{"success":true,"count":5,"items":[{"videoId":"B246K_G7mHU","chunkIndex":0,"totalChunks":5,"title":"Inside YC's AI Playbook — Part 1 of 5","thumbnail":"https://i.ytimg.com/vi/B246K_G7mHU/maxresdefault.jpg","duration":2790,"uploader":"Y Combinator","youtubeUrl":"https://www.youtube.com/watch?v=B246K_G7mHU","keywords":["ai","agents","startup","internal-tools","software-engineering","automation","data-access","productivity","business-operations"],"normalizedKeywords":["리더십·매니지먼트","엔지니어링","기술 트렌드"],"targetAudience":[{"who":"창업자","why":"회사 내부 업무를 AI로 재설계하는 조직 전략을 배울 수 있음"},{"who":"엔지니어","why":"에이전트 도구·DB 접근·사내 하네스 설계 감각을 얻을 수 있음"},{"who":"프로덕트 담당자","why":"반복적인 업무 루프를 제품화하는 접근을 참고할 수 있음"},{"who":"지식노동자","why":"질문 비용을 낮춰 더 자주 더 큰 질문을 하는 원리를 이해할 수 있음"}],"normalizedAudience":["창업자·스타트업","엔지니어·개발자","지식노동자 일반"],"summary":"이 영상은 YC가 내부 운영을 AI 네이티브하게 바꾸기 위해 어떤 과정을 거쳐 에이전트 인프라를 만들었는지 설명한다. 출발점은 재무팀과 엔지니어 사이의 비효율적인 왕복 루프였고, 이를 줄이기 위해 자연어 프롬프트로 업무를 표현하고 에이전트가 실제 데이터를 조회·실행하게 하는 구조로 전환했다. 특히 읽기 전용 SQL 쿼리와 모델 파일 접근 같은 제한적이지만 강력한 도구를 붙이자, 비기술자도 복잡한 질문을 직접 던질 수 있게 되었고 질문의 양과 수준 자체가 크게 올라갔다.\n\n핵심 메시지는 AI를 단순한 코파일럿이 아니라 조직 전체의 '공유 뇌'로 써야 한다는 것이다. YC처럼 중요한 맥락이 한 데이터베이스에 모여 있고, 업무 지식이 코드가 아니라 문장으로 운영될 때 에이전트는 엄청난 생산성 증폭기가 된다. 동시에 저자는 기존 회사들이 여전히 오래된 BI·백로그 구조에 묶여 있어 질문 비용이 높고, 그게 곧 조직의 속도와 사고의 범위를 제한한다고 본다.","insights":["AI는 보조 도구가 아니라 조직의 기본 운영층이 되어야 한다.","질문 비용이 낮아질수록 조직이 다루는 문제의 크기가 커진다.","맥락이 한곳에 모여 있어야 에이전트가 진짜로 유용해진다.","제한된 권한에 집착하면 AI의 생산성 잠재력을 놓치기 쉽다.","업무를 코드보다 자연어로 표현할수록 비기술자의 자율성이 커진다."],"keyClips":[{"clipId":"B246K_G7mHU:c0:1-4","startSegmentIndex":1,"endSegmentIndex":4,"startTime":1.59,"endTime":41.99,"durationSeconds":40.4,"preview":"AI는 코파일럿이 아니다","mustSee":true},{"clipId":"B246K_G7mHU:c0:15-28","startSegmentIndex":15,"endSegmentIndex":28,"startTime":136.56,"endTime":318.72,"durationSeconds":182.2,"preview":"재무팀에서 시작한 실험","mustSee":true},{"clipId":"B246K_G7mHU:c0:31-38","startSegmentIndex":31,"endSegmentIndex":38,"startTime":344.88,"endTime":455.039,"durationSeconds":110.2,"preview":"도구 권한의 역설","mustSee":true},{"clipId":"B246K_G7mHU:c0:39-49","startSegmentIndex":39,"endSegmentIndex":49,"startTime":455.039,"endTime":576.48,"durationSeconds":121.4,"preview":"질문이 많아지는 구조","mustSee":false},{"clipId":"B246K_G7mHU:c0:50-54","startSegmentIndex":50,"endSegmentIndex":54,"startTime":576.48,"endTime":607.279,"durationSeconds":30.8,"preview":"구식 조직의 한계","mustSee":false}],"curatedSegments":[{"segmentIndex":4,"text":"If you frame this as a way for everyone in an organization to get better at what they do using the like collective skill and instinct of the people they work with, it's incredibly powerful. Today we have a real treat.","startTime":19.6,"endTime":41.99,"durationSeconds":22,"level":"C1","overallScore":9,"rationale":"조직 학습 통찰과 구조 표현 풍부."},{"segmentIndex":28,"text":"like remove the software engineers from this crazy loop where they have to sort of understand these complicated workflows and give the finance team the tools that they could use to encode their own workflows not as you know not as Ruby uh but as English with prompts right >> I mean what's interesting is like uh we all funded companies like maybe even like two or three years ago when LLMs were out but like agent coding wasn't a thing yet and so the first thing actually was not a coding it was LLMs for writing SQL queries.","startTime":288.56,"endTime":318.72,"durationSeconds":30,"level":"C1","overallScore":9.2,"rationale":"핵심 전환 논리와 표현 밀도 매우 높음."},{"segmentIndex":31,"text":"I was really surprised too to be honest and so that we started with this kind of purpose-built thing for finance and then rewrote it to even more of a general agent loop right and it and it's this is now you see these all over the place now but um I the first kind of magical moment that I had was we had this agent loop and we had a tool registry a shared tool registry for kind of YC specific tools and the first tool that really was an unlock for me was I think a tool looking back that you actually built Jared.","startTime":344.88,"endTime":376.08,"durationSeconds":31,"level":"C1","overallScore":9.2,"rationale":"마법 같은 전환점 서사와 표현 풍부."},{"segmentIndex":45,"text":"It just turns out when all of that context is in one place with a little bit of additional uh information about how the schema is laid out, an agent can go and ask any or answer arbitrary questions about our business.","startTime":509.52,"endTime":521.279,"durationSeconds":12,"level":"C1","overallScore":9,"rationale":"맥락 집약의 힘을 잘 설명한 문장."},{"segmentIndex":48,"text":"It dramatically increased the number of questions that we would ask and dramatically increased the scale and complexity of the questions that we would dare to ask where like you know in the old days back when we were the in the old days back when we were using like BI tools to ask a question like that you know like what investors have invested like in space related companies that would be like several hours of writing SQL and so like unless it was really important you just wouldn't bother.","startTime":528.56,"endTime":550,"durationSeconds":21,"level":"C1","overallScore":9.2,"rationale":"행동 변화 통찰과 자연스런 표현 풍부."},{"segmentIndex":49,"text":"It's just another example of the you know this instance of Gvon's paradox that you get when you remove the amount of back and forth uh between different teams in order to get a thing done right if in order to answer ask some kind of complex question about YC I have to go and knock on you know the data science team's door and wait for them to get it through you know their backlog uh I'm just going to ask far fewer questions.","startTime":550,"endTime":576.48,"durationSeconds":26,"level":"C1","overallScore":9.2,"rationale":"마찰 감소의 원리를 설득력 있게 설명."