{"success":true,"count":10,"items":[{"videoId":"4D3hDmGhFhA","chunkIndex":0,"totalChunks":10,"title":"The AI paradox: More automation, more humans, more work | Dan Shipper — Part 1 of 10","thumbnail":"https://i.ytimg.com/vi/4D3hDmGhFhA/maxresdefault.jpg","duration":5646,"uploader":"Lenny's Podcast","youtubeUrl":"https://www.youtube.com/watch?v=4D3hDmGhFhA","keywords":["ai","automation","future-of-work","startup","productivity","saas","agents","product-design","career"],"normalizedKeywords":["비즈니스·전략","프로덕트","기술 트렌드"],"targetAudience":[{"who":"창업자","why":"AI가 일과 제품 구조를 어떻게 바꿀지 먼저 읽고 전략을 세울 수 있음"},{"who":"프로덕트 매니저","why":"에이전트 시대에 PM의 역할과 유리한 역량을 가늠할 수 있음"},{"who":"디자이너","why":"풀스택 디자이너가 왜 강해지는지와 일의 방식 변화를 이해할 수 있음"},{"who":"지식노동자","why":"AI 도입이 오히려 인간의 역할과 업무량을 어떻게 늘리는지 배울 수 있음"}],"normalizedAudience":["창업자·스타트업","프로덕트 매니저·기획자","프로덕트 디자이너"],"summary":"이 영상은 AI가 인간을 대체해 일자리를 없앤다는 통념에 반대하며, 오히려 더 많은 인간 개입과 더 많은 일이 생긴다고 주장한다. Dan Shipper는 AI를 적극적으로 쓰는 조직일수록 자동화가 완성되는 것이 아니라, 인간의 판단·창의성·감각이 더 중요해진다고 본다. 특히 모델이 '어제의 인간 역량'을 싸게 만들수록, 사람은 그 위에서 새로운 것을 조합하고 차별화하는 역할로 이동한다고 말한다.\n\n또한 앞으로의 일은 에이전트를 한 명 이상 붙여 쓰는 형태로 바뀌고, 실제 작업은 코드 에디터가 아니라 Codex나 Claude Code 같은 환경 안에서 이루어질 것이라고 예측한다. 하지만 SaaS가 사라지는 것은 아니며, 에이전트는 오히려 SaaS 사용자를 늘릴 것이라고 본다. 이 미래에서 특히 유리한 사람은 PM과 풀스택 디자이너처럼 AI를 다루면서도 문제 정의와 결과 감각을 함께 가진 사람들이다.","insights":["AI는 인간을 없애기보다 인간의 역할을 더 선명하게 만든다.","모델이 싸게 만드는 것은 '어제의 역량'이지 미래의 차별점이 아니다.","자동화가 늘수록 창의성과 감각의 희소성은 더 커진다.","에이전트 시대에는 소프트웨어를 쓰는 방식 자체가 바뀐다.","SaaS는 사라지지 않고, 에이전트 덕분에 더 많이 쓰인다."],"keyClips":[{"clipId":"4D3hDmGhFhA:c0:3-10","startSegmentIndex":3,"endSegmentIndex":10,"startTime":10.48,"endTime":57.32,"durationSeconds":46.8,"preview":"AI가 못 없애는 것","mustSee":false},{"clipId":"4D3hDmGhFhA:c0:11-17","startSegmentIndex":11,"endSegmentIndex":17,"startTime":57.32,"endTime":93.36,"durationSeconds":36,"preview":"일의 인터페이스 변화","mustSee":false},{"clipId":"4D3hDmGhFhA:c0:45-55","startSegmentIndex":45,"endSegmentIndex":55,"startTime":290.04,"endTime":384.32,"durationSeconds":94.3,"preview":"AI 조직의 학습법","mustSee":false},{"clipId":"4D3hDmGhFhA:c0:56-66","startSegmentIndex":56,"endSegmentIndex":66,"startTime":384.32,"endTime":468.2,"durationSeconds":83.9,"preview":"Claude Code 전환점","mustSee":true},{"clipId":"4D3hDmGhFhA:c0:78-85","startSegmentIndex":78,"endSegmentIndex":85,"startTime":557.8389999999999,"endTime":607.12,"durationSeconds":49.3,"preview":"미래의 승자","mustSee":false}],"curatedSegments":[{"segmentIndex":45,"text":"And there's a lot of work that's human work that uh it can't be scored until you write it down, but the act of thinking to prompt it or write it down um is uh is something that you can't measure, but like kind of means that even if the benchmarks get saturated, it doesn't mean the same thing as we you totally replace all senior engineers.","startTime":2703.56,"endTime":2707,"durationSeconds":3,"level":"C2","overallScore":9.8,"rationale":"측정 불가한 인간 일의 본질을 짚음."},{"segmentIndex":6,"text":"Automation is a lie. Every agent needs a human. We have so much automation, so much AI, and I also work way more. >> Creativity.","startTime":29,"endTime":35.2,"durationSeconds":6,"level":"B2","overallScore":8.4,"rationale":"핵심 주장과 대비가 매우 선명함."},{"segmentIndex":8,"text":">> What models do in general is they make yesterday's human competence cheap.","startTime":40.72,"endTime":44.44,"durationSeconds":4,"level":"B2","overallScore":8.6,"rationale":"일반화된 통찰이 매우 강함."},{"segmentIndex":41,"text":"is the minute the agent is like not really that useful anymore. And that's why it I think it has started to shift to a more uh one agent per company model because for now like the ideal is uh you basically set up a forward deployed engineer or someone with that sort of profile who's responsible for making sure that agent is working for the whole company.","startTime":907.68,"endTime":931.36,"durationSeconds":24,"level":"C1","overallScore":8.8,"rationale":"회사당 에이전트 모델의 이유를 설명."},{"segmentIndex":29,"text":"And I think the reverse is actually starting to happen and be like really, really valuable in a way that I did not expect, which is take the AI agent that you use all the time on your computer and put a browser in it so it can see everything you're doing.","startTime":1397.4,"endTime":1410.64,"durationSeconds":13,"level":"C1","overallScore":9,"rationale":"핵심 전환점을 명료하게 설명."},{"segmentIndex":73,"text":"Automation is a lie. Um in the sense that every time you automate something, in order to make sure the automation is working well, you need a human on top of it like making sure that it's working well.","startTime":2355.12,"endTime":2366.64,"durationSeconds":12,"level":"B2","overallScore":8.8,"rationale":"자동화 역설을 명확히 말해 통찰과 표현 강함."},{"segmentIndex":47,"text":"And so I think it's really important uh when we think about benchmark progress to think about it from that perspective, which is benchmarks rise on problems that we've framed that we can articulate, that we can score.","startTime":2711.12,"endTime":2745.16,"durationSeconds":34,"level":"C1","overallScore":8.8,"rationale":"벤치마크 해석 원칙을 정리한다."},{"segmentIndex":69,"text":"And I think that's a really important thing when we think about benchmarks is AI is a broadly distributed technology that any human can use and when we are benchmarking against humans, AI against humans, we're actually really always talking about one human using AI versus another human using AI cuz AI doesn't use itself.","startTime":2887.24,"endTime":2908.12,"durationSeconds":21,"level":"C1","overallScore":9,"rationale":"AI 대 인간 비교의 핵심 재정의."},{"segmentIndex":30,"text":"The pattern that I see so far and again, I don't have a total crystal ball, but I do feel like we've seen enough of the new model drops to like have some sense of how this is going is that what a new model drop does or what models do in general is they make yesterday's human competence cheap.","startTime":4418.6,"endTime":4436.48,"durationSeconds":18,"level":"C1","overallScore":8.8,"rationale":"모델의 경제적 효과를 날카롭게 정의."},{"segmentIndex":39,"text":"And I really think that structurally, because of the way the models work, because of the financial incentives of model companies to like make model of model companies to like make them um uh compliant and aligned, structurally, there are always going to be trailing behind those people who are taking the models and using them to make new expertise or make new things that haven't been done that way before for their very particular situation.","startTime":4502.2,"endTime":4527.72,"durationSeconds":26,"level":"C2","overallScore":8.8,"rationale":"구조적 이유로 인간 우위가 남는단 통찰."},{"segmentIndex":62,"text":"Um so, the way to ride the models is like not one specific thing cuz they're always changing, but it is to be curious and playful, to apply the model the new model to whatever it is that you care about, whether that's your job or something outside of your job and to keep turning over rocks uh because it may not work now, but it may work eventually, it probably will work eventually and the way that you use it matters.","startTime":4680.52,"endTime":4717.6,"durationSeconds":37,"level":"C1","overallScore":9,"rationale":"핵심 조언과 학습 표현이 매우 풍부함."},{"segmentIndex":18,"text":"And I think generally our intuitions about the future the model that I have of what our intuitions are about the future is the intuitions that people had in the Middle Ages about like what happened at the end of the horizon, you know, it's like are there dragons?","startTime":4914.6,"endTime":4931.12,"durationSeconds":17,"level":"C1","overallScore":8.8,"rationale":"미래 인식을 강한 비유로 통찰적으로 설명함."},{"segmentIndex":23,"text":"And until you get to that place where you're starting to see it, and I think we get to see it cuz we get to see it internally all the time, it's important not to let your mind get away from you and being like, this is going to happen and this like, this is going to happen and whatever cuz you're going to tell a story that sounds so real in the moment, but um later on you're like, actually it's much more complex than that and somewhere it's sort of a both everything has changed and nothing has.","startTime":4958.8,"endTime":4986.76,"durationSeconds":28,"level":"C1","overallScore":8.8,"rationale":"미래 예측의 함정을 깊이 있게 경고함."},{"segmentIndex":32,"text":"And try to like try to have fun. I think there's too much of I'm doing this because I have FOMO like it might I might lose my job or like I might miss out on this big thing or whatever and the best way to actually figure out interesting useful things to do with AI is to like do something enjoyable.","startTime":5052.28,"endTime":5071.32,"durationSeconds":19,"level":"C1","overallScore":9,"rationale":"FOMO보다 즐거움이 중요하단 조언이 강함."},{"segmentIndex":3,"text":">> The AI job apocalypse is not really a thing. I am super bullish on PMs and full-stack designers.","startTime":10.48,"endTime":18.28,"durationSeconds":8,"level":"B2","overallScore":7.8,"rationale":"강한 주장과 유용한 구어 표현."},{"segmentIndex":9,"text":"And so, it becomes commoditized. It's not valuable anymore. What humans do is we go in there and we're like,\"Yeah, we have all this frozen human competence from yesterday.","startTime":44.44,"endTime":52.76,"durationSeconds":8,"level":"C1","overallScore":7.8,"rationale":"개념 설명이 선명하고 표현도 좋음."},{"segmentIndex":11,"text":">> It's going to bifurcate in two main ways. One is everyone's going to have at least one agent that they talk to, that they can offload work to.","startTime":57.32,"endTime":64.879,"durationSeconds":8,"level":"B2","overallScore":8,"rationale":"예측과 표현 가치가 모두 높음."},{"segmentIndex":15,"text":"What agents do is increase the number of users of SaaS, not get rid of it.","startTime":80.52,"endTime":84.16,"durationSeconds":4,"level":"B2","overallScore":7.6,"rationale":"핵심 반박 논리가 분명함."},{"segmentIndex":42,"text":"So, just give us a sense of how you work. Thank you. Um I really appreciate the introduction. Um and yeah, I think what one of the things about predicting the future or the way that we think about predicting the future at Every is that you what you don't want to do is prognosticate.","startTime":266.52,"endTime":271.16,"durationSeconds":5,"level":"C1","overallScore":8,"rationale":"미래 예측 태도에 대한 통찰이 큼."},{"segmentIndex":50,"text":"And what I what that does, I think, is it creates this like little pocket of the future where we're all living in it together and we get to be a little bit further ahead cuz at any other company there's like a mix of people.","startTime":320.24,"endTime":330.44,"durationSeconds":10,"level":"C1","overallScore":8,"rationale":"조직이 미래를 앞서가는 이유 설명."}],"generatedAt":"2026-06-25T00:41:41.268Z","keyClipsTotalSec":1550},{"videoId":"4D3hDmGhFhA","chunkIndex":1,"totalChunks":10,"title":"The AI paradox: More automation, more humans, more work | Dan Shipper — Part 2 of 10","thumbnail":"https://i.ytimg.com/vi/4D3hDmGhFhA/maxresdefault.jpg","duration":5646,"uploader":"Lenny's Podcast","youtubeUrl":"https://www.youtube.com/watch?v=4D3hDmGhFhA","keywords":["artificial-intelligence","automation","agents","future-of-work","enterprise-software","productivity","slack","coding-tools","workflow","startup"],"normalizedKeywords":["기술 트렌드","비즈니스·전략","프로덕트"],"targetAudience":[{"who":"창업자","why":"AI 에이전트를 업무에 도입할 때의 제품·조직 모델을 이해할 수 있음"},{"who":"프로덕트 매니저","why":"새로운 업무 인터페이스가 어떻게 자리 잡는지 감을 잡을 수 있음"},{"who":"엔지니어","why":"에이전트가 왜 컴퓨터/터미널 기반으로 진화하는지 맥락을 얻을 수 있음"},{"who":"지식노동자","why":"앞으로 일하는 방식이 에이전트 중심으로 바뀔 가능성을 파악할 수 있음"}],"normalizedAudience":["창업자·스타트업","프로덕트 매니저·기획자","엔지니어·개발자"],"summary":"이 영상은 AI가 일자리를 단순히 대체하는 것이 아니라, 오히려 인간에게 더 많은 업무를 남기게 될 수 있다는 ‘패러독스’를 설명한다. 화자는 앞으로의 업무 변화가 크게 두 갈래로 진행된다고 보는데, 하나는 Slack 같은 곳에서 대화하며 일감을 넘기는 회사용 에이전트이고, 다른 하나는 Codex나 Claude Code 같은 환경이 이메일·문서·개발을 포함한 실제 작업의 운영체제가 되는 방향이다.\n\n특히 그는 개인용 에이전트가 직관적으로 매력적이지만, 실제로는 유지보수와 신뢰 문제 때문에 당분간 회사 단위의 중앙 에이전트가 먼저 자리 잡을 것이라고 본다. 에이전트가 유용하려면 ‘가꾸어 주는 사람’이 필요하며, 결국 AI 도구의 경쟁력은 모델 성능만이 아니라 운영 방식과 책임 구조에 달려 있다는 점을 강조한다.","insights":["AI가 강해질수록 인간의 일이 줄지 않고 재배치된다.","에이전트의 핵심 병목은 모델이 아니라 운영과 관리다.","개인용보다 회사용 중앙 에이전트가 먼저 확산될 가능성이 높다.","유용한 에이전트는 '돌봐주는 인간'이 있을 때만 지속된다.","업무의 승부처는 챗봇이 아니라 작업 환경 자체다."],"keyClips":[{"clipId":"4D3hDmGhFhA:c1:6-17","startSegmentIndex":6,"endSegmentIndex":17,"startTime":629.32,"endTime":755.8389999999999,"durationSeconds":126.5,"preview":"AI가 더 바쁜 이유","mustSee":false},{"clipId":"4D3hDmGhFhA:c1:19-29","startSegmentIndex":19,"endSegmentIndex":29,"startTime":759.68,"endTime":836.72,"durationSeconds":77,"preview":"개인 에이전트의 한계","mustSee":false},{"clipId":"4D3hDmGhFhA:c1:30-46","startSegmentIndex":30,"endSegmentIndex":46,"startTime":836.72,"endTime":984.44,"durationSeconds":147.7,"preview":"회사 에이전트의 부상","mustSee":true},{"clipId":"4D3hDmGhFhA:c1:53-60","startSegmentIndex":53,"endSegmentIndex":60,"startTime":1019.72,"endTime":1089.04,"durationSeconds":69.3,"preview":"Slack이 일의 입구","mustSee":false},{"clipId":"4D3hDmGhFhA:c1:63-72","startSegmentIndex":63,"endSegmentIndex":72,"startTime":1096.92,"endTime":1196.52,"durationSeconds":99.6,"preview":"작업환경이 OS다","mustSee":true}],"curatedSegments":[{"segmentIndex":45,"text":"And there's a lot of work that's human work that uh it can't be scored until you write it down, but the act of thinking to prompt it or write it down um is uh is something that you can't measure, but like kind of means that even if the benchmarks get saturated, it doesn't mean the same thing as we you totally replace all senior engineers.","startTime":2703.56,"endTime":2707,"durationSeconds":3,"level":"C2","overallScore":9.8,"rationale":"측정 불가한 인간 일의 본질을 짚음."