of part of the reason Quad Code works is this idea of latent demand where we bring the tool to where people are and it makes existing workflows a little bit easier. But also because it's a it's in a terminal, it's like a little surprising.","startTime":732.52,"endTime":745.16,"durationSeconds":13,"level":"C1","overallScore":9,"rationale":"제품 성공 원인을 원리 수준에서 설명함."},{"segmentIndex":48,"text":"Um, so we really just think in exponentials. And if you kind of look at the exponential of the percent of code that was written by Claude at that point, and if you just trace the line, it's pretty obvious we're going to cross 100% by the end of the year, even if it just does not match intuition at all.","startTime":876.76,"endTime":890.92,"durationSeconds":14,"level":"C1","overallScore":9,"rationale":"지수 추세 해석이 통찰·표현 모두 강함."},{"segmentIndex":54,"text":"Um you just have to give people space. You have to give them maybe the word is like safety. So, it's like psychological safety that it's okay to fail.","startTime":926.8,"endTime":931,"durationSeconds":4,"level":"C1","overallScore":8.8,"rationale":"혁신 문화의 조건을 구체적으로 제시함."},{"segmentIndex":24,"text":"So, I think a lot of the people that will be rewarded most over the next few years, they won't just be AI native and they don't just know how to use these tools really well, but also they're curious and they're generalists and they cross over multiple disciplines and can think about the broader problem they're solving rather than just engineering part of it.","startTime":2530.2,"endTime":2538.12,"durationSeconds":8,"level":"C1","overallScore":8.8,"rationale":"미래 인재상 통찰이 매우 응축됨."},{"segmentIndex":31,"text":"I do think that there is a future where I think by the end of the year what we're going to start to see is these start to get even murkier where I think in some places the title software engineer is going to start to go away and it's just going to be replaced by builder or maybe it's just everyone's going to be a product manager and everyone codes or something like this.","startTime":2587.96,"endTime":2606.2,"durationSeconds":18,"level":"C1","overallScore":8.8,"rationale":"직함 변화 전망이 매우 선명함."},{"segmentIndex":66,"text":"It's whatever people are doing if you can make that a little bit easier then that's just going to be a much better product that people enjoy more and this is just this principle of latent demand which I think is just the single most important principle in product.","startTime":2836.2,"endTime":2849.84,"durationSeconds":14,"level":"C1","overallScore":8.8,"rationale":"핵심 제품 원칙을 강하게 일반화함."},{"segmentIndex":68,"text":"Latent demand is this idea that if you build a product in a way that can be hacked or can be kind of mis [clears throat] used by people in a way it wasn't really designed for it to do kind of something that they want to do then this helps you as the product builder learn where to take the product next.","startTime":2857.64,"endTime":2875.08,"durationSeconds":17,"level":"C1","overallScore":9,"rationale":"개념 정의와 시사점이 매우 선명하다."},{"segmentIndex":8,"text":"So when you see people abusing the product in this way, using it in a way that it wasn't designed in order to do something that is useful for them, it's just such a strong indicator that you should just build a product and people are going to like that.","startTime":3036.24,"endTime":3047.2,"durationSeconds":11,"level":"C1","overallScore":9,"rationale":"제품 통찰이 강하고 표현도 풍부함."},{"segmentIndex":44,"text":"But this is kind of the way that we learn, um, both on the product side and on the safety side is we have to release things a little bit earlier than we think so that we can get the feedback, so that we can talk to users.","startTime":3230.04,"endTime":3238.68,"durationSeconds":9,"level":"C1","overallScore":9,"rationale":"학습 원칙이 선명하고 표현도 풍부함."},{"segmentIndex":45,"text":"But actually almost always you get better results if you just give the model tools, you give it a goal, and you let it figure it out.","startTime":3846.28,"endTime":3851.52,"durationSeconds":5,"level":"B2","overallScore":8.8,"rationale":"핵심 원칙과 실용 표현이 함께 강하다."},{"segmentIndex":53,"text":"Uh and it's actually a really simple idea. His idea was that the more general model will always outperform the more specific model.","startTime":3895.4,"endTime":3902.12,"durationSeconds":7,"level":"B2","overallScore":8.8,"rationale":"일반 모델 우위를 압축한 핵심 통찰."