},{"segmentIndex":7,"text":"Interestingly, uh I now realize YC itself is a context engineering uh sort of process in that like people we're frequently teaching people you have perfect context about what's going on in your brain, but great communication is replicating that same context in someone else's brain.","startTime":1245.2,"endTime":1264.32,"durationSeconds":19,"level":"C1","overallScore":8.8,"rationale":"강한 통찰과 추상적 설명이 공존."},{"segmentIndex":16,"text":"and now lived in the context of that meeting transcript and handing that back to the agent and saying given you know what you've learned by reading through this context improve the two sentence description skill and they got noticeably better after that like this thing is now better than I am I would I thing is now better than I am I would argue at writing those >> this is how super intelligence happens inside organizations I mean this two sentence pitch thing sounds like something kind of small but uh embedded in it is actually something very powerful I'm sure you guys have heard um Jack Dorsey talk about what he's doing with block.","startTime":1363.6,"endTime":1397.52,"durationSeconds":34,"level":"C1","overallScore":8.8,"rationale":"조직 학습 메커니즘 통찰이 강함."},{"segmentIndex":19,"text":"like you can look at the operation of any organization as uh the aggregate of you know I mean the two sentence pitch at YC is that's sort of one of like thousands of things that I would argue we do for founders but you know we just walked through a very concrete way where someone wrote a prompt used it used a bunch more other people used it uh a bunch of artifacts came off of that around literally like the transcript of using it becomes a thing that can be used to metaprompt and improve in an automated fashion on a daily basis the operation of that one skill and then suddenly that one skill you just said it that skill is now better than any of us individually than be you know when before we actually had access to that and so this is like a particular like needle pin prick in the fabric of like how any organization does things and then how do you build super intelligence inside a company you do that on everything you do.","startTime":1412.96,"endTime":1472.799,"durationSeconds":60,"level":"C1","overallScore":8.8,"rationale":"구체 예시로 일반 원리까지 확장함."},{"segmentIndex":24,"text":"But the real idea underneath was just kind of that the potential for AI is to shift control of software from the developer to the user, right?","startTime":1983.76,"endTime":2012.96,"durationSeconds":29,"level":"C1","overallScore":8.8,"rationale":"핵심 통찰이 매우 선명하고 구조도 좋음."},{"segmentIndex":27,"text":"I think the conclusion that this essay points to is that as we get better at building AI native software, it's going to look a lot more like the agent wrapping software deterministic tools rather than deterministic software wrapping AI, right?","startTime":2034.48,"endTime":2053.04,"durationSeconds":19,"level":"C1","overallScore":9,"rationale":"추상적 통찰과 고급 구조가 매우 풍부함."},{"segmentIndex":54,"text":"like they were complaining about like I had made you know this monolithic writer chat interface and it was like full of bugs because I was like reimplementing things that openclaw and telegram already do and now they just use open andclaw telegram and my retrieval system with like all the same data that I extracted it out and with our MCP and it's working great like basically you know Gary's list 2.0 know the next rewrite thankfully is not half a million lines of uh Rails code that is like insane to actually you it's rigid it's uh takes a long time it like takes like 10 times long you know even though it was 1/100th the amount of time to do it like by hand you don't have to do it by hand like that half a million lines of code in Rails is easily like 10,000 lines of like TypeScript and like maybe 2,000 lines of markdown and all of that is way more dynamic like you could just say like actually for the second paragraph uh I really like including a biography of like the politician we're focusing on and it's like I don't have to code that in Rails.","startTime":2239.44,"endTime":2302.96,"durationSeconds":64,"level":"C1","overallScore":9,"rationale":"구식 개발과 신방식 대비가 매우 선명함."},{"segmentIndex":57,"text":"Like this is actually the dawn of just in time software and I can see it right now.","startTime":2321.599,"endTime":2327.119,"durationSeconds":6,"level":"B2","overallScore":8.6,"rationale":"핵심 개념 선언이라 통찰 밀도가 높음."},{"segmentIndex":35,"text":"And that's what I hope that we are building with things like Gbrain and you know Hermes agent and openclaw like the ability to run your own software to change your own prompts to test all of it to have your own private repo that like you know is only yours um to be able to choose which model to use and maybe it's an openweight model like to me that's sort of the white pill for AI is uh we could have corporate control, no control of your own prompts and like literally the AI happens to you.","startTime":2635.44,"endTime":2669.92,"durationSeconds":34,"level":"C1","overallScore":9,"rationale":"핵심 비전과 표현 자원이 매우 풍부함."},{"segmentIndex":1,"text":"How do you build super intelligence inside a company? >> Part of the key thing is not to just use AI as a copilot.","startTime":1.59,"endTime":7.12,"durationSeconds":6,"level":"B2","overallScore":7.8,"rationale":"핵심 원칙 제시와 유용한 표현 포함."