},{"segmentIndex":6,"text":"Automation is a lie. Every agent needs a human. We have so much automation, so much AI, and I also work way more. >> Creativity.","startTime":29,"endTime":35.2,"durationSeconds":6,"level":"B2","overallScore":8.4,"rationale":"핵심 주장과 대비가 매우 선명함."},{"segmentIndex":8,"text":">> What models do in general is they make yesterday's human competence cheap.","startTime":40.72,"endTime":44.44,"durationSeconds":4,"level":"B2","overallScore":8.6,"rationale":"일반화된 통찰이 매우 강함."},{"segmentIndex":41,"text":"is the minute the agent is like not really that useful anymore. And that's why it I think it has started to shift to a more uh one agent per company model because for now like the ideal is uh you basically set up a forward deployed engineer or someone with that sort of profile who's responsible for making sure that agent is working for the whole company.","startTime":907.68,"endTime":931.36,"durationSeconds":24,"level":"C1","overallScore":8.8,"rationale":"회사당 에이전트 모델의 이유를 설명."},{"segmentIndex":29,"text":"And I think the reverse is actually starting to happen and be like really, really valuable in a way that I did not expect, which is take the AI agent that you use all the time on your computer and put a browser in it so it can see everything you're doing.","startTime":1397.4,"endTime":1410.64,"durationSeconds":13,"level":"C1","overallScore":9,"rationale":"핵심 전환점을 명료하게 설명."},{"segmentIndex":73,"text":"Automation is a lie. Um in the sense that every time you automate something, in order to make sure the automation is working well, you need a human on top of it like making sure that it's working well.","startTime":2355.12,"endTime":2366.64,"durationSeconds":12,"level":"B2","overallScore":8.8,"rationale":"자동화 역설을 명확히 말해 통찰과 표현 강함."},{"segmentIndex":47,"text":"And so I think it's really important uh when we think about benchmark progress to think about it from that perspective, which is benchmarks rise on problems that we've framed that we can articulate, that we can score.","startTime":2711.12,"endTime":2745.16,"durationSeconds":34,"level":"C1","overallScore":8.8,"rationale":"벤치마크 해석 원칙을 정리한다."},{"segmentIndex":69,"text":"And I think that's a really important thing when we think about benchmarks is AI is a broadly distributed technology that any human can use and when we are benchmarking against humans, AI against humans, we're actually really always talking about one human using AI versus another human using AI cuz AI doesn't use itself.","startTime":2887.24,"endTime":2908.12,"durationSeconds":21,"level":"C1","overallScore":9,"rationale":"AI 대 인간 비교의 핵심 재정의."},{"segmentIndex":30,"text":"The pattern that I see so far and again, I don't have a total crystal ball, but I do feel like we've seen enough of the new model drops to like have some sense of how this is going is that what a new model drop does or what models do in general is they make yesterday's human competence cheap.","startTime":4418.6,"endTime":4436.48,"durationSeconds":18,"level":"C1","overallScore":8.8,"rationale":"모델의 경제적 효과를 날카롭게 정의."},{"segmentIndex":39,"text":"And I really think that structurally, because of the way the models work, because of the financial incentives of model companies to like make model of model companies to like make them um uh compliant and aligned, structurally, there are always going to be trailing behind those people who are taking the models and using them to make new expertise or make new things that haven't been done that way before for their very particular situation.","startTime":4502.2,"endTime":4527.72,"durationSeconds":26,"level":"C2","overallScore":8.8,"rationale":"구조적 이유로 인간 우위가 남는단 통찰."},{"segmentIndex":62,"text":"Um so, the way to ride the models is like not one specific thing cuz they're always changing, but it is to be curious and playful, to apply the model the new model to whatever it is that you care about, whether that's your job or something outside of your job and to keep turning over rocks uh because it may not work now, but it may work eventually, it probably will work eventually and the way that you use it matters.","startTime":4680.52,"endTime":4717.6,"durationSeconds":37,"level":"C1","overallScore":9,"rationale":"핵심 조언과 학습 표현이 매우 풍부함."},{"segmentIndex":18,"text":"And I think generally our intuitions about the future the model that I have of what our intuitions are about the future is the intuitions that people had in the Middle Ages about like what happened at the end of the horizon, you know, it's like are there dragons?","startTime":4914.6,"endTime":4931.12,"durationSeconds":17,"level":"C1","overallScore":8.8,"rationale":"미래 인식을 강한 비유로 통찰적으로 설명함."},{"segmentIndex":23,"text":"And until you get to that place where you're starting to see it, and I think we get to see it cuz we get to see it internally all the time, it's important not to let your mind get away from you and being like, this is going to happen and this like, this is going to happen and whatever cuz you're going to tell a story that sounds so real in the moment, but um later on you're like, actually it's much more complex than that and somewhere it's sort of a both everything has changed and nothing has.","startTime":4958.8,"endTime":4986.76,"durationSeconds":28,"level":"C1","overallScore":8.8,"rationale":"미래 예측의 함정을 깊이 있게 경고함."},{"segmentIndex":32,"text":"And try to like try to have fun. I think there's too much of I'm doing this because I have FOMO like it might I might lose my job or like I might miss out on this big thing or whatever and the best way to actually figure out interesting useful things to do with AI is to like do something enjoyable.","startTime":5052.28,"endTime":5071.32,"durationSeconds":19,"level":"C1","overallScore":9,"rationale":"FOMO보다 즐거움이 중요하단 조언이 강함."},{"segmentIndex":3,"text":">> The AI job apocalypse is not really a thing. I am super bullish on PMs and full-stack designers.","startTime":10.48,"endTime":18.28,"durationSeconds":8,"level":"B2","overallScore":7.8,"rationale":"강한 주장과 유용한 구어 표현."},{"segmentIndex":9,"text":"And so, it becomes commoditized. It's not valuable anymore. What humans do is we go in there and we're like,\"Yeah, we have all this frozen human competence from yesterday.","startTime":44.44,"endTime":52.76,"durationSeconds":8,"level":"C1","overallScore":7.8,"rationale":"개념 설명이 선명하고 표현도 좋음."},{"segmentIndex":11,"text":">> It's going to bifurcate in two main ways. One is everyone's going to have at least one agent that they talk to, that they can offload work to.","startTime":57.32,"endTime":64.879,"durationSeconds":8,"level":"B2","overallScore":8,"rationale":"예측과 표현 가치가 모두 높음."},{"segmentIndex":15,"text":"What agents do is increase the number of users of SaaS, not get rid of it.","startTime":80.52,"endTime":84.16,"durationSeconds":4,"level":"B2","overallScore":7.6,"rationale":"핵심 반박 논리가 분명함."},{"segmentIndex":42,"text":"So, just give us a sense of how you work. Thank you. Um I really appreciate the introduction. Um and yeah, I think what one of the things about predicting the future or the way that we think about predicting the future at Every is that you what you don't want to do is prognosticate.","startTime":266.52,"endTime":271.16,"durationSeconds":5,"level":"C1","overallScore":8,"rationale":"미래 예측 태도에 대한 통찰이 큼."},{"segmentIndex":50,"text":"And what I what that does, I think, is it creates this like little pocket of the future where we're all living in it together and we get to be a little bit further ahead cuz at any other company there's like a mix of people.","startTime":320.24,"endTime":330.44,"durationSeconds":10,"level":"C1","overallScore":8,"rationale":"조직이 미래를 앞서가는 이유 설명."}],"generatedAt":"2026-06-25T00:42:11.838Z","keyClipsTotalSec":1550},{"videoId":"4D3hDmGhFhA","chunkIndex":2,"totalChunks":10,"title":"The AI paradox: More automation, more humans, more work | Dan Shipper — Part 3 of 10","thumbnail":"https://i.ytimg.com/vi/4D3hDmGhFhA/maxresdefault.jpg","duration":5646,"uploader":"Lenny's Podcast","youtubeUrl":"https://www.youtube.com/watch?v=4D3hDmGhFhA","keywords":["ai-agents","automation","knowledge-work","saas","productivity","openai","anthropic","cursor","workflow","desktop-app"],"normalizedKeywords":["비즈니스·전략","프로덕트","기술 트렌드"],"targetAudience":[{"who":"SaaS 창업자","why":"AI 에이전트 시대에 제품 구조와 수익모델이 어떻게 바뀌는지 직접 참고할 수 있음"},{"who":"프로덕트 매니저","why":"인간 중심 UX에서 인간+에이전트 협업 UX로의 전환 포인트를 이해할 수 있음"},{"who":"엔지니어","why":"에이전트 하니스, 브라우저 내장, 로그/승인/롤백 같은 구현 요구를 알 수 있음"},{"who":"지식노동자","why":"이메일·문서·리서치 같은 반복 업무가 어떻게 자동화되는지 감을 잡을 수 있음"}],"normalizedAudience":["창업자·스타트업","프로덕트 매니저·기획자","엔지니어·개발자"],"summary":"이 구간은 AI 제품의 경쟁축이 '모델 성능'에서 '모델을 제대로 쓰게 만드는 하니스(harness)'로 이동하고 있다고 주장한다. OpenAI의 Codex, Anthropic의 Claude Code, Cursor를 비교하면서, 앞으로는 AI가 SaaS 안에 들어가는 것보다 오히려 SaaS가 에이전트 안에서 실행되는 구조가 더 중요해질 것이라고 본다. 화자는 실제로 Codex와 브라우저, 문서 편집기, 이메일 에이전트를 함께 쓰며 인간과 에이전트가 같은 작업 공간에서 협업하는 방식이 강력하다고 강조한다.\n\n또한 이런 변화는 제품 설계와 비즈니스 모델까지 바꾼다고 말한다. 사용자가 자기 AI를 가져와 SaaS를 사용하면 토큰 비용 부담이 줄고, 제품은 인간용 UI만이 아니라 에이전트가 쓰기 좋은 CLI, 로그, 승인, 롤백, 대량 요청 처리까지 고려해야 한다. 결국 'AI 기능 추가'가 아니라 '인간과 에이전트가 함께 일하는 새 UX'를 만드는 것이 핵심이라는 메시지다.","insights":["AI 경쟁의 본질은 모델이 아니라 하니스다.","미래의 SaaS는 AI를 내장하기보다 AI가 들어오는 곳이 된다.","인간과 에이전트는 분리된 도구가 아니라 같은 작업면을 공유해야 한다.","사용자가 AI를 가져오면 SaaS의 토큰 비용 구조가 바뀐다.","에이전트 시대의 제품은 승인·로그·롤백이 기본값이어야 한다."],"keyClips":[{"clipId":"4D3hDmGhFhA:c2:1-11","startSegmentIndex":1,"endSegmentIndex":11,"startTime":1200.83,"endTime":1274,"durationSeconds":73.2,"preview":"모델보다 하니스","mustSee":false},{"clipId":"4D3hDmGhFhA:c2:12-20","startSegmentIndex":12,"endSegmentIndex":20,"startTime":1274,"endTime":1340.08,"durationSeconds":66.1,"preview":"같은 공간의 협업","mustSee":false},{"clipId":"4D3hDmGhFhA:c2:21-32","startSegmentIndex":21,"endSegmentIndex":32,"startTime":1340.08,"endTime":1425.64,"durationSeconds":85.6,"preview":"브라우저가 들어간 에이전트","mustSee":true},{"clipId":"4D3hDmGhFhA:c2:33-43","startSegmentIndex":33,"endSegmentIndex":43,"startTime":1425.64,"endTime":1539.48,"durationSeconds":113.8,"preview":"SaaS와 토큰 경제","mustSee":false},{"clipId":"4D3hDmGhFhA:c2:44-57","startSegmentIndex":44,"endSegmentIndex":57,"startTime":1539.48,"endTime":1655.32,"durationSeconds":115.8,"preview":"플랫폼의 새 전쟁","mustSee":false},{"clipId":"4D3hDmGhFhA:c2:58-77","startSegmentIndex":58,"endSegmentIndex":77,"startTime":1655.32,"endTime":1808.72,"durationSeconds":153.4,"preview":"에이전트 UX의 조건","mustSee":true}],"curatedSegments":[{"segmentIndex":45,"text":"And there's a lot of work that's human work that uh it can't be scored until you write it down, but the act of thinking to prompt it or write it down um is uh is something that you can't measure, but like kind of means that even if the benchmarks get saturated, it doesn't mean the same thing as we you totally replace all senior engineers.","startTime":2703.56,"endTime":2707,"durationSeconds":3,"level":"C2","overallScore":9.8,"rationale":"측정 불가한 인간 일의 본질을 짚음."},{"segmentIndex":6,"text":"Automation is a lie. Every agent needs a human. We have so much automation, so much AI, and I also work way more. >> Creativity.","startTime":29,"endTime":35.2,"durationSeconds":6,"level":"B2","overallScore":8.4,"rationale":"핵심 주장과 대비가 매우 선명함."},{"segmentIndex":8,"text":">> What models do in general is they make yesterday's human competence cheap.","startTime":40.72,"endTime":44.44,"durationSeconds":4,"level":"B2","overallScore":8.6,"rationale":"일반화된 통찰이 매우 강함."},{"segmentIndex":41,"text":"is the minute the agent is like not really that useful anymore. And that's why it I think it has started to shift to a more uh one agent per company model because for now like the ideal is uh you basically set up a forward deployed engineer or someone with that sort of profile who's responsible for making sure that agent is working for the whole company.","startTime":907.68,"endTime":931.36,"durationSeconds":24,"level":"C1","overallScore":8.8,"rationale":"회사당 에이전트 모델의 이유를 설명."},{"segmentIndex":29,"text":"And I think the reverse is actually starting to happen and be like really, really valuable in a way that I did not expect, which is take the AI agent that you use all the time on your computer and put a browser in it so it can see everything you're doing.","startTime":1397.4,"endTime":1410.64,"durationSeconds":13,"level":"C1","overallScore":9,"rationale":"핵심 전환점을 명료하게 설명."},{"segmentIndex":73,"text":"Automation is a lie. Um in the sense that every time you automate something, in order to make sure the automation is working well, you need a human on top of it like making sure that it's working well.","startTime":2355.12,"endTime":2366.64,"durationSeconds":12,"level":"B2","overallScore":8.8,"rationale":"자동화 역설을 명확히 말해 통찰과 표현 강함."},{"segmentIndex":47,"text":"And so I think it's really important uh when we think about benchmark progress to think about it from that perspective, which is benchmarks rise on problems that we've framed that we can articulate, that we can score.","startTime":2711.12,"endTime":2745.16,"durationSeconds":34,"level":"C1","overallScore":8.8,"rationale":"벤치마크 해석 원칙을 정리한다."