},{"segmentIndex":79,"text":"But if you build for the model 6 months out, when that model comes out, you're just going to hit the ground running, and the product is going to click and start to work.","startTime":4026.56,"endTime":4035.84,"durationSeconds":9,"level":"B2","overallScore":9,"rationale":"행동 지침과 표현이 모두 강함."},{"segmentIndex":48,"text":"And they're just following the momentum and not thinking about it. I think the best results that I see are people thinking from first principles and just developing their own common sense.","startTime":5069.56,"endTime":5077.64,"durationSeconds":8,"level":"B2","overallScore":8.8,"rationale":"성과 원인을 원칙 수준에서 설명함."},{"segmentIndex":4,"text":"I have never enjoyed coding as much as I do today because I don't have to deal with all the minutiae.","startTime":13.32,"endTime":18.04,"durationSeconds":5,"level":"B2","overallScore":7.6,"rationale":"코딩 만족도의 이유를 설명해 통찰적."},{"segmentIndex":11,"text":"The title software engineer is going to start [music] to go away. It's just going to be replaced by builder and it's going to be painful for a lot of people.","startTime":47.6,"endTime":58.39,"durationSeconds":11,"level":"B2","overallScore":7.8,"rationale":"직함 변화 전망과 여파를 압축 제시."},{"segmentIndex":55,"text":"Um and so the this kind of like mission drivenness just really, really resonated with me, and I just know personally it's something I need in order to be happy. Um and I that's just the thing that I really missed.","startTime":311.68,"endTime":321,"durationSeconds":9,"level":"C1","overallScore":8,"rationale":"가치관과 행복 조건을 깊이 있게 설명."},{"segmentIndex":56,"text":"And I found that, you know, whatever the work might be, no matter how exciting, even if it's building a really cool product, it's just not really a substitute for that.","startTime":321,"endTime":330.6,"durationSeconds":10,"level":"C1","overallScore":7.8,"rationale":"일의 의미에 대한 보편적 통찰이 있음."},{"segmentIndex":74,"text":"And you know, for Anthropic for a long time, we were building the models in this way that kind of fit our mental model of the way that we build safe AGI, where the model starts by being really good at coding, then it gets really good at tool use, then it gets really good at computer use.","startTime":443.919,"endTime":457.8,"durationSeconds":14,"level":"C1","overallScore":8,"rationale":"제품 철학과 발전 경로 통찰이 크다."}],"generatedAt":"2026-06-25T00:02:06.586Z","keyClipsTotalSec":1420},{"videoId":"We7BZVKbCVw","chunkIndex":8,"totalChunks":9,"title":"Head of Claude Code: What happens after coding is solved | Boris Cherny — Part 9 of 9","thumbnail":"https://i.ytimg.com/vi/We7BZVKbCVw/maxresdefault.jpg","duration":5265,"uploader":"Lenny's Podcast","youtubeUrl":"https://www.youtube.com/watch?v=We7BZVKbCVw","keywords":["ai-agents","productivity","automation","browser-automation","workflow","tooling","product-management","twitter","common-sense","tech-productivity"],"normalizedKeywords":["프로덕트","리더십·매니지먼트","기술 트렌드"],"targetAudience":[{"who":"프로덕트 매니저","why":"AI 에이전트의 실제 사용법과 업무 자동화 아이디어를 얻을 수 있음"},{"who":"창업자","why":"새로운 제품이 어떻게 빠르게 확산되고 평가되는지 감을 잡을 수 있음"},{"who":"지식노동자","why":"반복 업무를 자동화해 시간을 아끼는 실전 활용법이 많음"}],"normalizedAudience":["프로덕트 매니저·기획자","창업자·스타트업","지식노동자 일반"],"summary":"이 영상은 Claude Code/CoWork 같은 AI 에이전트가 이제는 단순한 코딩 보조를 넘어, 브라우저를 열고 실제 업무를 대신 처리하는 단계에 들어왔다는 점을 강조한다. 화자는 이메일 응답, Slack 알림, 스프레드시트 기반 프로젝트 관리, PDF 양식 처리 같은 반복 작업을 에이전트에 맡기는 방식으로 생산성을 크게 높였다고 말한다. 핵심 메시지는 '모델이 준비됐는지'보다 '사용자가 공통 감각으로 어디에 붙일지'가 더 중요하다는 것이다.\n\n후반부에서는 이런 도구를 잘 쓰는 방법을 세 가지로 정리한다: 먼저 단순한 도구 하나를 맡기고, 그다음 여러 툴을 연결하고, 마지막으로 여러 작업을 병렬로 돌리라는 것. 