},{"segmentIndex":3,"text":"And you need to start recording all the artifacts. >> It's like a shared organizational brain. It's like the closest thing to us being able to like connect our brains.","startTime":11.679,"endTime":19.6,"durationSeconds":8,"level":"B2","overallScore":8,"rationale":"비유가 강하고 학습 표현도 좋음."},{"segmentIndex":11,"text":"And it's been I think a very powerful symbiotic relationship for us to actually be adopting these tools and like transforming our own organization which was started way pre- AI into a super AI native organization ourselves and Pete has really been leading the charge for that.","startTime":98.15899999999999,"endTime":115.2,"durationSeconds":17,"level":"C1","overallScore":8,"rationale":"조직 변화 통찰과 고급 표현 풍부."},{"segmentIndex":21,"text":"The finance team would describe to our software engineers how you know this complicated financial workflow worked and then the software engineers would go and build some purpose-built software where there was a deterministic workflow encapsulating everything that they had been told and then hand it back to the finance team and so on.","startTime":226.239,"endTime":244.879,"durationSeconds":19,"level":"C1","overallScore":8.2,"rationale":"비효율 프로세스 설명이 통찰적임."},{"segmentIndex":27,"text":"And so the original impetus was why don't we try to build some tools at YC that we could use to run agents that would give the finance team control over their own software, right?","startTime":276.72,"endTime":288.56,"durationSeconds":12,"level":"B2","overallScore":8,"rationale":"핵심 아이디어와 실용적 구조가 강함."},{"segmentIndex":34,"text":"I remember I built those tools and I felt a little bit like I was breaking the rules because initially we started with very limited tools that had very narrowly scoped domains and I kept getting frustrated because they weren't powerful enough to do the things that I wanted and so I was like what if we just gave the thing like access complete access to the production database where it could just like trample on anything and I sort of like surreptitiously pushed it out maybe late at Right.","startTime":389.6,"endTime":419.759,"durationSeconds":30,"level":"C1","overallScore":8.4,"rationale":"실험 서사와 고급 구어 표현이 많음."}],"generatedAt":"2026-06-25T00:28:38.939Z","keyClipsTotalSec":1574},{"videoId":"B246K_G7mHU","chunkIndex":1,"totalChunks":5,"title":"Inside YC's AI Playbook — Part 2 of 5","thumbnail":"https://i.ytimg.com/vi/B246K_G7mHU/maxresdefault.jpg","duration":2790,"uploader":"Y Combinator","youtubeUrl":"https://www.youtube.com/watch?v=B246K_G7mHU","keywords":["ai-agents","llm","enterprise-ai","knowledge-management","retrieval","tooling","mcp","automation","startup"],"normalizedKeywords":["엔지니어링","비즈니스·전략","기술 트렌드"],"targetAudience":[{"who":"AI 제품/플랫폼 담당자","why":"에이전트용 컨텍스트, 툴 레지스트리, 리졸버 설계가 핵심 논점이다."},{"who":"사내 자동화 담당자","why":"조직 전체에 AI를 확산시키는 인프라와 운영 방식이 구체적으로 나온다."},{"who":"엔지니어링 리더","why":"멀티플레이어 에이전트와 내부 도구 체계를 어떻게 묶을지 배울 수 있다."},{"who":"예비 창업자","why":"에이전트 인프라에서 반복적으로 등장하는 핵심 프리미티브를 이해할 수 있다."}],"normalizedAudience":["창업자·스타트업","엔지니어·개발자","지식노동자 일반"],"summary":"이 영상은 YC가 내부에서 AI 에이전트를 실제 업무에 적용하면서 발견한 핵심 원리를 정리한다. 화자는 에이전트가 사람처럼 유용해지려면 데이터를 여기저기 흩어두기보다 하나의 큰 컨텍스트 층으로 정규화/비정규화해 에이전트가 읽기 쉬운 형태로 만들어야 한다고 강조한다. 또 지금의 에이전트 생태계는 아직 '싱글 플레이어' 단계에 가깝고, 진짜 어려운 문제는 팀과 조직 전체에서 작동하는 멀티플레이어 하니스라는 점을 짚는다.\n\n이어 내부 툴 레지스트리, 스킬/리졸버 구조, 메타 스킬, 자기개선 루프(드림 사이클) 같은 공통 프리미티브를 통해 에이전트 시스템이 점점 표준화되고 있음을 설명한다. 결국 이 영상은 '에이전트 시대의 Unix 프리미티브'가 무엇인지 YC 사례로 보여주며, 조직이 AI를 도입할 때 어디에 투자해야 하는지 실무적으로 제안한다.","insights":["에이전트는 흩어진 데이터를 먹기보다 큰 테이블로 다뤄야 강해진다.","개인용 에이전트는 성숙했지만 팀용 하니스는 아직 미해결이다.","조직 AI의 핵심은 모델보다 컨텍스트 층과 툴 레지스트리다.","툴이 많아질수록 리졸버와 중복 제거가 시스템 품질을 좌우한다.","에이전트는 스스로 대화 로그를 학습하는 자기개선 루프로 진화한다."],"keyClips":[{"clipId":"B246K_G7mHU:c1:1-11","startSegmentIndex":1,"endSegmentIndex":11,"startTime":601.75,"endTime":732.24,"durationSeconds":130.5,"preview":"에이전트용 큰 테이블","mustSee":true},{"clipId":"B246K_G7mHU:c1:12-20","startSegmentIndex":12,"endSegmentIndex":20,"startTime":732.24,"endTime":844.56,"durationSeconds":112.3,"preview":"싱글플레이어의 한계","mustSee":true},{"clipId":"B246K_G7mHU:c1:21-33","startSegmentIndex":21,"endSegmentIndex":33,"startTime":844.56,"endTime":926.079,"durationSeconds":81.5,"preview":"툴 레지스트리의 힘","mustSee":true},{"clipId":"B246K_G7mHU:c1:34-46","startSegmentIndex":34,"endSegmentIndex":46,"startTime":926.079,"endTime":1047.36,"durationSeconds":121.3,"preview":"스킬과 리졸버 정리","mustSee":false},{"clipId":"B246K_G7mHU:c1:47-62","startSegmentIndex":47,"endSegmentIndex":62,"startTime":1047.36,"endTime":1195.12,"durationSeconds":147.8,"preview":"에이전트의 자가학습","mustSee":true}],"curatedSegments":[{"segmentIndex":4,"text":"If you frame this as a way for everyone in an organization to get better at what they do using the like collective skill and instinct of the people they work with, it's incredibly powerful. Today we have a real treat.","startTime":19.6,"endTime":41.99,"durationSeconds":22,"level":"C1","overallScore":9,"rationale":"조직 학습 통찰과 구조 표현 풍부."},{"segmentIndex":28,"text":"like remove the software engineers from this crazy loop where they have to sort of understand these complicated workflows and give the finance team the tools that they could use to encode their own workflows not as you know not as Ruby uh but as English with prompts right >> I mean what's interesting is like uh we all funded companies like maybe even like two or three years ago when LLMs were out but like agent coding wasn't a thing yet and so the first thing actually was not a coding it was LLMs for writing SQL queries.","startTime":288.56,"endTime":318.72,"durationSeconds":30,"level":"C1","overallScore":9.2,"rationale":"핵심 전환 논리와 표현 밀도 매우 높음."