},{"segmentIndex":69,"text":"And I think that's a really important thing when we think about benchmarks is AI is a broadly distributed technology that any human can use and when we are benchmarking against humans, AI against humans, we're actually really always talking about one human using AI versus another human using AI cuz AI doesn't use itself.","startTime":2887.24,"endTime":2908.12,"durationSeconds":21,"level":"C1","overallScore":9,"rationale":"AI 대 인간 비교의 핵심 재정의."},{"segmentIndex":30,"text":"The pattern that I see so far and again, I don't have a total crystal ball, but I do feel like we've seen enough of the new model drops to like have some sense of how this is going is that what a new model drop does or what models do in general is they make yesterday's human competence cheap.","startTime":4418.6,"endTime":4436.48,"durationSeconds":18,"level":"C1","overallScore":8.8,"rationale":"모델의 경제적 효과를 날카롭게 정의."},{"segmentIndex":39,"text":"And I really think that structurally, because of the way the models work, because of the financial incentives of model companies to like make model of model companies to like make them um uh compliant and aligned, structurally, there are always going to be trailing behind those people who are taking the models and using them to make new expertise or make new things that haven't been done that way before for their very particular situation.","startTime":4502.2,"endTime":4527.72,"durationSeconds":26,"level":"C2","overallScore":8.8,"rationale":"구조적 이유로 인간 우위가 남는단 통찰."},{"segmentIndex":62,"text":"Um so, the way to ride the models is like not one specific thing cuz they're always changing, but it is to be curious and playful, to apply the model the new model to whatever it is that you care about, whether that's your job or something outside of your job and to keep turning over rocks uh because it may not work now, but it may work eventually, it probably will work eventually and the way that you use it matters.","startTime":4680.52,"endTime":4717.6,"durationSeconds":37,"level":"C1","overallScore":9,"rationale":"핵심 조언과 학습 표현이 매우 풍부함."},{"segmentIndex":18,"text":"And I think generally our intuitions about the future the model that I have of what our intuitions are about the future is the intuitions that people had in the Middle Ages about like what happened at the end of the horizon, you know, it's like are there dragons?","startTime":4914.6,"endTime":4931.12,"durationSeconds":17,"level":"C1","overallScore":8.8,"rationale":"미래 인식을 강한 비유로 통찰적으로 설명함."},{"segmentIndex":23,"text":"And until you get to that place where you're starting to see it, and I think we get to see it cuz we get to see it internally all the time, it's important not to let your mind get away from you and being like, this is going to happen and this like, this is going to happen and whatever cuz you're going to tell a story that sounds so real in the moment, but um later on you're like, actually it's much more complex than that and somewhere it's sort of a both everything has changed and nothing has.","startTime":4958.8,"endTime":4986.76,"durationSeconds":28,"level":"C1","overallScore":8.8,"rationale":"미래 예측의 함정을 깊이 있게 경고함."},{"segmentIndex":32,"text":"And try to like try to have fun. I think there's too much of I'm doing this because I have FOMO like it might I might lose my job or like I might miss out on this big thing or whatever and the best way to actually figure out interesting useful things to do with AI is to like do something enjoyable.","startTime":5052.28,"endTime":5071.32,"durationSeconds":19,"level":"C1","overallScore":9,"rationale":"FOMO보다 즐거움이 중요하단 조언이 강함."},{"segmentIndex":3,"text":">> The AI job apocalypse is not really a thing. I am super bullish on PMs and full-stack designers.","startTime":10.48,"endTime":18.28,"durationSeconds":8,"level":"B2","overallScore":7.8,"rationale":"강한 주장과 유용한 구어 표현."},{"segmentIndex":9,"text":"And so, it becomes commoditized. It's not valuable anymore. What humans do is we go in there and we're like,\"Yeah, we have all this frozen human competence from yesterday.","startTime":44.44,"endTime":52.76,"durationSeconds":8,"level":"C1","overallScore":7.8,"rationale":"개념 설명이 선명하고 표현도 좋음."},{"segmentIndex":11,"text":">> It's going to bifurcate in two main ways. One is everyone's going to have at least one agent that they talk to, that they can offload work to.","startTime":57.32,"endTime":64.879,"durationSeconds":8,"level":"B2","overallScore":8,"rationale":"예측과 표현 가치가 모두 높음."},{"segmentIndex":15,"text":"What agents do is increase the number of users of SaaS, not get rid of it.","startTime":80.52,"endTime":84.16,"durationSeconds":4,"level":"B2","overallScore":7.6,"rationale":"핵심 반박 논리가 분명함."},{"segmentIndex":42,"text":"So, just give us a sense of how you work. Thank you. Um I really appreciate the introduction. Um and yeah, I think what one of the things about predicting the future or the way that we think about predicting the future at Every is that you what you don't want to do is prognosticate.","startTime":266.52,"endTime":271.16,"durationSeconds":5,"level":"C1","overallScore":8,"rationale":"미래 예측 태도에 대한 통찰이 큼."},{"segmentIndex":50,"text":"And what I what that does, I think, is it creates this like little pocket of the future where we're all living in it together and we get to be a little bit further ahead cuz at any other company there's like a mix of people.","startTime":320.24,"endTime":330.44,"durationSeconds":10,"level":"C1","overallScore":8,"rationale":"조직이 미래를 앞서가는 이유 설명."}],"generatedAt":"2026-06-25T00:42:21.814Z","keyClipsTotalSec":1550},{"videoId":"4D3hDmGhFhA","chunkIndex":3,"totalChunks":10,"title":"The AI paradox: More automation, more humans, more work | Dan Shipper — Part 4 of 10","thumbnail":"https://i.ytimg.com/vi/4D3hDmGhFhA/maxresdefault.jpg","duration":5646,"uploader":"Lenny's Podcast","youtubeUrl":"https://www.youtube.com/watch?v=4D3hDmGhFhA","keywords":["ai-agents","saas","software-business","automation","developer-tools","productivity","workflow","business-strategy","clis","future-of-work"],"normalizedKeywords":["비즈니스·전략","프로덕트","기술 트렌드"],"targetAudience":[{"who":"창업자","why":"AI 에이전트 시대의 제품 설계와 사업모델 변화를 이해하는 데 유용함"},{"who":"프로덕트 매니저","why":"사람과 에이전트가 함께 쓰는 제품 구조를 어떻게 설계할지 힌트를 줌"},{"who":"엔지니어","why":"CLI/GUI/에이전트 도구 흐름과 에이전트 연동 패턴을 배울 수 있음"},{"who":"투자자","why":"SaaS 수요와 마진이 왜 오히려 강화될 수 있는지 논리를 볼 수 있음"}],"normalizedAudience":["창업자·스타트업","프로덕트 매니저·기획자","엔지니어·개발자"],"summary":"이 구간은 AI 에이전트가 소프트웨어 사용 방식을 어떻게 바꿀지에 대한 Dan Shipper의 전망을 중심으로 전개된다. 사용자는 이제 Slack에서 회사 에이전트와 대화하고, 개인 작업은 Codex나 Cloud Code 같은 도구 안에서 수행하게 되며, 앱은 그 내부 브라우저 안에서 다뤄질 것이라고 본다. 그는 CLI가 주류 작업 표면이 되기엔 한계가 있고, GUI가 다시 더 자연스러운 인터페이스가 될 것이라고 주장한다.\n\n또한 그는 에이전트가 늘어날수록 SaaS가 사라지기보다 오히려 더 많이 쓰이게 될 것이라고 본다. 에이전트가 SaaS를 더 자주, 더 많이 호출하면서 수요가 증가하고, 기업은 AI를 덧붙이는 데 그치지 않고 인간과 AI가 함께 협업하기 좋은 소프트웨어를 만들어야 한다는 것이다. 마지막으로 자동화가 곧 인력 감축을 뜻하지 않으며, 오히려 자동화를 관리하고 검증할 인간의 역할이 더 중요해진다고 정리한다.","insights":["에이전트가 늘수록 SaaS 수요는 줄지 않고 늘어난다.","자동화는 인간을 없애지 않고 관리 역할을 만든다.","CLI보다 GUI가 에이전트 작업에 더 자연스럽다.","미래 제품은 AI를 붙이는 게 아니라 협업을 설계해야 한다.","에이전트 시대의 경쟁력은 '혼자'보다 '연결'에 있다."],"keyClips":[{"clipId":"4D3hDmGhFhA:c3:2-8","startSegmentIndex":2,"endSegmentIndex":8,"startTime":1815.32,"endTime":1870.88,"durationSeconds":55.6,"preview":"에이전트 지원 루프","mustSee":false},{"clipId":"4D3hDmGhFhA:c3:11-17","startSegmentIndex":11,"endSegmentIndex":17,"startTime":1884.8,"endTime":1955.72,"durationSeconds":70.9,"preview":"CLI 시대의 종말","mustSee":false},{"clipId":"4D3hDmGhFhA:c3:19-26","startSegmentIndex":19,"endSegmentIndex":26,"startTime":1958.92,"endTime":2014.84,"durationSeconds":55.9,"preview":"두 개의 작업 모드","mustSee":false},{"clipId":"4D3hDmGhFhA:c3:28-47","startSegmentIndex":28,"endSegmentIndex":47,"startTime":2018.44,"endTime":2182.76,"durationSeconds":164.3,"preview":"에이전트끼리 협업","mustSee":false},{"clipId":"4D3hDmGhFhA:c3:49-68","startSegmentIndex":49,"endSegmentIndex":68,"startTime":2186.44,"endTime":2327.16,"durationSeconds":140.7,"preview":"SaaS는 더 강해진다","mustSee":true},{"clipId":"4D3hDmGhFhA:c3:71-78","startSegmentIndex":71,"endSegmentIndex":78,"startTime":2338.2,"endTime":2407.04,"durationSeconds":68.8,"preview":"자동화와 인간","mustSee":false}],"curatedSegments":[{"segmentIndex":45,"text":"And there's a lot of work that's human work that uh it can't be scored until you write it down, but the act of thinking to prompt it or write it down um is uh is something that you can't measure, but like kind of means that even if the benchmarks get saturated, it doesn't mean the same thing as we you totally replace all senior engineers.","startTime":2703.56,"endTime":2707,"durationSeconds":3,"level":"C2","overallScore":9.8,"rationale":"측정 불가한 인간 일의 본질을 짚음."},{"segmentIndex":6,"text":"Automation is a lie. Every agent needs a human. We have so much automation, so much AI, and I also work way more. >> Creativity.","startTime":29,"endTime":35.2,"durationSeconds":6,"level":"B2","overallScore":8.4,"rationale":"핵심 주장과 대비가 매우 선명함."},{"segmentIndex":8,"text":">> What models do in general is they make yesterday's human competence cheap.","startTime":40.72,"endTime":44.44,"durationSeconds":4,"level":"B2","overallScore":8.6,"rationale":"일반화된 통찰이 매우 강함."},{"segmentIndex":41,"text":"is the minute the agent is like not really that useful anymore. And that's why it I think it has started to shift to a more uh one agent per company model because for now like the ideal is uh you basically set up a forward deployed engineer or someone with that sort of profile who's responsible for making sure that agent is working for the whole company.","startTime":907.68,"endTime":931.36,"durationSeconds":24,"level":"C1","overallScore":8.8,"rationale":"회사당 에이전트 모델의 이유를 설명."},{"segmentIndex":29,"text":"And I think the reverse is actually starting to happen and be like really, really valuable in a way that I did not expect, which is take the AI agent that you use all the time on your computer and put a browser in it so it can see everything you're doing.","startTime":1397.4,"endTime":1410.64,"durationSeconds":13,"level":"C1","overallScore":9,"rationale":"핵심 전환점을 명료하게 설명."},{"segmentIndex":73,"text":"Automation is a lie. Um in the sense that every time you automate something, in order to make sure the automation is working well, you need a human on top of it like making sure that it's working well.","startTime":2355.12,"endTime":2366.64,"durationSeconds":12,"level":"B2","overallScore":8.8,"rationale":"자동화 역설을 명확히 말해 통찰과 표현 강함."},{"segmentIndex":47,"text":"And so I think it's really important uh when we think about benchmark progress to think about it from that perspective, which is benchmarks rise on problems that we've framed that we can articulate, that we can score.","startTime":2711.12,"endTime":2745.16,"durationSeconds":34,"level":"C1","overallScore":8.8,"rationale":"벤치마크 해석 원칙을 정리한다."},{"segmentIndex":69,"text":"And I think that's a really important thing when we think about benchmarks is AI is a broadly distributed technology that any human can use and when we are benchmarking against humans, AI against humans, we're actually really always talking about one human using AI versus another human using AI cuz AI doesn't use itself.","startTime":2887.24,"endTime":2908.12,"durationSeconds":21,"level":"C1","overallScore":9,"rationale":"AI 대 인간 비교의 핵심 재정의."},{"segmentIndex":30,"text":"The pattern that I see so far and again, I don't have a total crystal ball, but I do feel like we've seen enough of the new model drops to like have some sense of how this is going is that what a new model drop does or what models do in general is they make yesterday's human competence cheap.","startTime":4418.6,"endTime":4436.48,"durationSeconds":18,"level":"C1","overallScore":8.8,"rationale":"모델의 경제적 효과를 날카롭게 정의."},{"segmentIndex":39,"text":"And I really think that structurally, because of the way the models work, because of the financial incentives of model companies to like make model of model companies to like make them um uh compliant and aligned, structurally, there are always going to be trailing behind those people who are taking the models and using them to make new expertise or make new things that haven't been done that way before for their very particular situation.","startTime":4502.2,"endTime":4527.72,"durationSeconds":26,"level":"C2","overallScore":8.8,"rationale":"구조적 이유로 인간 우위가 남는단 통찰."},{"segmentIndex":62,"text":"Um so, the way to ride the models is like not one specific thing cuz they're always changing, but it is to be curious and playful, to apply the model the new model to whatever it is that you care about, whether that's your job or something outside of your job and to keep turning over rocks uh because it may not work now, but it may work eventually, it probably will work eventually and the way that you use it matters.","startTime":4680.52,"endTime":4717.6,"durationSeconds":37,"level":"C1","overallScore":9,"rationale":"핵심 조언과 학습 표현이 매우 풍부함."},{"segmentIndex":18,"text":"And I think generally our intuitions about the future the model that I have of what our intuitions are about the future is the intuitions that people had in the Middle Ages about like what happened at the end of the horizon, you know, it's like are there dragons?","startTime":4914.6,"endTime":4931.12,"durationSeconds":17,"level":"C1","overallScore":8.8,"rationale":"미래 인식을 강한 비유로 통찰적으로 설명함."