또한 Boris Cherny는 제품 개선의 출발점으로 Twitter/X에서 사용자 버그와 피드백을 직접 듣는 습관을 소개하며, 좋은 제품은 빠른 학습 루프와 상식적인 판단에서 나온다고 정리한다.","insights":["에이전트의 가치는 '자동화'보다 '번거로움 제거'에 있다.","좋은 AI 활용은 먼저 단순 업무 하나를 맡기는 데서 시작한다.","툴 연결이 되면 이메일·Slack·스프레드시트가 하나의 흐름이 된다.","상식이 없으면 강한 모델도 엉뚱한 일을 더 빨리 할 뿐이다.","사용자 피드백을 직접 받는 사람은 제품 개선 속도가 압도적으로 빨라진다."],"keyClips":[{"clipId":"We7BZVKbCVw:c8:4-17","startSegmentIndex":4,"endSegmentIndex":17,"startTime":4822.76,"endTime":4899.04,"durationSeconds":76.3,"preview":"에이전트의 실제 효용","mustSee":true},{"clipId":"We7BZVKbCVw:c8:20-32","startSegmentIndex":20,"endSegmentIndex":32,"startTime":4912.52,"endTime":4989.52,"durationSeconds":77,"preview":"업무 자동화의 3단계","mustSee":false},{"clipId":"We7BZVKbCVw:c8:33-44","startSegmentIndex":33,"endSegmentIndex":44,"startTime":4989.52,"endTime":5048.76,"durationSeconds":59.2,"preview":"사용 사례로 학습하기","mustSee":false},{"clipId":"We7BZVKbCVw:c8:45-50","startSegmentIndex":45,"endSegmentIndex":50,"startTime":5048.76,"endTime":5091.44,"durationSeconds":42.7,"preview":"상식이 최고의 기준","mustSee":false},{"clipId":"We7BZVKbCVw:c8:51-68","startSegmentIndex":51,"endSegmentIndex":68,"startTime":5091.44,"endTime":5209.96,"durationSeconds":118.5,"preview":"피드백 루프의 힘","mustSee":true},{"clipId":"We7BZVKbCVw:c8:70-74","startSegmentIndex":70,"endSegmentIndex":74,"startTime":5213.56,"endTime":5236.64,"durationSeconds":23.1,"preview":"제품은 함께 만든다","mustSee":false}],"curatedSegments":[{"segmentIndex":6,"text":"In a year or two it's not going [music] to matter. Coding is virtually solved. I imagine a world where everyone is able to program.","startTime":22.72,"endTime":27.88,"durationSeconds":5,"level":"B2","overallScore":8.4,"rationale":"미래 변화에 대한 강한 관점 제시."},{"segmentIndex":83,"text":"It acts in the world. Um, and I think now with Co-work, we're starting to see the transition for non-technical folks also. And I think the craziest thing for me isn't even the number that we're at right now, but the pace at which we're growing because if you look at Claude code's growth rate kind of across any metric, it's continuing to accelerate. And I think honestly, that's just for me as an engineer, I find that to do good work, you really have to understand the layer under the layer at which you work.","startTime":501.88,"endTime":508.8,"durationSeconds":7,"level":"C1","overallScore":9,"rationale":"성장 해석과 일의 원칙이 매우 통찰적."},{"segmentIndex":26,"text":"It's just And when I think about it, kind of part of the reason Quad Code works is this idea of latent demand where we bring the tool to where people are and it makes existing workflows a little bit easier. But also because it's a it's in a terminal, it's like a little surprising.","startTime":732.52,"endTime":745.16,"durationSeconds":13,"level":"C1","overallScore":9,"rationale":"제품 성공 원인을 원리 수준에서 설명함."},{"segmentIndex":48,"text":"Um, so we really just think in exponentials. And if you kind of look at the exponential of the percent of code that was written by Claude at that point, and if you just trace the line, it's pretty obvious we're going to cross 100% by the end of the year, even if it just does not match intuition at all.","startTime":876.76,"endTime":890.92,"durationSeconds":14,"level":"C1","overallScore":9,"rationale":"지수 추세 해석이 통찰·표현 모두 강함."},{"segmentIndex":54,"text":"Um you just have to give people space. You have to give them maybe the word is like safety. So, it's like psychological safety that it's okay to fail.","startTime":926.8,"endTime":931,"durationSeconds":4,"level":"C1","overallScore":8.8,"rationale":"혁신 문화의 조건을 구체적으로 제시함."},{"segmentIndex":24,"text":"So, I think a lot of the people that will be rewarded most over the next few years, they won't just be AI native and they don't just know how to use these tools really well, but also they're curious and they're generalists and they cross over multiple disciplines and can think about the broader problem they're solving rather than just engineering part of it.","startTime":2530.2,"endTime":2538.12,"durationSeconds":8,"level":"C1","overallScore":8.8,"rationale":"미래 인재상 통찰이 매우 응축됨."},{"segmentIndex":31,"text":"I do think that there is a future where I think by the end of the year what we're going to start to see is these start to get even murkier where I think in some places the title software engineer is going to start to go away and it's just going to be replaced by builder or maybe it's just everyone's going to be a product manager and everyone codes or something like this.","startTime":2587.96,"endTime":2606.2,"durationSeconds":18,"level":"C1","overallScore":8.8,"rationale":"직함 변화 전망이 매우 선명함."},{"segmentIndex":66,"text":"It's whatever people are doing if you can make that a little bit easier then that's just going to be a much better product that people enjoy more and this is just this principle of latent demand which I think is just the single most important principle in product.","startTime":2836.2,"endTime":2849.84,"durationSeconds":14,"level":"C1","overallScore":8.8,"rationale":"핵심 제품 원칙을 강하게 일반화함."