},{"segmentIndex":31,"text":"I was really surprised too to be honest and so that we started with this kind of purpose-built thing for finance and then rewrote it to even more of a general agent loop right and it and it's this is now you see these all over the place now but um I the first kind of magical moment that I had was we had this agent loop and we had a tool registry a shared tool registry for kind of YC specific tools and the first tool that really was an unlock for me was I think a tool looking back that you actually built Jared.","startTime":344.88,"endTime":376.08,"durationSeconds":31,"level":"C1","overallScore":9.2,"rationale":"마법 같은 전환점 서사와 표현 풍부."},{"segmentIndex":45,"text":"It just turns out when all of that context is in one place with a little bit of additional uh information about how the schema is laid out, an agent can go and ask any or answer arbitrary questions about our business.","startTime":509.52,"endTime":521.279,"durationSeconds":12,"level":"C1","overallScore":9,"rationale":"맥락 집약의 힘을 잘 설명한 문장."},{"segmentIndex":48,"text":"It dramatically increased the number of questions that we would ask and dramatically increased the scale and complexity of the questions that we would dare to ask where like you know in the old days back when we were the in the old days back when we were using like BI tools to ask a question like that you know like what investors have invested like in space related companies that would be like several hours of writing SQL and so like unless it was really important you just wouldn't bother.","startTime":528.56,"endTime":550,"durationSeconds":21,"level":"C1","overallScore":9.2,"rationale":"행동 변화 통찰과 자연스런 표현 풍부."},{"segmentIndex":49,"text":"It's just another example of the you know this instance of Gvon's paradox that you get when you remove the amount of back and forth uh between different teams in order to get a thing done right if in order to answer ask some kind of complex question about YC I have to go and knock on you know the data science team's door and wait for them to get it through you know their backlog uh I'm just going to ask far fewer questions.","startTime":550,"endTime":576.48,"durationSeconds":26,"level":"C1","overallScore":9.2,"rationale":"마찰 감소의 원리를 설득력 있게 설명."},{"segmentIndex":7,"text":"Interestingly, uh I now realize YC itself is a context engineering uh sort of process in that like people we're frequently teaching people you have perfect context about what's going on in your brain, but great communication is replicating that same context in someone else's brain.","startTime":1245.2,"endTime":1264.32,"durationSeconds":19,"level":"C1","overallScore":8.8,"rationale":"강한 통찰과 추상적 설명이 공존."},{"segmentIndex":16,"text":"and now lived in the context of that meeting transcript and handing that back to the agent and saying given you know what you've learned by reading through this context improve the two sentence description skill and they got noticeably better after that like this thing is now better than I am I would I thing is now better than I am I would argue at writing those >> this is how super intelligence happens inside organizations I mean this two sentence pitch thing sounds like something kind of small but uh embedded in it is actually something very powerful I'm sure you guys have heard um Jack Dorsey talk about what he's doing with block.","startTime":1363.6,"endTime":1397.52,"durationSeconds":34,"level":"C1","overallScore":8.8,"rationale":"조직 학습 메커니즘 통찰이 강함."},{"segmentIndex":19,"text":"like you can look at the operation of any organization as uh the aggregate of you know I mean the two sentence pitch at YC is that's sort of one of like thousands of things that I would argue we do for founders but you know we just walked through a very concrete way where someone wrote a prompt used it used a bunch more other people used it uh a bunch of artifacts came off of that around literally like the transcript of using it becomes a thing that can be used to metaprompt and improve in an automated fashion on a daily basis the operation of that one skill and then suddenly that one skill you just said it that skill is now better than any of us individually than be you know when before we actually had access to that and so this is like a particular like needle pin prick in the fabric of like how any organization does things and then how do you build super intelligence inside a company you do that on everything you do.","startTime":1412.96,"endTime":1472.799,"durationSeconds":60,"level":"C1","overallScore":8.8,"rationale":"구체 예시로 일반 원리까지 확장함."},{"segmentIndex":24,"text":"But the real idea underneath was just kind of that the potential for AI is to shift control of software from the developer to the user, right?","startTime":1983.76,"endTime":2012.96,"durationSeconds":29,"level":"C1","overallScore":8.8,"rationale":"핵심 통찰이 매우 선명하고 구조도 좋음."},{"segmentIndex":27,"text":"I think the conclusion that this essay points to is that as we get better at building AI native software, it's going to look a lot more like the agent wrapping software deterministic tools rather than deterministic software wrapping AI, right?","startTime":2034.48,"endTime":2053.04,"durationSeconds":19,"level":"C1","overallScore":9,"rationale":"추상적 통찰과 고급 구조가 매우 풍부함."