},{"segmentIndex":23,"text":"And until you get to that place where you're starting to see it, and I think we get to see it cuz we get to see it internally all the time, it's important not to let your mind get away from you and being like, this is going to happen and this like, this is going to happen and whatever cuz you're going to tell a story that sounds so real in the moment, but um later on you're like, actually it's much more complex than that and somewhere it's sort of a both everything has changed and nothing has.","startTime":4958.8,"endTime":4986.76,"durationSeconds":28,"level":"C1","overallScore":8.8,"rationale":"미래 예측의 함정을 깊이 있게 경고함."},{"segmentIndex":32,"text":"And try to like try to have fun. I think there's too much of I'm doing this because I have FOMO like it might I might lose my job or like I might miss out on this big thing or whatever and the best way to actually figure out interesting useful things to do with AI is to like do something enjoyable.","startTime":5052.28,"endTime":5071.32,"durationSeconds":19,"level":"C1","overallScore":9,"rationale":"FOMO보다 즐거움이 중요하단 조언이 강함."},{"segmentIndex":3,"text":">> The AI job apocalypse is not really a thing. I am super bullish on PMs and full-stack designers.","startTime":10.48,"endTime":18.28,"durationSeconds":8,"level":"B2","overallScore":7.8,"rationale":"강한 주장과 유용한 구어 표현."},{"segmentIndex":9,"text":"And so, it becomes commoditized. It's not valuable anymore. What humans do is we go in there and we're like,\"Yeah, we have all this frozen human competence from yesterday.","startTime":44.44,"endTime":52.76,"durationSeconds":8,"level":"C1","overallScore":7.8,"rationale":"개념 설명이 선명하고 표현도 좋음."},{"segmentIndex":11,"text":">> It's going to bifurcate in two main ways. One is everyone's going to have at least one agent that they talk to, that they can offload work to.","startTime":57.32,"endTime":64.879,"durationSeconds":8,"level":"B2","overallScore":8,"rationale":"예측과 표현 가치가 모두 높음."},{"segmentIndex":15,"text":"What agents do is increase the number of users of SaaS, not get rid of it.","startTime":80.52,"endTime":84.16,"durationSeconds":4,"level":"B2","overallScore":7.6,"rationale":"핵심 반박 논리가 분명함."},{"segmentIndex":42,"text":"So, just give us a sense of how you work. Thank you. Um I really appreciate the introduction. Um and yeah, I think what one of the things about predicting the future or the way that we think about predicting the future at Every is that you what you don't want to do is prognosticate.","startTime":266.52,"endTime":271.16,"durationSeconds":5,"level":"C1","overallScore":8,"rationale":"미래 예측 태도에 대한 통찰이 큼."},{"segmentIndex":50,"text":"And what I what that does, I think, is it creates this like little pocket of the future where we're all living in it together and we get to be a little bit further ahead cuz at any other company there's like a mix of people.","startTime":320.24,"endTime":330.44,"durationSeconds":10,"level":"C1","overallScore":8,"rationale":"조직이 미래를 앞서가는 이유 설명."}],"generatedAt":"2026-06-25T00:42:59.391Z","keyClipsTotalSec":1550},{"videoId":"4D3hDmGhFhA","chunkIndex":4,"totalChunks":10,"title":"The AI paradox: More automation, more humans, more work | Dan Shipper — Part 5 of 10","thumbnail":"https://i.ytimg.com/vi/4D3hDmGhFhA/maxresdefault.jpg","duration":5646,"uploader":"Lenny's Podcast","youtubeUrl":"https://www.youtube.com/watch?v=4D3hDmGhFhA","keywords":["artificial-intelligence","machine-learning","benchmarking","software-engineering","automation","agents","coding-tools","startup","productivity","saas"],"normalizedKeywords":["엔지니어링","비즈니스·전략","기술 트렌드"],"targetAudience":[{"who":"엔지니어","why":"AI 코딩 도구의 실제 한계와 활용법을 이해하는 데 유용함"},{"who":"창업자","why":"에이전트 시대에 제품과 팀 운영이 어떻게 바뀌는지 볼 수 있음"},{"who":"프로덕트 매니저","why":"벤치마크가 제품 인식과 의사결정에 미치는 영향을 배울 수 있음"},{"who":"지식노동자","why":"자동화가 늘어도 일이 줄지 않는 이유를 설명해 줌"}],"normalizedAudience":["엔지니어·개발자","창업자·스타트업","프로덕트 매니저·기획자"],"summary":"이 영상은 AI 자동화가 늘어났는데도 왜 사람의 일이 더 많아지는지에 대한 역설을 다룬다. 화자는 자신이 만든 '시니어 엔지니어 벤치마크' 사례를 통해, 현재의 AI는 겉보기보다 훨씬 덜 자율적이며 벤치마크는 모델의 실제 능력보다 문제를 어떻게 프레이밍했는지에 따라 점수를 왜곡한다고 주장한다.\n\n핵심 메시지는 '벤치마크 점수 상승 = 인간 대체'가 아니라는 점이다. AI는 아직도 인간의 판단·감독·재프레이밍이 필요하고, 실제 업무에서는 한 인간이 AI를 쓰는 방식의 차이가 성능을 가른다. 그래서 그는 여전히 엔지니어를 고용하고, 기업은 에이전트가 제품을 쉽게 쓰게 만드는 방향으로 준비해야 한다고 말한다.","insights":["벤치마크가 높아져도 인간 대체를 뜻하진 않는다.","AI 성능은 '프레이밍'을 얼마나 잘했는지에 크게 좌우된다.","모델은 스스로 큰 구조를 갈아엎기보다 작은 패치로 회피한다.","실제 업무의 병목은 자동화보다 인간의 판단과 감독이다.","에이전트 시대의 제품은 '사람+에이전트 동시 사용'을 전제로 해야 한다."],"keyClips":[{"clipId":"4D3hDmGhFhA:c4:1-8","startSegmentIndex":1,"endSegmentIndex":8,"startTime":2402.07,"endTime":2451.12,"durationSeconds":49,"preview":"자동화 역설의 출발점","mustSee":false},{"clipId":"4D3hDmGhFhA:c4:9-17","startSegmentIndex":9,"endSegmentIndex":17,"startTime":2451.12,"endTime":2505.52,"durationSeconds":54.4,"preview":"벤치마크의 함정","mustSee":false},{"clipId":"4D3hDmGhFhA:c4:18-31","startSegmentIndex":18,"endSegmentIndex":31,"startTime":2505.52,"endTime":2583.36,"durationSeconds":77.8,"preview":"시니어 엔지니어 기준","mustSee":false},{"clipId":"4D3hDmGhFhA:c4:35-48","startSegmentIndex":35,"endSegmentIndex":48,"startTime":2608.52,"endTime":2752.36,"durationSeconds":143.8,"preview":"프롬프트와 인간판단","mustSee":true},{"clipId":"4D3hDmGhFhA:c4:60-70","startSegmentIndex":60,"endSegmentIndex":70,"startTime":2820.8,"endTime":2917.04,"durationSeconds":96.2,"preview":"AI는 인간의 도구","mustSee":false},{"clipId":"4D3hDmGhFhA:c4:71-82","startSegmentIndex":71,"endSegmentIndex":82,"startTime":2917.04,"endTime":3004.52,"durationSeconds":87.5,"preview":"에이전트 제품의 미래","mustSee":false}],"curatedSegments":[{"segmentIndex":45,"text":"And there's a lot of work that's human work that uh it can't be scored until you write it down, but the act of thinking to prompt it or write it down um is uh is something that you can't measure, but like kind of means that even if the benchmarks get saturated, it doesn't mean the same thing as we you totally replace all senior engineers.","startTime":2703.56,"endTime":2707,"durationSeconds":3,"level":"C2","overallScore":9.8,"rationale":"측정 불가한 인간 일의 본질을 짚음."},{"segmentIndex":6,"text":"Automation is a lie. Every agent needs a human. We have so much automation, so much AI, and I also work way more. >> Creativity.","startTime":29,"endTime":35.2,"durationSeconds":6,"level":"B2","overallScore":8.4,"rationale":"핵심 주장과 대비가 매우 선명함."},{"segmentIndex":8,"text":">> What models do in general is they make yesterday's human competence cheap.","startTime":40.72,"endTime":44.44,"durationSeconds":4,"level":"B2","overallScore":8.6,"rationale":"일반화된 통찰이 매우 강함."},{"segmentIndex":41,"text":"is the minute the agent is like not really that useful anymore. And that's why it I think it has started to shift to a more uh one agent per company model because for now like the ideal is uh you basically set up a forward deployed engineer or someone with that sort of profile who's responsible for making sure that agent is working for the whole company.","startTime":907.68,"endTime":931.36,"durationSeconds":24,"level":"C1","overallScore":8.8,"rationale":"회사당 에이전트 모델의 이유를 설명."},{"segmentIndex":29,"text":"And I think the reverse is actually starting to happen and be like really, really valuable in a way that I did not expect, which is take the AI agent that you use all the time on your computer and put a browser in it so it can see everything you're doing.","startTime":1397.4,"endTime":1410.64,"durationSeconds":13,"level":"C1","overallScore":9,"rationale":"핵심 전환점을 명료하게 설명."},{"segmentIndex":73,"text":"Automation is a lie. Um in the sense that every time you automate something, in order to make sure the automation is working well, you need a human on top of it like making sure that it's working well.","startTime":2355.12,"endTime":2366.64,"durationSeconds":12,"level":"B2","overallScore":8.8,"rationale":"자동화 역설을 명확히 말해 통찰과 표현 강함."},{"segmentIndex":47,"text":"And so I think it's really important uh when we think about benchmark progress to think about it from that perspective, which is benchmarks rise on problems that we've framed that we can articulate, that we can score.","startTime":2711.12,"endTime":2745.16,"durationSeconds":34,"level":"C1","overallScore":8.8,"rationale":"벤치마크 해석 원칙을 정리한다."},{"segmentIndex":69,"text":"And I think that's a really important thing when we think about benchmarks is AI is a broadly distributed technology that any human can use and when we are benchmarking against humans, AI against humans, we're actually really always talking about one human using AI versus another human using AI cuz AI doesn't use itself.","startTime":2887.24,"endTime":2908.12,"durationSeconds":21,"level":"C1","overallScore":9,"rationale":"AI 대 인간 비교의 핵심 재정의."},{"segmentIndex":30,"text":"The pattern that I see so far and again, I don't have a total crystal ball, but I do feel like we've seen enough of the new model drops to like have some sense of how this is going is that what a new model drop does or what models do in general is they make yesterday's human competence cheap.","startTime":4418.6,"endTime":4436.48,"durationSeconds":18,"level":"C1","overallScore":8.8,"rationale":"모델의 경제적 효과를 날카롭게 정의."},{"segmentIndex":39,"text":"And I really think that structurally, because of the way the models work, because of the financial incentives of model companies to like make model of model companies to like make them um uh compliant and aligned, structurally, there are always going to be trailing behind those people who are taking the models and using them to make new expertise or make new things that haven't been done that way before for their very particular situation.","startTime":4502.2,"endTime":4527.72,"durationSeconds":26,"level":"C2","overallScore":8.8,"rationale":"구조적 이유로 인간 우위가 남는단 통찰."},{"segmentIndex":62,"text":"Um so, the way to ride the models is like not one specific thing cuz they're always changing, but it is to be curious and playful, to apply the model the new model to whatever it is that you care about, whether that's your job or something outside of your job and to keep turning over rocks uh because it may not work now, but it may work eventually, it probably will work eventually and the way that you use it matters.","startTime":4680.52,"endTime":4717.6,"durationSeconds":37,"level":"C1","overallScore":9,"rationale":"핵심 조언과 학습 표현이 매우 풍부함."},{"segmentIndex":18,"text":"And I think generally our intuitions about the future the model that I have of what our intuitions are about the future is the intuitions that people had in the Middle Ages about like what happened at the end of the horizon, you know, it's like are there dragons?","startTime":4914.6,"endTime":4931.12,"durationSeconds":17,"level":"C1","overallScore":8.8,"rationale":"미래 인식을 강한 비유로 통찰적으로 설명함."},{"segmentIndex":23,"text":"And until you get to that place where you're starting to see it, and I think we get to see it cuz we get to see it internally all the time, it's important not to let your mind get away from you and being like, this is going to happen and this like, this is going to happen and whatever cuz you're going to tell a story that sounds so real in the moment, but um later on you're like, actually it's much more complex than that and somewhere it's sort of a both everything has changed and nothing has.","startTime":4958.8,"endTime":4986.76,"durationSeconds":28,"level":"C1","overallScore":8.8,"rationale":"미래 예측의 함정을 깊이 있게 경고함."},{"segmentIndex":32,"text":"And try to like try to have fun. I think there's too much of I'm doing this because I have FOMO like it might I might lose my job or like I might miss out on this big thing or whatever and the best way to actually figure out interesting useful things to do with AI is to like do something enjoyable.","startTime":5052.28,"endTime":5071.32,"durationSeconds":19,"level":"C1","overallScore":9,"rationale":"FOMO보다 즐거움이 중요하단 조언이 강함."},{"segmentIndex":3,"text":">> The AI job apocalypse is not really a thing. I am super bullish on PMs and full-stack designers.","startTime":10.48,"endTime":18.28,"durationSeconds":8,"level":"B2","overallScore":7.8,"rationale":"강한 주장과 유용한 구어 표현."},{"segmentIndex":9,"text":"And so, it becomes commoditized. It's not valuable anymore. What humans do is we go in there and we're like,\"Yeah, we have all this frozen human competence from yesterday.","startTime":44.44,"endTime":52.76,"durationSeconds":8,"level":"C1","overallScore":7.8,"rationale":"개념 설명이 선명하고 표현도 좋음."},{"segmentIndex":11,"text":">> It's going to bifurcate in two main ways. One is everyone's going to have at least one agent that they talk to, that they can offload work to.","startTime":57.32,"endTime":64.879,"durationSeconds":8,"level":"B2","overallScore":8,"rationale":"예측과 표현 가치가 모두 높음."},{"segmentIndex":15,"text":"What agents do is increase the number of users of SaaS, not get rid of it.","startTime":80.52,"endTime":84.16,"durationSeconds":4,"level":"B2","overallScore":7.6,"rationale":"핵심 반박 논리가 분명함."},{"segmentIndex":42,"text":"So, just give us a sense of how you work. Thank you. Um I really appreciate the introduction. Um and yeah, I think what one of the things about predicting the future or the way that we think about predicting the future at Every is that you what you don't want to do is prognosticate.","startTime":266.52,"endTime":271.16,"durationSeconds":5,"level":"C1","overallScore":8,"rationale":"미래 예측 태도에 대한 통찰이 큼."