},{"segmentIndex":68,"text":"Latent demand is this idea that if you build a product in a way that can be hacked or can be kind of mis [clears throat] used by people in a way it wasn't really designed for it to do kind of something that they want to do then this helps you as the product builder learn where to take the product next.","startTime":2857.64,"endTime":2875.08,"durationSeconds":17,"level":"C1","overallScore":9,"rationale":"개념 정의와 시사점이 매우 선명하다."},{"segmentIndex":8,"text":"So when you see people abusing the product in this way, using it in a way that it wasn't designed in order to do something that is useful for them, it's just such a strong indicator that you should just build a product and people are going to like that.","startTime":3036.24,"endTime":3047.2,"durationSeconds":11,"level":"C1","overallScore":9,"rationale":"제품 통찰이 강하고 표현도 풍부함."},{"segmentIndex":44,"text":"But this is kind of the way that we learn, um, both on the product side and on the safety side is we have to release things a little bit earlier than we think so that we can get the feedback, so that we can talk to users.","startTime":3230.04,"endTime":3238.68,"durationSeconds":9,"level":"C1","overallScore":9,"rationale":"학습 원칙이 선명하고 표현도 풍부함."},{"segmentIndex":45,"text":"But actually almost always you get better results if you just give the model tools, you give it a goal, and you let it figure it out.","startTime":3846.28,"endTime":3851.52,"durationSeconds":5,"level":"B2","overallScore":8.8,"rationale":"핵심 원칙과 실용 표현이 함께 강하다."},{"segmentIndex":53,"text":"Uh and it's actually a really simple idea. His idea was that the more general model will always outperform the more specific model.","startTime":3895.4,"endTime":3902.12,"durationSeconds":7,"level":"B2","overallScore":8.8,"rationale":"일반 모델 우위를 압축한 핵심 통찰."},{"segmentIndex":79,"text":"But if you build for the model 6 months out, when that model comes out, you're just going to hit the ground running, and the product is going to click and start to work.","startTime":4026.56,"endTime":4035.84,"durationSeconds":9,"level":"B2","overallScore":9,"rationale":"행동 지침과 표현이 모두 강함."},{"segmentIndex":48,"text":"And they're just following the momentum and not thinking about it. I think the best results that I see are people thinking from first principles and just developing their own common sense.","startTime":5069.56,"endTime":5077.64,"durationSeconds":8,"level":"B2","overallScore":8.8,"rationale":"성과 원인을 원칙 수준에서 설명함."},{"segmentIndex":4,"text":"I have never enjoyed coding as much as I do today because I don't have to deal with all the minutiae.","startTime":13.32,"endTime":18.04,"durationSeconds":5,"level":"B2","overallScore":7.6,"rationale":"코딩 만족도의 이유를 설명해 통찰적."},{"segmentIndex":11,"text":"The title software engineer is going to start [music] to go away. It's just going to be replaced by builder and it's going to be painful for a lot of people.","startTime":47.6,"endTime":58.39,"durationSeconds":11,"level":"B2","overallScore":7.8,"rationale":"직함 변화 전망과 여파를 압축 제시."},{"segmentIndex":55,"text":"Um and so the this kind of like mission drivenness just really, really resonated with me, and I just know personally it's something I need in order to be happy. Um and I that's just the thing that I really missed.","startTime":311.68,"endTime":321,"durationSeconds":9,"level":"C1","overallScore":8,"rationale":"가치관과 행복 조건을 깊이 있게 설명."},{"segmentIndex":56,"text":"And I found that, you know, whatever the work might be, no matter how exciting, even if it's building a really cool product, it's just not really a substitute for that.","startTime":321,"endTime":330.6,"durationSeconds":10,"level":"C1","overallScore":7.8,"rationale":"일의 의미에 대한 보편적 통찰이 있음."},{"segmentIndex":74,"text":"And you know, for Anthropic for a long time, we were building the models in this way that kind of fit our mental model of the way that we build safe AGI, where the model starts by being really good at coding, then it gets really good at tool use, then it gets really good at computer use.","startTime":443.919,"endTime":457.8,"durationSeconds":14,"level":"C1","overallScore":8,"rationale":"제품 철학과 발전 경로 통찰이 크다."}],"generatedAt":"2026-06-25T00:02:34.146Z","keyClipsTotalSec":1420}]}