},{"segmentIndex":54,"text":"like they were complaining about like I had made you know this monolithic writer chat interface and it was like full of bugs because I was like reimplementing things that openclaw and telegram already do and now they just use open andclaw telegram and my retrieval system with like all the same data that I extracted it out and with our MCP and it's working great like basically you know Gary's list 2.0 know the next rewrite thankfully is not half a million lines of uh Rails code that is like insane to actually you it's rigid it's uh takes a long time it like takes like 10 times long you know even though it was 1/100th the amount of time to do it like by hand you don't have to do it by hand like that half a million lines of code in Rails is easily like 10,000 lines of like TypeScript and like maybe 2,000 lines of markdown and all of that is way more dynamic like you could just say like actually for the second paragraph uh I really like including a biography of like the politician we're focusing on and it's like I don't have to code that in Rails.","startTime":2239.44,"endTime":2302.96,"durationSeconds":64,"level":"C1","overallScore":9,"rationale":"구식 개발과 신방식 대비가 매우 선명함."},{"segmentIndex":57,"text":"Like this is actually the dawn of just in time software and I can see it right now.","startTime":2321.599,"endTime":2327.119,"durationSeconds":6,"level":"B2","overallScore":8.6,"rationale":"핵심 개념 선언이라 통찰 밀도가 높음."},{"segmentIndex":35,"text":"And that's what I hope that we are building with things like Gbrain and you know Hermes agent and openclaw like the ability to run your own software to change your own prompts to test all of it to have your own private repo that like you know is only yours um to be able to choose which model to use and maybe it's an openweight model like to me that's sort of the white pill for AI is uh we could have corporate control, no control of your own prompts and like literally the AI happens to you.","startTime":2635.44,"endTime":2669.92,"durationSeconds":34,"level":"C1","overallScore":9,"rationale":"핵심 비전과 표현 자원이 매우 풍부함."},{"segmentIndex":1,"text":"How do you build super intelligence inside a company? >> Part of the key thing is not to just use AI as a copilot.","startTime":1.59,"endTime":7.12,"durationSeconds":6,"level":"B2","overallScore":7.8,"rationale":"핵심 원칙 제시와 유용한 표현 포함."},{"segmentIndex":3,"text":"And you need to start recording all the artifacts. >> It's like a shared organizational brain. It's like the closest thing to us being able to like connect our brains.","startTime":11.679,"endTime":19.6,"durationSeconds":8,"level":"B2","overallScore":8,"rationale":"비유가 강하고 학습 표현도 좋음."},{"segmentIndex":11,"text":"And it's been I think a very powerful symbiotic relationship for us to actually be adopting these tools and like transforming our own organization which was started way pre- AI into a super AI native organization ourselves and Pete has really been leading the charge for that.","startTime":98.15899999999999,"endTime":115.2,"durationSeconds":17,"level":"C1","overallScore":8,"rationale":"조직 변화 통찰과 고급 표현 풍부."},{"segmentIndex":21,"text":"The finance team would describe to our software engineers how you know this complicated financial workflow worked and then the software engineers would go and build some purpose-built software where there was a deterministic workflow encapsulating everything that they had been told and then hand it back to the finance team and so on.","startTime":226.239,"endTime":244.879,"durationSeconds":19,"level":"C1","overallScore":8.2,"rationale":"비효율 프로세스 설명이 통찰적임."},{"segmentIndex":27,"text":"And so the original impetus was why don't we try to build some tools at YC that we could use to run agents that would give the finance team control over their own software, right?","startTime":276.72,"endTime":288.56,"durationSeconds":12,"level":"B2","overallScore":8,"rationale":"핵심 아이디어와 실용적 구조가 강함."},{"segmentIndex":34,"text":"I remember I built those tools and I felt a little bit like I was breaking the rules because initially we started with very limited tools that had very narrowly scoped domains and I kept getting frustrated because they weren't powerful enough to do the things that I wanted and so I was like what if we just gave the thing like access complete access to the production database where it could just like trample on anything and I sort of like surreptitiously pushed it out maybe late at Right.","startTime":389.6,"endTime":419.759,"durationSeconds":30,"level":"C1","overallScore":8.4,"rationale":"실험 서사와 고급 구어 표현이 많음."}],"generatedAt":"2026-06-25T00:29:08.424Z","keyClipsTotalSec":1574},{"videoId":"B246K_G7mHU","chunkIndex":2,"totalChunks":5,"title":"Inside YC's AI Playbook — Part 3 of 5","thumbnail":"https://i.ytimg.com/vi/B246K_G7mHU/maxresdefault.jpg","duration":2790,"uploader":"Y Combinator","youtubeUrl":"https://www.youtube.com/watch?v=B246K_G7mHU","keywords":["startup","ai","organization","leadership","knowledge-sharing","automation","productivity","communication","transparency","founder"],"normalizedKeywords":["리더십·매니지먼트","비즈니스·전략","기술 트렌드"],"targetAudience":[{"who":"창업자","why":"AI를 조직 운영의 핵심 레이어로 바꾸는 방식이 직접적 도움됨"},{"who":"스타트업 리더","why":"고신뢰·고투명 조직을 어떻게 설계할지 감을 잡을 수 있음"},{"who":"지식노동자","why":"회의·문서·커뮤니케이션을 AI로 개선하는 실용 팁이 많음"}],"normalizedAudience":["창업자·스타트업","지식노동자 일반"],"summary":"이 영상은 YC가 AI를 단순한 보조 도구가 아니라 조직의 운영 레이어로 활용하는 방식을 설명한다. 핵심 예시는 '두 문장 설명' 스킬로, 파트너들의 경험과 회의 전사를 바탕으로 에이전트가 더 나은 설명을 만들도록 개선해 가는 과정이다. 이를 통해 조직의 암묵지를 프롬프트와 기록으로 축적하면, 개별 사람보다 나은 성능의 '공유된 조직 두뇌'를 만들 수 있다고 주장한다.\n\n또한 이런 AI 네이티브 조직이 작동하려면 높은 신뢰, 기본 공개, 그리고 어느 정도의 평등성이 필요하다고 말한다. 회의 녹음과 내부 브로드캐스트 같은 투명성은 사생활 리스크처럼 보이지만, 실제로는 다른 사람이 쓰는 방식을 학습하게 하고 사회적 통제를 통해 남용을 줄인다. 