},{"segmentIndex":50,"text":"And what I what that does, I think, is it creates this like little pocket of the future where we're all living in it together and we get to be a little bit further ahead cuz at any other company there's like a mix of people.","startTime":320.24,"endTime":330.44,"durationSeconds":10,"level":"C1","overallScore":8,"rationale":"조직이 미래를 앞서가는 이유 설명."}],"generatedAt":"2026-06-25T00:43:20.154Z","keyClipsTotalSec":1550},{"videoId":"4D3hDmGhFhA","chunkIndex":5,"totalChunks":10,"title":"The AI paradox: More automation, more humans, more work | Dan Shipper — Part 6 of 10","thumbnail":"https://i.ytimg.com/vi/4D3hDmGhFhA/maxresdefault.jpg","duration":5646,"uploader":"Lenny's Podcast","youtubeUrl":"https://www.youtube.com/watch?v=4D3hDmGhFhA","keywords":["ai","automation","future-of-work","agentic-workflows","engineering","product-management","data-science","leadership","organizational-design","consulting"],"normalizedKeywords":["리더십·매니지먼트","엔지니어링","기술 트렌드"],"targetAudience":[{"who":"기술 조직 리더","why":"AI 도입 후 역할 재편과 책임 분배를 설계하는 데 유용함"},{"who":"엔지니어","why":"에이전트 운영, 코드 리뷰, 새로운 업무 형태를 이해할 수 있음"},{"who":"PM·비기술 직군","why":"비기술 인력의 기술 업무 참여와 경계 변화가 무엇인지 알 수 있음"}],"normalizedAudience":["엔지니어·개발자","지식노동자 일반"],"summary":"이 구간은 AI가 단순히 일을 없애는 것이 아니라, 오히려 새로운 일과 새로운 역할을 폭발적으로 만든다는 관점을 밀도 있게 보여준다. 비기술 인력도 PR을 올리고 코드를 만지면서 업무 경계가 무너지고, 그 결과 기술 팀은 더 많은 산출물을 리뷰·통합·정리하는 역할을 떠안게 된다. 동시에 '모두가 다 할 수 있는 시대'가 되면서 직무 정체성이 흔들리지만, 결국에는 일반화된 업무 방식이 새로운 표준으로 자리 잡을 것이라고 본다.\n\n핵심은 자동화의 본질이 인간 제거가 아니라 인간의 역할 재배치라는 점이다. 특히 forward deployed engineer처럼 에이전트를 직접 관리하고 조정하는 새로운 직무가 부상하고, 데이터 사이언스/엔지니어링 조직은 잡음을 걸러내는 시스템과 에이전트를 따로 만들어야 한다. 마지막으로 CEO와 리더는 이 변화를 남에게 맡길 수 없고, 직접 AI를 써보며 감각을 쌓아야 한다는 메시지로 정리된다.","insights":["AI는 일을 없애기보다 일의 형태를 바꾼다.","생산성 증가의 대가로 리뷰·통합 노동이 커진다.","에이전트가 늘수록 인간의 감독 역할도 중요해진다.","직무 경계는 흐려지지만, 결국 각자의 책임은 다시 정리된다.","리더는 AI를 위임하지 말고 직접 써봐야 감각이 생긴다."],"keyClips":[{"clipId":"4D3hDmGhFhA:c5:3-11","startSegmentIndex":3,"endSegmentIndex":11,"startTime":3010.12,"endTime":3086,"durationSeconds":75.9,"preview":"일의 형태가 바뀐다","mustSee":false},{"clipId":"4D3hDmGhFhA:c5:14-23","startSegmentIndex":14,"endSegmentIndex":23,"startTime":3095.48,"endTime":3172.56,"durationSeconds":77.1,"preview":"홍수처럼 늘어난 PR","mustSee":false},{"clipId":"4D3hDmGhFhA:c5:24-31","startSegmentIndex":24,"endSegmentIndex":31,"startTime":3172.56,"endTime":3225.08,"durationSeconds":52.5,"preview":"직무 경계의 혼란","mustSee":false},{"clipId":"4D3hDmGhFhA:c5:32-46","startSegmentIndex":32,"endSegmentIndex":46,"startTime":3225.08,"endTime":3345.72,"durationSeconds":120.6,"preview":"에이전트엔 인간이 필요","mustSee":false},{"clipId":"4D3hDmGhFhA:c5:52-62","startSegmentIndex":52,"endSegmentIndex":62,"startTime":3377.84,"endTime":3505.84,"durationSeconds":128,"preview":"잡음 제거가 핵심","mustSee":false},{"clipId":"4D3hDmGhFhA:c5:64-78","startSegmentIndex":64,"endSegmentIndex":78,"startTime":3509.96,"endTime":3604.16,"durationSeconds":94.2,"preview":"리더가 직접 써야","mustSee":true}],"curatedSegments":[{"segmentIndex":45,"text":"And there's a lot of work that's human work that uh it can't be scored until you write it down, but the act of thinking to prompt it or write it down um is uh is something that you can't measure, but like kind of means that even if the benchmarks get saturated, it doesn't mean the same thing as we you totally replace all senior engineers.","startTime":2703.56,"endTime":2707,"durationSeconds":3,"level":"C2","overallScore":9.8,"rationale":"측정 불가한 인간 일의 본질을 짚음."},{"segmentIndex":6,"text":"Automation is a lie. Every agent needs a human. We have so much automation, so much AI, and I also work way more. >> Creativity.","startTime":29,"endTime":35.2,"durationSeconds":6,"level":"B2","overallScore":8.4,"rationale":"핵심 주장과 대비가 매우 선명함."},{"segmentIndex":8,"text":">> What models do in general is they make yesterday's human competence cheap.","startTime":40.72,"endTime":44.44,"durationSeconds":4,"level":"B2","overallScore":8.6,"rationale":"일반화된 통찰이 매우 강함."},{"segmentIndex":41,"text":"is the minute the agent is like not really that useful anymore. And that's why it I think it has started to shift to a more uh one agent per company model because for now like the ideal is uh you basically set up a forward deployed engineer or someone with that sort of profile who's responsible for making sure that agent is working for the whole company.","startTime":907.68,"endTime":931.36,"durationSeconds":24,"level":"C1","overallScore":8.8,"rationale":"회사당 에이전트 모델의 이유를 설명."},{"segmentIndex":29,"text":"And I think the reverse is actually starting to happen and be like really, really valuable in a way that I did not expect, which is take the AI agent that you use all the time on your computer and put a browser in it so it can see everything you're doing.","startTime":1397.4,"endTime":1410.64,"durationSeconds":13,"level":"C1","overallScore":9,"rationale":"핵심 전환점을 명료하게 설명."},{"segmentIndex":73,"text":"Automation is a lie. Um in the sense that every time you automate something, in order to make sure the automation is working well, you need a human on top of it like making sure that it's working well.","startTime":2355.12,"endTime":2366.64,"durationSeconds":12,"level":"B2","overallScore":8.8,"rationale":"자동화 역설을 명확히 말해 통찰과 표현 강함."},{"segmentIndex":47,"text":"And so I think it's really important uh when we think about benchmark progress to think about it from that perspective, which is benchmarks rise on problems that we've framed that we can articulate, that we can score.","startTime":2711.12,"endTime":2745.16,"durationSeconds":34,"level":"C1","overallScore":8.8,"rationale":"벤치마크 해석 원칙을 정리한다."},{"segmentIndex":69,"text":"And I think that's a really important thing when we think about benchmarks is AI is a broadly distributed technology that any human can use and when we are benchmarking against humans, AI against humans, we're actually really always talking about one human using AI versus another human using AI cuz AI doesn't use itself.","startTime":2887.24,"endTime":2908.12,"durationSeconds":21,"level":"C1","overallScore":9,"rationale":"AI 대 인간 비교의 핵심 재정의."},{"segmentIndex":30,"text":"The pattern that I see so far and again, I don't have a total crystal ball, but I do feel like we've seen enough of the new model drops to like have some sense of how this is going is that what a new model drop does or what models do in general is they make yesterday's human competence cheap.","startTime":4418.6,"endTime":4436.48,"durationSeconds":18,"level":"C1","overallScore":8.8,"rationale":"모델의 경제적 효과를 날카롭게 정의."},{"segmentIndex":39,"text":"And I really think that structurally, because of the way the models work, because of the financial incentives of model companies to like make model of model companies to like make them um uh compliant and aligned, structurally, there are always going to be trailing behind those people who are taking the models and using them to make new expertise or make new things that haven't been done that way before for their very particular situation.","startTime":4502.2,"endTime":4527.72,"durationSeconds":26,"level":"C2","overallScore":8.8,"rationale":"구조적 이유로 인간 우위가 남는단 통찰."},{"segmentIndex":62,"text":"Um so, the way to ride the models is like not one specific thing cuz they're always changing, but it is to be curious and playful, to apply the model the new model to whatever it is that you care about, whether that's your job or something outside of your job and to keep turning over rocks uh because it may not work now, but it may work eventually, it probably will work eventually and the way that you use it matters.","startTime":4680.52,"endTime":4717.6,"durationSeconds":37,"level":"C1","overallScore":9,"rationale":"핵심 조언과 학습 표현이 매우 풍부함."},{"segmentIndex":18,"text":"And I think generally our intuitions about the future the model that I have of what our intuitions are about the future is the intuitions that people had in the Middle Ages about like what happened at the end of the horizon, you know, it's like are there dragons?","startTime":4914.6,"endTime":4931.12,"durationSeconds":17,"level":"C1","overallScore":8.8,"rationale":"미래 인식을 강한 비유로 통찰적으로 설명함."},{"segmentIndex":23,"text":"And until you get to that place where you're starting to see it, and I think we get to see it cuz we get to see it internally all the time, it's important not to let your mind get away from you and being like, this is going to happen and this like, this is going to happen and whatever cuz you're going to tell a story that sounds so real in the moment, but um later on you're like, actually it's much more complex than that and somewhere it's sort of a both everything has changed and nothing has.","startTime":4958.8,"endTime":4986.76,"durationSeconds":28,"level":"C1","overallScore":8.8,"rationale":"미래 예측의 함정을 깊이 있게 경고함."},{"segmentIndex":32,"text":"And try to like try to have fun. I think there's too much of I'm doing this because I have FOMO like it might I might lose my job or like I might miss out on this big thing or whatever and the best way to actually figure out interesting useful things to do with AI is to like do something enjoyable.","startTime":5052.28,"endTime":5071.32,"durationSeconds":19,"level":"C1","overallScore":9,"rationale":"FOMO보다 즐거움이 중요하단 조언이 강함."},{"segmentIndex":3,"text":">> The AI job apocalypse is not really a thing. I am super bullish on PMs and full-stack designers.","startTime":10.48,"endTime":18.28,"durationSeconds":8,"level":"B2","overallScore":7.8,"rationale":"강한 주장과 유용한 구어 표현."},{"segmentIndex":9,"text":"And so, it becomes commoditized. It's not valuable anymore. What humans do is we go in there and we're like,\"Yeah, we have all this frozen human competence from yesterday.","startTime":44.44,"endTime":52.76,"durationSeconds":8,"level":"C1","overallScore":7.8,"rationale":"개념 설명이 선명하고 표현도 좋음."},{"segmentIndex":11,"text":">> It's going to bifurcate in two main ways. One is everyone's going to have at least one agent that they talk to, that they can offload work to.","startTime":57.32,"endTime":64.879,"durationSeconds":8,"level":"B2","overallScore":8,"rationale":"예측과 표현 가치가 모두 높음."},{"segmentIndex":15,"text":"What agents do is increase the number of users of SaaS, not get rid of it.","startTime":80.52,"endTime":84.16,"durationSeconds":4,"level":"B2","overallScore":7.6,"rationale":"핵심 반박 논리가 분명함."},{"segmentIndex":42,"text":"So, just give us a sense of how you work. Thank you. Um I really appreciate the introduction. Um and yeah, I think what one of the things about predicting the future or the way that we think about predicting the future at Every is that you what you don't want to do is prognosticate.","startTime":266.52,"endTime":271.16,"durationSeconds":5,"level":"C1","overallScore":8,"rationale":"미래 예측 태도에 대한 통찰이 큼."},{"segmentIndex":50,"text":"And what I what that does, I think, is it creates this like little pocket of the future where we're all living in it together and we get to be a little bit further ahead cuz at any other company there's like a mix of people.","startTime":320.24,"endTime":330.44,"durationSeconds":10,"level":"C1","overallScore":8,"rationale":"조직이 미래를 앞서가는 이유 설명."}],"generatedAt":"2026-06-25T00:53:44.047Z","keyClipsTotalSec":1550},{"videoId":"4D3hDmGhFhA","chunkIndex":6,"totalChunks":10,"title":"The AI paradox: More automation, more humans, more work | Dan Shipper — Part 7 of 10","thumbnail":"https://i.ytimg.com/vi/4D3hDmGhFhA/maxresdefault.jpg","duration":5646,"uploader":"Lenny's Podcast","youtubeUrl":"https://www.youtube.com/watch?v=4D3hDmGhFhA","keywords":["ai","automation","sales","recruiting","product-management","writing","knowledge-work","agents","business","future-of-work"],"normalizedKeywords":["프로덕트","비즈니스·전략","기술 트렌드"],"targetAudience":[{"who":"프로덕트 매니저","why":"AI 시대에 PM의 역할과 경쟁력이 어떻게 달라지는지 직접적으로 다룸"},{"who":"지식노동자","why":"문서, 이메일, 기획 등 일반 사무 업무가 AI로 바뀌는 흐름을 이해할 수 있음"},{"who":"창업자","why":"조직에서 AI를 어떻게 도입하고 어떤 인재가 유리한지 판단하는 데 도움됨"}],"normalizedAudience":["프로덕트 매니저·기획자","지식노동자 일반","창업자·스타트업"],"summary":"이 구간은 AI가 특정 직무를 대체하는 방식보다, 조직 안에서 일의 형태를 어떻게 바꾸는지에 초점을 맞춘다. 특히 sales, recruiting, 문서 작성, 이메일, 전략 기획처럼 지식노동의 상단과 중간 단계를 AI가 크게 보조하면서, 사람은 리뷰·판단·조율에 더 많이 시간을 쓰게 된다고 말한다.\n\n또한 AI가 만든 결과물에 대한 반감은 점점 약해지고, 오히려 잘 지시된 AI 산출물이 인간이 대충 쓴 문서보다 더 낫다는 태도를 보여준다. 마지막으로, 이런 환경에서 가장 유리한 사람은 강한 제품 감각과 약간의 기술 이해를 함께 가진 PM이며, 실제 사례로 Marcus를 들어 그 조합이 얼마나 강력한지 강조한다.","insights":["AI는 일을 없애기보다, 일을 '검토와 조율' 중심으로 바꾼다.","좋은 AI 산출물은 인간의 대충 쓴 문서보다 실용적일 수 있다.","문서와 이메일은 이미 '사람이 직접 써야만 한다'는 영역이 아니다.","PM은 기술을 완전하게 아는 사람보다, 제품 감각과 도구 활용력이 중요해진다.","AI 시대의 경쟁력은 코딩 능력보다 문제를 잘 정의하는 힘이다."],"keyClips":[{"clipId":"4D3hDmGhFhA:c6:2-14","startSegmentIndex":2,"endSegmentIndex":14,"startTime":3605.96,"endTime":3706,"durationSeconds":100,"preview":"세일즈와 AI 변화","mustSee":false},{"clipId":"4D3hDmGhFhA:c6:15-23","startSegmentIndex":15,"endSegmentIndex":23,"startTime":3706,"endTime":3789.08,"durationSeconds":83.1,"preview":"일의 형태가 바뀐다","mustSee":false},{"clipId":"4D3hDmGhFhA:c6:24-58","startSegmentIndex":24,"endSegmentIndex":58,"startTime":3789.08,"endTime":4100.759,"durationSeconds":311.7,"preview":"AI 글쓰기의 일상화","mustSee":true},{"clipId":"4D3hDmGhFhA:c6:61-70","startSegmentIndex":61,"endSegmentIndex":70,"startTime":4119.759,"endTime":4209.56,"durationSeconds":89.8,"preview":"PM이 유리한 이유","mustSee":true}],"curatedSegments":[{"segmentIndex":45,"text":"And there's a lot of work that's human work that uh it can't be scored until you write it down, but the act of thinking to prompt it or write it down um is uh is something that you can't measure, but like kind of means that even if the benchmarks get saturated, it doesn't mean the same thing as we you totally replace all senior engineers.","startTime":2703.56,"endTime":2707,"durationSeconds":3,"level":"C2","overallScore":9.8,"rationale":"측정 불가한 인간 일의 본질을 짚음."