결국 AI 시대의 경쟁력은 기술 자체보다도, 조직이 맥락을 얼마나 열어두고 학습 가능한 형태로 축적하느냐에 달려 있다는 메시지다.","insights":["좋은 소통은 내 머리의 맥락을 남의 머리로 복제하는 일이다.","조직의 암묵지는 전사와 피드백을 거치면 자동 개선 자산이 된다.","AI는 보조 도구가 아니라 업무를 구성하는 기반이어야 한다.","투명성은 보안의 적이 아니라 집단 학습의 가속장치가 될 수 있다.","AI 네이티브 조직은 신뢰·평등·맥락 개방을 전제로 한다."],"keyClips":[{"clipId":"B246K_G7mHU:c2:2-16","startSegmentIndex":2,"endSegmentIndex":16,"startTime":1210.559,"endTime":1397.52,"durationSeconds":187,"preview":"두 문장 설명의 힘","mustSee":false},{"clipId":"B246K_G7mHU:c2:17-23","startSegmentIndex":17,"endSegmentIndex":23,"startTime":1397.52,"endTime":1516.96,"durationSeconds":119.4,"preview":"슈퍼조직의 작동원리","mustSee":true},{"clipId":"B246K_G7mHU:c2:24-31","startSegmentIndex":24,"endSegmentIndex":31,"startTime":1516.96,"endTime":1599.76,"durationSeconds":82.8,"preview":"코파일럿을 넘어서","mustSee":false},{"clipId":"B246K_G7mHU:c2:32-38","startSegmentIndex":32,"endSegmentIndex":38,"startTime":1599.76,"endTime":1652.08,"durationSeconds":52.3,"preview":"공유된 조직 두뇌","mustSee":false},{"clipId":"B246K_G7mHU:c2:39-59","startSegmentIndex":39,"endSegmentIndex":59,"startTime":1652.08,"endTime":1799.279,"durationSeconds":147.2,"preview":"신뢰 기반 AI 조직","mustSee":true}],"curatedSegments":[{"segmentIndex":4,"text":"If you frame this as a way for everyone in an organization to get better at what they do using the like collective skill and instinct of the people they work with, it's incredibly powerful. Today we have a real treat.","startTime":19.6,"endTime":41.99,"durationSeconds":22,"level":"C1","overallScore":9,"rationale":"조직 학습 통찰과 구조 표현 풍부."},{"segmentIndex":28,"text":"like remove the software engineers from this crazy loop where they have to sort of understand these complicated workflows and give the finance team the tools that they could use to encode their own workflows not as you know not as Ruby uh but as English with prompts right >> I mean what's interesting is like uh we all funded companies like maybe even like two or three years ago when LLMs were out but like agent coding wasn't a thing yet and so the first thing actually was not a coding it was LLMs for writing SQL queries.","startTime":288.56,"endTime":318.72,"durationSeconds":30,"level":"C1","overallScore":9.2,"rationale":"핵심 전환 논리와 표현 밀도 매우 높음."},{"segmentIndex":31,"text":"I was really surprised too to be honest and so that we started with this kind of purpose-built thing for finance and then rewrote it to even more of a general agent loop right and it and it's this is now you see these all over the place now but um I the first kind of magical moment that I had was we had this agent loop and we had a tool registry a shared tool registry for kind of YC specific tools and the first tool that really was an unlock for me was I think a tool looking back that you actually built Jared.","startTime":344.88,"endTime":376.08,"durationSeconds":31,"level":"C1","overallScore":9.2,"rationale":"마법 같은 전환점 서사와 표현 풍부."},{"segmentIndex":45,"text":"It just turns out when all of that context is in one place with a little bit of additional uh information about how the schema is laid out, an agent can go and ask any or answer arbitrary questions about our business.","startTime":509.52,"endTime":521.279,"durationSeconds":12,"level":"C1","overallScore":9,"rationale":"맥락 집약의 힘을 잘 설명한 문장."},{"segmentIndex":48,"text":"It dramatically increased the number of questions that we would ask and dramatically increased the scale and complexity of the questions that we would dare to ask where like you know in the old days back when we were the in the old days back when we were using like BI tools to ask a question like that you know like what investors have invested like in space related companies that would be like several hours of writing SQL and so like unless it was really important you just wouldn't bother.","startTime":528.56,"endTime":550,"durationSeconds":21,"level":"C1","overallScore":9.2,"rationale":"행동 변화 통찰과 자연스런 표현 풍부."},{"segmentIndex":49,"text":"It's just another example of the you know this instance of Gvon's paradox that you get when you remove the amount of back and forth uh between different teams in order to get a thing done right if in order to answer ask some kind of complex question about YC I have to go and knock on you know the data science team's door and wait for them to get it through you know their backlog uh I'm just going to ask far fewer questions.","startTime":550,"endTime":576.48,"durationSeconds":26,"level":"C1","overallScore":9.2,"rationale":"마찰 감소의 원리를 설득력 있게 설명."},{"segmentIndex":7,"text":"Interestingly, uh I now realize YC itself is a context engineering uh sort of process in that like people we're frequently teaching people you have perfect context about what's going on in your brain, but great communication is replicating that same context in someone else's brain.","startTime":1245.2,"endTime":1264.32,"durationSeconds":19,"level":"C1","overallScore":8.8,"rationale":"강한 통찰과 추상적 설명이 공존."},{"segmentIndex":16,"text":"and now lived in the context of that meeting transcript and handing that back to the agent and saying given you know what you've learned by reading through this context improve the two sentence description skill and they got noticeably better after that like this thing is now better than I am I would I thing is now better than I am I would argue at writing those >> this is how super intelligence happens inside organizations I mean this two sentence pitch thing sounds like something kind of small but uh embedded in it is actually something very powerful I'm sure you guys have heard um Jack Dorsey talk about what he's doing with block.","startTime":1363.6,"endTime":1397.52,"durationSeconds":34,"level":"C1","overallScore":8.8,"rationale":"조직 학습 메커니즘 통찰이 강함."},{"segmentIndex":19,"text":"like you can look at the operation of any organization as uh the aggregate of you know I mean the two sentence pitch at YC is that's sort of one of like thousands of things that I would argue we do for founders but you know we just walked through a very concrete way where someone wrote a prompt used it used a bunch more other people used it uh a bunch of artifacts came off of that around literally like the transcript of using it becomes a thing that can be used to metaprompt and improve in an automated fashion on a daily basis the operation of that one skill and then suddenly that one skill you just said it that skill is now better than any of us individually than be you know when before we actually had access to that and so this is like a particular like needle pin prick in the fabric of like how any organization does things and then how do you build super intelligence inside a company you do that on everything you do.","startTime":1412.96,"endTime":1472.799,"durationSeconds":60,"level":"C1","overallScore":8.8,"rationale":"구체 예시로 일반 원리까지 확장함."},{"segmentIndex":24,"text":"But the real idea underneath was just kind of that the potential for AI is to shift control of software from the developer to the user, right?","startTime":1983.76,"endTime":2012.96,"durationSeconds":29,"level":"C1","overallScore":8.8,"rationale":"핵심 통찰이 매우 선명하고 구조도 좋음."},{"segmentIndex":27,"text":"I think the conclusion that this essay points to is that as we get better at building AI native software, it's going to look a lot more like the agent wrapping software deterministic tools rather than deterministic software wrapping AI, right?","startTime":2034.48,"endTime":2053.04,"durationSeconds":19,"level":"C1","overallScore":9,"rationale":"추상적 통찰과 고급 구조가 매우 풍부함."},{"segmentIndex":54,"text":"like they were complaining about like I had made you know this monolithic writer chat interface and it was like full of bugs because I was like reimplementing things that openclaw and telegram already do and now they just use open andclaw telegram and my retrieval system with like all the same data that I extracted it out and with our MCP and it's working great like basically you know Gary's list 2.0 know the next rewrite thankfully is not half a million lines of uh Rails code that is like insane to actually you it's rigid it's uh takes a long time it like takes like 10 times long you know even though it was 1/100th the amount of time to do it like by hand you don't have to do it by hand like that half a million lines of code in Rails is easily like 10,000 lines of like TypeScript and like maybe 2,000 lines of markdown and all of that is way more dynamic like you could just say like actually for the second paragraph uh I really like including a biography of like the politician we're focusing on and it's like I don't have to code that in Rails.","startTime":2239.44,"endTime":2302.96,"durationSeconds":64,"level":"C1","overallScore":9,"rationale":"구식 개발과 신방식 대비가 매우 선명함."},{"segmentIndex":57,"text":"Like this is actually the dawn of just in time software and I can see it right now.","startTime":2321.599,"endTime":2327.119,"durationSeconds":6,"level":"B2","overallScore":8.6,"rationale":"핵심 개념 선언이라 통찰 밀도가 높음."},{"segmentIndex":35,"text":"And that's what I hope that we are building with things like Gbrain and you know Hermes agent and openclaw like the ability to run your own software to change your own prompts to test all of it to have your own private repo that like you know is only yours um to be able to choose which model to use and maybe it's an openweight model like to me that's sort of the white pill for AI is uh we could have corporate control, no control of your own prompts and like literally the AI happens to you.","startTime":2635.44,"endTime":2669.92,"durationSeconds":34,"level":"C1","overallScore":9,"rationale":"핵심 비전과 표현 자원이 매우 풍부함."},{"segmentIndex":1,"text":"How do you build super intelligence inside a company? >> Part of the key thing is not to just use AI as a copilot.","startTime":1.59,"endTime":7.12,"durationSeconds":6,"level":"B2","overallScore":7.8,"rationale":"핵심 원칙 제시와 유용한 표현 포함."},{"segmentIndex":3,"text":"And you need to start recording all the artifacts. >> It's like a shared organizational brain. It's like the closest thing to us being able to like connect our brains.","startTime":11.679,"endTime":19.6,"durationSeconds":8,"level":"B2","overallScore":8,"rationale":"비유가 강하고 학습 표현도 좋음."},{"segmentIndex":11,"text":"And it's been I think a very powerful symbiotic relationship for us to actually be adopting these tools and like transforming our own organization which was started way pre- AI into a super AI native organization ourselves and Pete has really been leading the charge for that.","startTime":98.15899999999999,"endTime":115.2,"durationSeconds":17,"level":"C1","overallScore":8,"rationale":"조직 변화 통찰과 고급 표현 풍부."},{"segmentIndex":21,"text":"The finance team would describe to our software engineers how you know this complicated financial workflow worked and then the software engineers would go and build some purpose-built software where there was a deterministic workflow encapsulating everything that they had been told and then hand it back to the finance team and so on.","startTime":226.239,"endTime":244.879,"durationSeconds":19,"level":"C1","overallScore":8.2,"rationale":"비효율 프로세스 설명이 통찰적임."},{"segmentIndex":27,"text":"And so the original impetus was why don't we try to build some tools at YC that we could use to run agents that would give the finance team control over their own software, right?","startTime":276.72,"endTime":288.56,"durationSeconds":12,"level":"B2","overallScore":8,"rationale":"핵심 아이디어와 실용적 구조가 강함."