},{"segmentIndex":6,"text":"Automation is a lie. Every agent needs a human. We have so much automation, so much AI, and I also work way more. >> Creativity.","startTime":29,"endTime":35.2,"durationSeconds":6,"level":"B2","overallScore":8.4,"rationale":"핵심 주장과 대비가 매우 선명함."},{"segmentIndex":8,"text":">> What models do in general is they make yesterday's human competence cheap.","startTime":40.72,"endTime":44.44,"durationSeconds":4,"level":"B2","overallScore":8.6,"rationale":"일반화된 통찰이 매우 강함."},{"segmentIndex":41,"text":"is the minute the agent is like not really that useful anymore. And that's why it I think it has started to shift to a more uh one agent per company model because for now like the ideal is uh you basically set up a forward deployed engineer or someone with that sort of profile who's responsible for making sure that agent is working for the whole company.","startTime":907.68,"endTime":931.36,"durationSeconds":24,"level":"C1","overallScore":8.8,"rationale":"회사당 에이전트 모델의 이유를 설명."},{"segmentIndex":29,"text":"And I think the reverse is actually starting to happen and be like really, really valuable in a way that I did not expect, which is take the AI agent that you use all the time on your computer and put a browser in it so it can see everything you're doing.","startTime":1397.4,"endTime":1410.64,"durationSeconds":13,"level":"C1","overallScore":9,"rationale":"핵심 전환점을 명료하게 설명."},{"segmentIndex":73,"text":"Automation is a lie. Um in the sense that every time you automate something, in order to make sure the automation is working well, you need a human on top of it like making sure that it's working well.","startTime":2355.12,"endTime":2366.64,"durationSeconds":12,"level":"B2","overallScore":8.8,"rationale":"자동화 역설을 명확히 말해 통찰과 표현 강함."},{"segmentIndex":47,"text":"And so I think it's really important uh when we think about benchmark progress to think about it from that perspective, which is benchmarks rise on problems that we've framed that we can articulate, that we can score.","startTime":2711.12,"endTime":2745.16,"durationSeconds":34,"level":"C1","overallScore":8.8,"rationale":"벤치마크 해석 원칙을 정리한다."},{"segmentIndex":69,"text":"And I think that's a really important thing when we think about benchmarks is AI is a broadly distributed technology that any human can use and when we are benchmarking against humans, AI against humans, we're actually really always talking about one human using AI versus another human using AI cuz AI doesn't use itself.","startTime":2887.24,"endTime":2908.12,"durationSeconds":21,"level":"C1","overallScore":9,"rationale":"AI 대 인간 비교의 핵심 재정의."},{"segmentIndex":30,"text":"The pattern that I see so far and again, I don't have a total crystal ball, but I do feel like we've seen enough of the new model drops to like have some sense of how this is going is that what a new model drop does or what models do in general is they make yesterday's human competence cheap.","startTime":4418.6,"endTime":4436.48,"durationSeconds":18,"level":"C1","overallScore":8.8,"rationale":"모델의 경제적 효과를 날카롭게 정의."},{"segmentIndex":39,"text":"And I really think that structurally, because of the way the models work, because of the financial incentives of model companies to like make model of model companies to like make them um uh compliant and aligned, structurally, there are always going to be trailing behind those people who are taking the models and using them to make new expertise or make new things that haven't been done that way before for their very particular situation.","startTime":4502.2,"endTime":4527.72,"durationSeconds":26,"level":"C2","overallScore":8.8,"rationale":"구조적 이유로 인간 우위가 남는단 통찰."},{"segmentIndex":62,"text":"Um so, the way to ride the models is like not one specific thing cuz they're always changing, but it is to be curious and playful, to apply the model the new model to whatever it is that you care about, whether that's your job or something outside of your job and to keep turning over rocks uh because it may not work now, but it may work eventually, it probably will work eventually and the way that you use it matters.","startTime":4680.52,"endTime":4717.6,"durationSeconds":37,"level":"C1","overallScore":9,"rationale":"핵심 조언과 학습 표현이 매우 풍부함."},{"segmentIndex":18,"text":"And I think generally our intuitions about the future the model that I have of what our intuitions are about the future is the intuitions that people had in the Middle Ages about like what happened at the end of the horizon, you know, it's like are there dragons?","startTime":4914.6,"endTime":4931.12,"durationSeconds":17,"level":"C1","overallScore":8.8,"rationale":"미래 인식을 강한 비유로 통찰적으로 설명함."},{"segmentIndex":23,"text":"And until you get to that place where you're starting to see it, and I think we get to see it cuz we get to see it internally all the time, it's important not to let your mind get away from you and being like, this is going to happen and this like, this is going to happen and whatever cuz you're going to tell a story that sounds so real in the moment, but um later on you're like, actually it's much more complex than that and somewhere it's sort of a both everything has changed and nothing has.","startTime":4958.8,"endTime":4986.76,"durationSeconds":28,"level":"C1","overallScore":8.8,"rationale":"미래 예측의 함정을 깊이 있게 경고함."},{"segmentIndex":32,"text":"And try to like try to have fun. I think there's too much of I'm doing this because I have FOMO like it might I might lose my job or like I might miss out on this big thing or whatever and the best way to actually figure out interesting useful things to do with AI is to like do something enjoyable.","startTime":5052.28,"endTime":5071.32,"durationSeconds":19,"level":"C1","overallScore":9,"rationale":"FOMO보다 즐거움이 중요하단 조언이 강함."},{"segmentIndex":3,"text":">> The AI job apocalypse is not really a thing. I am super bullish on PMs and full-stack designers.","startTime":10.48,"endTime":18.28,"durationSeconds":8,"level":"B2","overallScore":7.8,"rationale":"강한 주장과 유용한 구어 표현."},{"segmentIndex":9,"text":"And so, it becomes commoditized. It's not valuable anymore. What humans do is we go in there and we're like,\"Yeah, we have all this frozen human competence from yesterday.","startTime":44.44,"endTime":52.76,"durationSeconds":8,"level":"C1","overallScore":7.8,"rationale":"개념 설명이 선명하고 표현도 좋음."},{"segmentIndex":11,"text":">> It's going to bifurcate in two main ways. One is everyone's going to have at least one agent that they talk to, that they can offload work to.","startTime":57.32,"endTime":64.879,"durationSeconds":8,"level":"B2","overallScore":8,"rationale":"예측과 표현 가치가 모두 높음."},{"segmentIndex":15,"text":"What agents do is increase the number of users of SaaS, not get rid of it.","startTime":80.52,"endTime":84.16,"durationSeconds":4,"level":"B2","overallScore":7.6,"rationale":"핵심 반박 논리가 분명함."},{"segmentIndex":42,"text":"So, just give us a sense of how you work. Thank you. Um I really appreciate the introduction. Um and yeah, I think what one of the things about predicting the future or the way that we think about predicting the future at Every is that you what you don't want to do is prognosticate.","startTime":266.52,"endTime":271.16,"durationSeconds":5,"level":"C1","overallScore":8,"rationale":"미래 예측 태도에 대한 통찰이 큼."},{"segmentIndex":50,"text":"And what I what that does, I think, is it creates this like little pocket of the future where we're all living in it together and we get to be a little bit further ahead cuz at any other company there's like a mix of people.","startTime":320.24,"endTime":330.44,"durationSeconds":10,"level":"C1","overallScore":8,"rationale":"조직이 미래를 앞서가는 이유 설명."}],"generatedAt":"2026-06-25T00:54:02.614Z","keyClipsTotalSec":1550},{"videoId":"4D3hDmGhFhA","chunkIndex":7,"totalChunks":10,"title":"The AI paradox: More automation, more humans, more work | Dan Shipper — Part 8 of 10","thumbnail":"https://i.ytimg.com/vi/4D3hDmGhFhA/maxresdefault.jpg","duration":5646,"uploader":"Lenny's Podcast","youtubeUrl":"https://www.youtube.com/watch?v=4D3hDmGhFhA","keywords":["artificial-intelligence","productivity","product-management","design","software-development","future-of-work","career","automation","startup","model-crafting"],"normalizedKeywords":["프로덕트","디자인","커리어·성장"],"targetAudience":[{"who":"프로덕트 매니저","why":"AI를 활용해 무엇을 만들지·어떻게 우선순위를 정할지에 대한 관점이 직접적이다."},{"who":"프로덕트 디자이너","why":"디자이너가 AI로 직접 만들고 전달하는 방식의 변화가 핵심이다."},{"who":"엔지니어","why":"모델이 코드와 개발 업무를 바꾸는 방식, 그리고 엔지니어 수요가 왜 유지되는지 다룬다."},{"who":"지식노동자","why":"직무를 막론하고 AI를 업무에 어떻게 붙여야 뒤처지지 않는지에 대한 조언이 중심이다."}],"normalizedAudience":["프로덕트 매니저·기획자","프로덕트 디자이너","엔지니어·개발자"],"summary":"이 구간은 AI가 일자리를 한꺼번에 없애기보다, '어제까지 인간이 하던 역량'을 싸게 만들어서 업무 방식을 재편한다는 관점을 강하게 밀어붙인다. 그래서 PM과 디자이너, 엔지니어 모두 사라지기보다 더 많이 필요해질 수 있으며, 특히 AI를 잘 다루는 사람은 더 빠르게 만들고 더 넓게 문제를 볼 수 있다고 말한다.\n\n핵심 처방은 단순하다. 모델을 두려워하며 피하지 말고, 자신의 일에 붙여서 계속 시험해 보라는 것이다. 새 모델이 나올 때마다 '돌 밑을 다시 들어올려' 무엇이 가능한지 확인하고, 현실의 구체적인 업무와 결합해 쓰는 사람이 진짜 최전선에 선다는 메시지다.","insights":["AI는 일자리를 즉시 없애기보다 기존 역량을 싸게 만든다.","차별화는 모델 기본값이 아니라 인간의 재해석에서 나온다.","PM·디자이너는 AI로 더 많이 만들수록 오히려 더 강해진다.","살아남는 전략은 회피가 아니라 모델을 업무에 계속 붙이는 것이다.","AI의 전선은 샌프란시스코가 아니라 실제 사용 현장이다."],"keyClips":[{"clipId":"4D3hDmGhFhA:c7:1-8","startSegmentIndex":1,"endSegmentIndex":8,"startTime":4201.23,"endTime":4250.04,"durationSeconds":48.8,"preview":"PM이 더 강해진다","mustSee":false},{"clipId":"4D3hDmGhFhA:c7:12-16","startSegmentIndex":12,"endSegmentIndex":16,"startTime":4264.28,"endTime":4325.72,"durationSeconds":61.4,"preview":"디자이너의 역전","mustSee":true},{"clipId":"4D3hDmGhFhA:c7:26-40","startSegmentIndex":26,"endSegmentIndex":40,"startTime":4391.16,"endTime":4534.56,"durationSeconds":143.4,"preview":"일자리 공포 재해석","mustSee":false},{"clipId":"4D3hDmGhFhA:c7:50-62","startSegmentIndex":50,"endSegmentIndex":62,"startTime":4584.84,"endTime":4717.6,"durationSeconds":132.8,"preview":"모델을 타는 법","mustSee":false},{"clipId":"4D3hDmGhFhA:c7:63-74","startSegmentIndex":63,"endSegmentIndex":74,"startTime":4717.6,"endTime":4809.12,"durationSeconds":91.5,"preview":"현장이 최전선이다","mustSee":false}],"curatedSegments":[{"segmentIndex":45,"text":"And there's a lot of work that's human work that uh it can't be scored until you write it down, but the act of thinking to prompt it or write it down um is uh is something that you can't measure, but like kind of means that even if the benchmarks get saturated, it doesn't mean the same thing as we you totally replace all senior engineers.","startTime":2703.56,"endTime":2707,"durationSeconds":3,"level":"C2","overallScore":9.8,"rationale":"측정 불가한 인간 일의 본질을 짚음."},{"segmentIndex":6,"text":"Automation is a lie. Every agent needs a human. We have so much automation, so much AI, and I also work way more. >> Creativity.","startTime":29,"endTime":35.2,"durationSeconds":6,"level":"B2","overallScore":8.4,"rationale":"핵심 주장과 대비가 매우 선명함."},{"segmentIndex":8,"text":">> What models do in general is they make yesterday's human competence cheap.","startTime":40.72,"endTime":44.44,"durationSeconds":4,"level":"B2","overallScore":8.6,"rationale":"일반화된 통찰이 매우 강함."},{"segmentIndex":41,"text":"is the minute the agent is like not really that useful anymore. And that's why it I think it has started to shift to a more uh one agent per company model because for now like the ideal is uh you basically set up a forward deployed engineer or someone with that sort of profile who's responsible for making sure that agent is working for the whole company.","startTime":907.68,"endTime":931.36,"durationSeconds":24,"level":"C1","overallScore":8.8,"rationale":"회사당 에이전트 모델의 이유를 설명."},{"segmentIndex":29,"text":"And I think the reverse is actually starting to happen and be like really, really valuable in a way that I did not expect, which is take the AI agent that you use all the time on your computer and put a browser in it so it can see everything you're doing.","startTime":1397.4,"endTime":1410.64,"durationSeconds":13,"level":"C1","overallScore":9,"rationale":"핵심 전환점을 명료하게 설명."},{"segmentIndex":73,"text":"Automation is a lie. Um in the sense that every time you automate something, in order to make sure the automation is working well, you need a human on top of it like making sure that it's working well.","startTime":2355.12,"endTime":2366.64,"durationSeconds":12,"level":"B2","overallScore":8.8,"rationale":"자동화 역설을 명확히 말해 통찰과 표현 강함."},{"segmentIndex":47,"text":"And so I think it's really important uh when we think about benchmark progress to think about it from that perspective, which is benchmarks rise on problems that we've framed that we can articulate, that we can score.","startTime":2711.12,"endTime":2745.16,"durationSeconds":34,"level":"C1","overallScore":8.8,"rationale":"벤치마크 해석 원칙을 정리한다."},{"segmentIndex":69,"text":"And I think that's a really important thing when we think about benchmarks is AI is a broadly distributed technology that any human can use and when we are benchmarking against humans, AI against humans, we're actually really always talking about one human using AI versus another human using AI cuz AI doesn't use itself.","startTime":2887.24,"endTime":2908.12,"durationSeconds":21,"level":"C1","overallScore":9,"rationale":"AI 대 인간 비교의 핵심 재정의."},{"segmentIndex":30,"text":"The pattern that I see so far and again, I don't have a total crystal ball, but I do feel like we've seen enough of the new model drops to like have some sense of how this is going is that what a new model drop does or what models do in general is they make yesterday's human competence cheap.","startTime":4418.6,"endTime":4436.48,"durationSeconds":18,"level":"C1","overallScore":8.8,"rationale":"모델의 경제적 효과를 날카롭게 정의."},{"segmentIndex":39,"text":"And I really think that structurally, because of the way the models work, because of the financial incentives of model companies to like make model of model companies to like make them um uh compliant and aligned, structurally, there are always going to be trailing behind those people who are taking the models and using them to make new expertise or make new things that haven't been done that way before for their very particular situation.","startTime":4502.2,"endTime":4527.72,"durationSeconds":26,"level":"C2","overallScore":8.8,"rationale":"구조적 이유로 인간 우위가 남는단 통찰."},{"segmentIndex":62,"text":"Um so, the way to ride the models is like not one specific thing cuz they're always changing, but it is to be curious and playful, to apply the model the new model to whatever it is that you care about, whether that's your job or something outside of your job and to keep turning over rocks uh because it may not work now, but it may work eventually, it probably will work eventually and the way that you use it matters.","startTime":4680.52,"endTime":4717.6,"durationSeconds":37,"level":"C1","overallScore":9,"rationale":"핵심 조언과 학습 표현이 매우 풍부함."},{"segmentIndex":18,"text":"And I think generally our intuitions about the future the model that I have of what our intuitions are about the future is the intuitions that people had in the Middle Ages about like what happened at the end of the horizon, you know, it's like are there dragons?","startTime":4914.6,"endTime":4931.12,"durationSeconds":17,"level":"C1","overallScore":8.8,"rationale":"미래 인식을 강한 비유로 통찰적으로 설명함."},{"segmentIndex":23,"text":"And until you get to that place where you're starting to see it, and I think we get to see it cuz we get to see it internally all the time, it's important not to let your mind get away from you and being like, this is going to happen and this like, this is going to happen and whatever cuz you're going to tell a story that sounds so real in the moment, but um later on you're like, actually it's much more complex than that and somewhere it's sort of a both everything has changed and nothing has.","startTime":4958.8,"endTime":4986.76,"durationSeconds":28,"level":"C1","overallScore":8.8,"rationale":"미래 예측의 함정을 깊이 있게 경고함."},{"segmentIndex":32,"text":"And try to like try to have fun. I think there's too much of I'm doing this because I have FOMO like it might I might lose my job or like I might miss out on this big thing or whatever and the best way to actually figure out interesting useful things to do with AI is to like do something enjoyable.","startTime":5052.28,"endTime":5071.32,"durationSeconds":19,"level":"C1","overallScore":9,"rationale":"FOMO보다 즐거움이 중요하단 조언이 강함."},{"segmentIndex":3,"text":">> The AI job apocalypse is not really a thing. I am super bullish on PMs and full-stack designers.","startTime":10.48,"endTime":18.28,"durationSeconds":8,"level":"B2","overallScore":7.8,"rationale":"강한 주장과 유용한 구어 표현."},{"segmentIndex":9,"text":"And so, it becomes commoditized. It's not valuable anymore. What humans do is we go in there and we're like,\"Yeah, we have all this frozen human competence from yesterday.","startTime":44.44,"endTime":52.76,"durationSeconds":8,"level":"C1","overallScore":7.8,"rationale":"개념 설명이 선명하고 표현도 좋음."},{"segmentIndex":11,"text":">> It's going to bifurcate in two main ways. One is everyone's going to have at least one agent that they talk to, that they can offload work to.","startTime":57.32,"endTime":64.879,"durationSeconds":8,"level":"B2","overallScore":8,"rationale":"예측과 표현 가치가 모두 높음."},{"segmentIndex":15,"text":"What agents do is increase the number of users of SaaS, not get rid of it.","startTime":80.52,"endTime":84.16,"durationSeconds":4,"level":"B2","overallScore":7.6,"rationale":"핵심 반박 논리가 분명함."},{"segmentIndex":42,"text":"So, just give us a sense of how you work. Thank you. Um I really appreciate the introduction. Um and yeah, I think what one of the things about predicting the future or the way that we think about predicting the future at Every is that you what you don't want to do is prognosticate.","startTime":266.52,"endTime":271.16,"durationSeconds":5,"level":"C1","overallScore":8,"rationale":"미래 예측 태도에 대한 통찰이 큼."},{"segmentIndex":50,"text":"And what I what that does, I think, is it creates this like little pocket of the future where we're all living in it together and we get to be a little bit further ahead cuz at any other company there's like a mix of people.","startTime":320.24,"endTime":330.44,"durationSeconds":10,"level":"C1","overallScore":8,"rationale":"조직이 미래를 앞서가는 이유 설명."}],"generatedAt":"2026-06-25T00:44:27.471Z","keyClipsTotalSec":1550},{"videoId":"4D3hDmGhFhA","chunkIndex":8,"totalChunks":10,"title":"The AI paradox: More automation, more humans, more work | Dan Shipper — Part 9 of 10","thumbnail":"https://i.ytimg.com/vi/4D3hDmGhFhA/maxresdefault.jpg","duration":5646,"uploader":"Lenny's Podcast","youtubeUrl":"https://www.youtube.com/watch?v=4D3hDmGhFhA","keywords":["artificial-intelligence","future-of-work","automation","productivity","startup","silicon-valley","agentic-ai","writing","books","career-growth"],"normalizedKeywords":["기술 트렌드","비즈니스·전략","커리어·성장"],"targetAudience":[{"who":"창업자","why":"AI를 제품·업무 방식에 어떻게 녹일지 실전 감각을 얻을 수 있음"},{"who":"엔지니어","why":"코딩이 어떻게 바뀌는지와 에이전트 도구 활용법을 직접 들을 수 있음"},{"who":"지식노동자","why":"업무 자동화 시대에 무엇을 시도해야 하는지 실용적 힌트를 줌"}],"normalizedAudience":["창업자·스타트업","엔지니어·개발자","지식노동자 일반"],"summary":"이 구간은 AI가 세상을 완전히 뒤엎을 것이라는 과장과, 아무것도 바뀌지 않을 것이라는 냉소 사이에서 균형을 잡으려는 관점을 제시한다. 화자는 AI가 널리 접근 가능해졌다는 점을 긍정하면서도, 실제 업무는 여전히 이메일·Slack·SaaS 중심으로 이어지고 역할만 크게 재배치되고 있다고 본다. 즉 '모든 것이 바뀌었다'와 '대부분은 그대로다'가 동시에 참이라는 시각이다. 또한 AI 회사들이 위험을 과장하며 공포를 유통하는 PR 전략은 비효율적이고 종종 잘못됐다고 비판한 뒤, 앞으로 1년 동안은 각자의 업무 흐름에 모델과 에이전트 도구를 직접 넣어보고, 즐거운 문제를 골라 AI로 풀어보라고 조언한다.\n\n후반부에서는 책, 다큐, 삶의 모토 같은 개인적 추천이 이어지지만, 핵심 메시지는 'AI 시대의 적응은 거창한 예측보다 직접 써보는 습관'이라는 점으로 모인다. 화자는 재미와 호기심이 있어야 실제로 유용한 사용처를 발견할 수 있다고 강조한다.","insights":["AI는 세상을 뒤엎기보다 역할과 흐름을 재배치한다.","과장된 미래 예측은 현실 적응을 오히려 방해한다.","도구를 직접 써봐야 AI의 진짜 유용성이 보인다.","FOMO보다 즐거움이 더 좋은 학습 동기다.","업무 자동화는 거창한 혁명보다 작은 워크플로 변화다."],"keyClips":[{"clipId":"4D3hDmGhFhA:c8:2-7","startSegmentIndex":2,"endSegmentIndex":7,"startTime":4810.12,"endTime":4862.72,"durationSeconds":52.6,"preview":"AI의 민주화 효과","mustSee":false},{"clipId":"4D3hDmGhFhA:c8:8-14","startSegmentIndex":8,"endSegmentIndex":14,"startTime":4862.72,"endTime":4897.2,"durationSeconds":34.5,"preview":"변한 것과 그대로인 것","mustSee":false},{"clipId":"4D3hDmGhFhA:c8:18-24","startSegmentIndex":18,"endSegmentIndex":24,"startTime":4914.6,"endTime":4991.68,"durationSeconds":77.1,"preview":"미래 과장 경계하기","mustSee":false},{"clipId":"4D3hDmGhFhA:c8:25-28","startSegmentIndex":25,"endSegmentIndex":28,"startTime":4991.68,"endTime":5011.48,"durationSeconds":19.8,"preview":"공포 마케팅 비판","mustSee":false},{"clipId":"4D3hDmGhFhA:c8:29-37","startSegmentIndex":29,"endSegmentIndex":37,"startTime":5011.48,"endTime":5098,"durationSeconds":86.5,"preview":"올해의 AI 실천법","mustSee":true},{"clipId":"4D3hDmGhFhA:c8:44-62","startSegmentIndex":44,"endSegmentIndex":62,"startTime":5131.68,"endTime":5271.68,"durationSeconds":140,"preview":"추천 책과 지적 관심","mustSee":false},{"clipId":"4D3hDmGhFhA:c8:67-77","startSegmentIndex":67,"endSegmentIndex":77,"startTime":5295.04,"endTime":5401.2,"durationSeconds":106.2,"preview":"극한 도전의 매력","mustSee":false}],"curatedSegments":[{"segmentIndex":45,"text":"And there's a lot of work that's human work that uh it can't be scored until you write it down, but the act of thinking to prompt it or write it down um is uh is something that you can't measure, but like kind of means that even if the benchmarks get saturated, it doesn't mean the same thing as we you totally replace all senior engineers.","startTime":2703.56,"endTime":2707,"durationSeconds":3,"level":"C2","overallScore":9.8,"rationale":"측정 불가한 인간 일의 본질을 짚음."},{"segmentIndex":6,"text":"Automation is a lie. Every agent needs a human. We have so much automation, so much AI, and I also work way more. >> Creativity.","startTime":29,"endTime":35.2,"durationSeconds":6,"level":"B2","overallScore":8.4,"rationale":"핵심 주장과 대비가 매우 선명함."},{"segmentIndex":8,"text":">> What models do in general is they make yesterday's human competence cheap.","startTime":40.72,"endTime":44.44,"durationSeconds":4,"level":"B2","overallScore":8.6,"rationale":"일반화된 통찰이 매우 강함."},{"segmentIndex":41,"text":"is the minute the agent is like not really that useful anymore. And that's why it I think it has started to shift to a more uh one agent per company model because for now like the ideal is uh you basically set up a forward deployed engineer or someone with that sort of profile who's responsible for making sure that agent is working for the whole company.","startTime":907.68,"endTime":931.36,"durationSeconds":24,"level":"C1","overallScore":8.8,"rationale":"회사당 에이전트 모델의 이유를 설명."},{"segmentIndex":29,"text":"And I think the reverse is actually starting to happen and be like really, really valuable in a way that I did not expect, which is take the AI agent that you use all the time on your computer and put a browser in it so it can see everything you're doing.","startTime":1397.4,"endTime":1410.64,"durationSeconds":13,"level":"C1","overallScore":9,"rationale":"핵심 전환점을 명료하게 설명."},{"segmentIndex":73,"text":"Automation is a lie. Um in the sense that every time you automate something, in order to make sure the automation is working well, you need a human on top of it like making sure that it's working well.","startTime":2355.12,"endTime":2366.64,"durationSeconds":12,"level":"B2","overallScore":8.8,"rationale":"자동화 역설을 명확히 말해 통찰과 표현 강함."},{"segmentIndex":47,"text":"And so I think it's really important uh when we think about benchmark progress to think about it from that perspective, which is benchmarks rise on problems that we've framed that we can articulate, that we can score.","startTime":2711.12,"endTime":2745.16,"durationSeconds":34,"level":"C1","overallScore":8.8,"rationale":"벤치마크 해석 원칙을 정리한다."},{"segmentIndex":69,"text":"And I think that's a really important thing when we think about benchmarks is AI is a broadly distributed technology that any human can use and when we are benchmarking against humans, AI against humans, we're actually really always talking about one human using AI versus another human using AI cuz AI doesn't use itself.","startTime":2887.24,"endTime":2908.12,"durationSeconds":21,"level":"C1","overallScore":9,"rationale":"AI 대 인간 비교의 핵심 재정의."},{"segmentIndex":30,"text":"The pattern that I see so far and again, I don't have a total crystal ball, but I do feel like we've seen enough of the new model drops to like have some sense of how this is going is that what a new model drop does or what models do in general is they make yesterday's human competence cheap.","startTime":4418.6,"endTime":4436.48,"durationSeconds":18,"level":"C1","overallScore":8.8,"rationale":"모델의 경제적 효과를 날카롭게 정의."},{"segmentIndex":39,"text":"And I really think that structurally, because of the way the models work, because of the financial incentives of model companies to like make model of model companies to like make them um uh compliant and aligned, structurally, there are always going to be trailing behind those people who are taking the models and using them to make new expertise or make new things that haven't been done that way before for their very particular situation.","startTime":4502.2,"endTime":4527.72,"durationSeconds":26,"level":"C2","overallScore":8.8,"rationale":"구조적 이유로 인간 우위가 남는단 통찰."},{"segmentIndex":62,"text":"Um so, the way to ride the models is like not one specific thing cuz they're always changing, but it is to be curious and playful, to apply the model the new model to whatever it is that you care about, whether that's your job or something outside of your job and to keep turning over rocks uh because it may not work now, but it may work eventually, it probably will work eventually and the way that you use it matters.","startTime":4680.52,"endTime":4717.6,"durationSeconds":37,"level":"C1","overallScore":9,"rationale":"핵심 조언과 학습 표현이 매우 풍부함."},{"segmentIndex":18,"text":"And I think generally our intuitions about the future the model that I have of what our intuitions are about the future is the intuitions that people had in the Middle Ages about like what happened at the end of the horizon, you know, it's like are there dragons?","startTime":4914.6,"endTime":4931.12,"durationSeconds":17,"level":"C1","overallScore":8.8,"rationale":"미래 인식을 강한 비유로 통찰적으로 설명함."},{"segmentIndex":23,"text":"And until you get to that place where you're starting to see it, and I think we get to see it cuz we get to see it internally all the time, it's important not to let your mind get away from you and being like, this is going to happen and this like, this is going to happen and whatever cuz you're going to tell a story that sounds so real in the moment, but um later on you're like, actually it's much more complex than that and somewhere it's sort of a both everything has changed and nothing has.","startTime":4958.8,"endTime":4986.76,"durationSeconds":28,"level":"C1","overallScore":8.8,"rationale":"미래 예측의 함정을 깊이 있게 경고함."},{"segmentIndex":32,"text":"And try to like try to have fun. I think there's too much of I'm doing this because I have FOMO like it might I might lose my job or like I might miss out on this big thing or whatever and the best way to actually figure out interesting useful things to do with AI is to like do something enjoyable.","startTime":5052.28,"endTime":5071.32,"durationSeconds":19,"level":"C1","overallScore":9,"rationale":"FOMO보다 즐거움이 중요하단 조언이 강함."},{"segmentIndex":3,"text":">> The AI job apocalypse is not really a thing. I am super bullish on PMs and full-stack designers.","startTime":10.48,"endTime":18.28,"durationSeconds":8,"level":"B2","overallScore":7.8,"rationale":"강한 주장과 유용한 구어 표현."},{"segmentIndex":9,"text":"And so, it becomes commoditized. It's not valuable anymore. What humans do is we go in there and we're like,\"Yeah, we have all this frozen human competence from yesterday.","startTime":44.44,"endTime":52.76,"durationSeconds":8,"level":"C1","overallScore":7.8,"rationale":"개념 설명이 선명하고 표현도 좋음."},{"segmentIndex":11,"text":">> It's going to bifurcate in two main ways. One is everyone's going to have at least one agent that they talk to, that they can offload work to.","startTime":57.32,"endTime":64.879,"durationSeconds":8,"level":"B2","overallScore":8,"rationale":"예측과 표현 가치가 모두 높음."},{"segmentIndex":15,"text":"What agents do is increase the number of users of SaaS, not get rid of it.","startTime":80.52,"endTime":84.16,"durationSeconds":4,"level":"B2","overallScore":7.6,"rationale":"핵심 반박 논리가 분명함."},{"segmentIndex":42,"text":"So, just give us a sense of how you work. Thank you. Um I really appreciate the introduction. Um and yeah, I think what one of the things about predicting the future or the way that we think about predicting the future at Every is that you what you don't want to do is prognosticate.","startTime":266.52,"endTime":271.16,"durationSeconds":5,"level":"C1","overallScore":8,"rationale":"미래 예측 태도에 대한 통찰이 큼."},{"segmentIndex":50,"text":"And what I what that does, I think, is it creates this like little pocket of the future where we're all living in it together and we get to be a little bit further ahead cuz at any other company there's like a mix of people.","startTime":320.24,"endTime":330.44,"durationSeconds":10,"level":"C1","overallScore":8,"rationale":"조직이 미래를 앞서가는 이유 설명."}],"generatedAt":"2026-06-25T00:45:01.565Z","keyClipsTotalSec":1550},{"videoId":"4D3hDmGhFhA","chunkIndex":9,"totalChunks":10,"title":"The AI paradox: More automation, more humans, more work | Dan Shipper — Part 10 of 10","thumbnail":"https://i.ytimg.com/vi/4D3hDmGhFhA/maxresdefault.jpg","duration":5646,"uploader":"Lenny's Podcast","youtubeUrl":"https://www.youtube.com/watch?v=4D3hDmGhFhA","keywords":["artificial-intelligence","automation","knowledge-work","productivity","career","tools","software","workflow","future-of-work"],"normalizedKeywords":["기술 트렌드","커리어·성장","비즈니스·전략"],"targetAudience":[{"who":"지식노동자","why":"AI가 일과 커리어를 어떻게 바꾸는지 현실적으로 점검할 수 있음"},{"who":"창업자","why":"업무 도구 선택과 AI 활용 습관이 생산성에 미치는 영향을 볼 수 있음"},{"who":"개발자","why":"Codex와 Claude 같은 코딩/업무 도구 선택 관점을 참고할 수 있음"}],"normalizedAudience":["지식노동자 일반","창업자·스타트업","엔지니어·개발자"],"summary":"이 구간은 AI 시대에 불안과 업무 도구 선택을 어떻게 다뤄야 하는지에 대한 대화다. 화자는 AI가 내 일을 바꿀까 두렵다면, 그 두려움에 압도되기보다 '여유와 힘의 관점'에서 바라보는 것이 훨씬 생산적이라고 말한다. 이어서 최근 가장 과소평가된 도구로 Codex를 추천하며, 특히 브라우저 작업까지 포함한 실제 업무 흐름에서 작업 방식을 크게 바꿔준다고 설명한다.\n\n또한 Claude와 OpenAI를 한쪽만 고집하기보다 상황에 따라 둘 다 쓰는 것이 유리하다고 보고, AI 도구 시장은 한 회사가 영구히 지배하는 구조가 아니라 앞서거니 뒤서거니 하는 경쟁 구도라고 정리한다. 마지막에는 중요한 건 도구를 함께 실험하고 생활 속에서 잘 활용해보는 태도라고 강조한다.","insights":["AI 불안은 감정이 아니라 작업 방식의 문제로 다뤄야 한다.","여유와 힘의 관점이 있어야 문제 해결이 더 빨라진다.","도구는 하나를 믿기보다 상황별로 갈아타는 게 유리하다.","좋은 AI 활용은 혼자 소유보다 함께 실험할 때 빨라진다."],"keyClips":[{"clipId":"4D3hDmGhFhA:c9:1-8","startSegmentIndex":1,"endSegmentIndex":8,"startTime":5400.87,"endTime":5465.84,"durationSeconds":65,"preview":"불안은 관점이 좌우","mustSee":false},{"clipId":"4D3hDmGhFhA:c9:9-13","startSegmentIndex":9,"endSegmentIndex":13,"startTime":5465.84,"endTime":5519.8,"durationSeconds":54,"preview":"과소평가된 코덱스","mustSee":false},{"clipId":"4D3hDmGhFhA:c9:14-21","startSegmentIndex":14,"endSegmentIndex":21,"startTime":5519.8,"endTime":5571.6,"durationSeconds":51.8,"preview":"도구는 혼용이 답","mustSee":true},{"clipId":"4D3hDmGhFhA:c9:24-28","startSegmentIndex":24,"endSegmentIndex":28,"startTime":5581.52,"endTime":5620.08,"durationSeconds":38.6,"preview":"함께 써야 빨라진다","mustSee":false}],"curatedSegments":[{"segmentIndex":45,"text":"And there's a lot of work that's human work that uh it can't be scored until you write it down, but the act of thinking to prompt it or write it down um is uh is something that you can't measure, but like kind of means that even if the benchmarks get saturated, it doesn't mean the same thing as we you totally replace all senior engineers.","startTime":2703.56,"endTime":2707,"durationSeconds":3,"level":"C2","overallScore":9.8,"rationale":"측정 불가한 인간 일의 본질을 짚음."},{"segmentIndex":6,"text":"Automation is a lie. Every agent needs a human. We have so much automation, so much AI, and I also work way more. >> Creativity.","startTime":29,"endTime":35.2,"durationSeconds":6,"level":"B2","overallScore":8.4,"rationale":"핵심 주장과 대비가 매우 선명함."},{"segmentIndex":8,"text":">> What models do in general is they make yesterday's human competence cheap.","startTime":40.72,"endTime":44.44,"durationSeconds":4,"level":"B2","overallScore":8.6,"rationale":"일반화된 통찰이 매우 강함."},{"segmentIndex":41,"text":"is the minute the agent is like not really that useful anymore. And that's why it I think it has started to shift to a more uh one agent per company model because for now like the ideal is uh you basically set up a forward deployed engineer or someone with that sort of profile who's responsible for making sure that agent is working for the whole company.","startTime":907.68,"endTime":931.36,"durationSeconds":24,"level":"C1","overallScore":8.8,"rationale":"회사당 에이전트 모델의 이유를 설명."},{"segmentIndex":29,"text":"And I think the reverse is actually starting to happen and be like really, really valuable in a way that I did not expect, which is take the AI agent that you use all the time on your computer and put a browser in it so it can see everything you're doing.","startTime":1397.4,"endTime":1410.64,"durationSeconds":13,"level":"C1","overallScore":9,"rationale":"핵심 전환점을 명료하게 설명."},{"segmentIndex":73,"text":"Automation is a lie. Um in the sense that every time you automate something, in order to make sure the automation is working well, you need a human on top of it like making sure that it's working well.","startTime":2355.12,"endTime":2366.64,"durationSeconds":12,"level":"B2","overallScore":8.8,"rationale":"자동화 역설을 명확히 말해 통찰과 표현 강함."},{"segmentIndex":47,"text":"And so I think it's really important uh when we think about benchmark progress to think about it from that perspective, which is benchmarks rise on problems that we've framed that we can articulate, that we can score.","startTime":2711.12,"endTime":2745.16,"durationSeconds":34,"level":"C1","overallScore":8.8,"rationale":"벤치마크 해석 원칙을 정리한다."},{"segmentIndex":69,"text":"And I think that's a really important thing when we think about benchmarks is AI is a broadly distributed technology that any human can use and when we are benchmarking against humans, AI against humans, we're actually really always talking about one human using AI versus another human using AI cuz AI doesn't use itself.","startTime":2887.24,"endTime":2908.12,"durationSeconds":21,"level":"C1","overallScore":9,"rationale":"AI 대 인간 비교의 핵심 재정의."},{"segmentIndex":30,"text":"The pattern that I see so far and again, I don't have a total crystal ball, but I do feel like we've seen enough of the new model drops to like have some sense of how this is going is that what a new model drop does or what models do in general is they make yesterday's human competence cheap.","startTime":4418.6,"endTime":4436.48,"durationSeconds":18,"level":"C1","overallScore":8.8,"rationale":"모델의 경제적 효과를 날카롭게 정의."},{"segmentIndex":39,"text":"And I really think that structurally, because of the way the models work, because of the financial incentives of model companies to like make model of model companies to like make them um uh compliant and aligned, structurally, there are always going to be trailing behind those people who are taking the models and using them to make new expertise or make new things that haven't been done that way before for their very particular situation.","startTime":4502.2,"endTime":4527.72,"durationSeconds":26,"level":"C2","overallScore":8.8,"rationale":"구조적 이유로 인간 우위가 남는단 통찰."},{"segmentIndex":62,"text":"Um so, the way to ride the models is like not one specific thing cuz they're always changing, but it is to be curious and playful, to apply the model the new model to whatever it is that you care about, whether that's your job or something outside of your job and to keep turning over rocks uh because it may not work now, but it may work eventually, it probably will work eventually and the way that you use it matters.","startTime":4680.52,"endTime":4717.6,"durationSeconds":37,"level":"C1","overallScore":9,"rationale":"핵심 조언과 학습 표현이 매우 풍부함."},{"segmentIndex":18,"text":"And I think generally our intuitions about the future the model that I have of what our intuitions are about the future is the intuitions that people had in the Middle Ages about like what happened at the end of the horizon, you know, it's like are there dragons?","startTime":4914.6,"endTime":4931.12,"durationSeconds":17,"level":"C1","overallScore":8.8,"rationale":"미래 인식을 강한 비유로 통찰적으로 설명함."},{"segmentIndex":23,"text":"And until you get to that place where you're starting to see it, and I think we get to see it cuz we get to see it internally all the time, it's important not to let your mind get away from you and being like, this is going to happen and this like, this is going to happen and whatever cuz you're going to tell a story that sounds so real in the moment, but um later on you're like, actually it's much more complex than that and somewhere it's sort of a both everything has changed and nothing has.","startTime":4958.8,"endTime":4986.76,"durationSeconds":28,"level":"C1","overallScore":8.8,"rationale":"미래 예측의 함정을 깊이 있게 경고함."},{"segmentIndex":32,"text":"And try to like try to have fun. I think there's too much of I'm doing this because I have FOMO like it might I might lose my job or like I might miss out on this big thing or whatever and the best way to actually figure out interesting useful things to do with AI is to like do something enjoyable.","startTime":5052.28,"endTime":5071.32,"durationSeconds":19,"level":"C1","overallScore":9,"rationale":"FOMO보다 즐거움이 중요하단 조언이 강함."},{"segmentIndex":3,"text":">> The AI job apocalypse is not really a thing. I am super bullish on PMs and full-stack designers.","startTime":10.48,"endTime":18.28,"durationSeconds":8,"level":"B2","overallScore":7.8,"rationale":"강한 주장과 유용한 구어 표현."},{"segmentIndex":9,"text":"And so, it becomes commoditized. It's not valuable anymore. What humans do is we go in there and we're like,\"Yeah, we have all this frozen human competence from yesterday.","startTime":44.44,"endTime":52.76,"durationSeconds":8,"level":"C1","overallScore":7.8,"rationale":"개념 설명이 선명하고 표현도 좋음."},{"segmentIndex":11,"text":">> It's going to bifurcate in two main ways. One is everyone's going to have at least one agent that they talk to, that they can offload work to.","startTime":57.32,"endTime":64.879,"durationSeconds":8,"level":"B2","overallScore":8,"rationale":"예측과 표현 가치가 모두 높음."},{"segmentIndex":15,"text":"What agents do is increase the number of users of SaaS, not get rid of it.","startTime":80.52,"endTime":84.16,"durationSeconds":4,"level":"B2","overallScore":7.6,"rationale":"핵심 반박 논리가 분명함."},{"segmentIndex":42,"text":"So, just give us a sense of how you work. Thank you. Um I really appreciate the introduction. Um and yeah, I think what one of the things about predicting the future or the way that we think about predicting the future at Every is that you what you don't want to do is prognosticate.","startTime":266.52,"endTime":271.16,"durationSeconds":5,"level":"C1","overallScore":8,"rationale":"미래 예측 태도에 대한 통찰이 큼."},{"segmentIndex":50,"text":"And what I what that does, I think, is it creates this like little pocket of the future where we're all living in it together and we get to be a little bit further ahead cuz at any other company there's like a mix of people.","startTime":320.24,"endTime":330.44,"durationSeconds":10,"level":"C1","overallScore":8,"rationale":"조직이 미래를 앞서가는 이유 설명."}],"generatedAt":"2026-06-25T00:45:24.276Z","keyClipsTotalSec":1550}]}