},{"segmentIndex":34,"text":"I remember I built those tools and I felt a little bit like I was breaking the rules because initially we started with very limited tools that had very narrowly scoped domains and I kept getting frustrated because they weren't powerful enough to do the things that I wanted and so I was like what if we just gave the thing like access complete access to the production database where it could just like trample on anything and I sort of like surreptitiously pushed it out maybe late at Right.","startTime":389.6,"endTime":419.759,"durationSeconds":30,"level":"C1","overallScore":8.4,"rationale":"실험 서사와 고급 구어 표현이 많음."}],"generatedAt":"2026-06-25T00:29:30.813Z","keyClipsTotalSec":1574},{"videoId":"B246K_G7mHU","chunkIndex":3,"totalChunks":5,"title":"Inside YC's AI Playbook — Part 4 of 5","thumbnail":"https://i.ytimg.com/vi/B246K_G7mHU/maxresdefault.jpg","duration":2790,"uploader":"Y Combinator","youtubeUrl":"https://www.youtube.com/watch?v=B246K_G7mHU","keywords":["ai-native","software-design","chat-interface","automation","startups","developer-tools","productivity","future-of-work","enterprise-ai"],"normalizedKeywords":["엔지니어링","프로덕트","기술 트렌드"],"targetAudience":[{"who":"창업자","why":"AI를 제품과 조직 운영에 어떻게 녹일지 판단하는 데 직접적 힌트를 준다"},{"who":"엔지니어","why":"AI 네이티브 소프트웨어 구조와 에이전트 기반 구현 원리를 배울 수 있다"},{"who":"프로덕트 매니저","why":"AI 제품의 인터페이스와 사용자 제어권 설계에 대한 관점을 얻을 수 있다"}],"normalizedAudience":["창업자·스타트업","엔지니어·개발자","프로덕트 매니저·기획자"],"summary":"이 대화는 AI가 소프트웨어와 조직을 어떻게 바꾸는지에 대한 YC식 관점을 압축해서 보여준다. 핵심은 AI를 기존 소프트웨어에 얇게 덧붙이는 것이 아니라, 에이전트가 도구를 감싸는 방식으로 재설계해야 한다는 주장이다. 그렇게 하면 비용이 급격히 내려가는 기술을 지금 당장 활용해 '2028년의 생산성'을 먼저 살 수 있고, 신입 직원도 회사의 맥락을 훨씬 빠르게 익힐 수 있다.\n\n또한 화자들은 chat이 AI의 기본 인터페이스가 될 가능성에 강하게 수렴한다. 텍스트·음성·파일을 모두 받아들이는 대화형 방식이 인간의 사고 표현에 가장 가깝고, 복잡한 UI보다 신뢰와 즉시성을 살리기 쉽다고 본다. 실제로 자신들이 만든 대규모 코드베이스와 툴링 사례를 들어, 미래의 소프트웨어는 사전에 거대한 시스템을 짜는 방식보다 '작고 단순한 기본 골격 위에서 즉시 생성되는 just-in-time software'에 가까워질 것이라 결론낸다.","insights":["AI는 미래를 앞당겨 쓰는 도구라, 지금의 비용으로 2년 뒤 생산성을 산다.","좋은 AI 소프트웨어는 기능 추가보다 사용자 통제권의 재배치에 가깝다.","신입 온보딩의 병목은 지식 부족이 아니라 질문하기 어려움이다.","chat은 가장 단순해서가 아니라 인간 사고에 가장 가까워서 강하다.","미래의 소프트웨어는 미리 다 만드는 것보다 필요할 때 생성되는 쪽이다."],"keyClips":[{"clipId":"B246K_G7mHU:c3:1-5","startSegmentIndex":1,"endSegmentIndex":5,"startTime":1800.31,"endTime":1841.12,"durationSeconds":40.8,"preview":"AI가 만드는 시간차익","mustSee":false},{"clipId":"B246K_G7mHU:c3:9-17","startSegmentIndex":9,"endSegmentIndex":17,"startTime":1860.559,"endTime":1939.919,"durationSeconds":79.4,"preview":"온보딩을 자동화하라","mustSee":false},{"clipId":"B246K_G7mHU:c3:20-27","startSegmentIndex":20,"endSegmentIndex":27,"startTime":1952.32,"endTime":2053.04,"durationSeconds":100.7,"preview":"AI는 사용자에게로","mustSee":true},{"clipId":"B246K_G7mHU:c3:29-39","startSegmentIndex":29,"endSegmentIndex":39,"startTime":2059.2,"endTime":2150.72,"durationSeconds":91.5,"preview":"챗이 기본 인터페이스","mustSee":false},{"clipId":"B246K_G7mHU:c3:44-57","startSegmentIndex":44,"endSegmentIndex":57,"startTime":2169.52,"endTime":2327.119,"durationSeconds":157.6,"preview":"지금의 소프트웨어 혁명","mustSee":true},{"clipId":"B246K_G7mHU:c3:58-64","startSegmentIndex":58,"endSegmentIndex":64,"startTime":2327.119,"endTime":2382.32,"durationSeconds":55.2,"preview":"작게 시작해야 강하다","mustSee":false}],"curatedSegments":[{"segmentIndex":4,"text":"If you frame this as a way for everyone in an organization to get better at what they do using the like collective skill and instinct of the people they work with, it's incredibly powerful. Today we have a real treat.","startTime":19.6,"endTime":41.99,"durationSeconds":22,"level":"C1","overallScore":9,"rationale":"조직 학습 통찰과 구조 표현 풍부."},{"segmentIndex":28,"text":"like remove the software engineers from this crazy loop where they have to sort of understand these complicated workflows and give the finance team the tools that they could use to encode their own workflows not as you know not as Ruby uh but as English with prompts right >> I mean what's interesting is like uh we all funded companies like maybe even like two or three years ago when LLMs were out but like agent coding wasn't a thing yet and so the first thing actually was not a coding it was LLMs for writing SQL queries.","startTime":288.56,"endTime":318.72,"durationSeconds":30,"level":"C1","overallScore":9.2,"rationale":"핵심 전환 논리와 표현 밀도 매우 높음."},{"segmentIndex":31,"text":"I was really surprised too to be honest and so that we started with this kind of purpose-built thing for finance and then rewrote it to even more of a general agent loop right and it and it's this is now you see these all over the place now but um I the first kind of magical moment that I had was we had this agent loop and we had a tool registry a shared tool registry for kind of YC specific tools and the first tool that really was an unlock for me was I think a tool looking back that you actually built Jared.","startTime":344.88,"endTime":376.08,"durationSeconds":31,"level":"C1","overallScore":9.2,"rationale":"마법 같은 전환점 서사와 표현 풍부."},{"segmentIndex":45,"text":"It just turns out when all of that context is in one place with a little bit of additional uh information about how the schema is laid out, an agent can go and ask any or answer arbitrary questions about our business.","startTime":509.52,"endTime":521.279,"durationSeconds":12,"level":"C1","overallScore":9,"rationale":"맥락 집약의 힘을 잘 설명한 문장."},{"segmentIndex":48,"text":"It dramatically increased the number of questions that we would ask and dramatically increased the scale and complexity of the questions that we would dare to ask where like you know in the old days back when we were the in the old days back when we were using like BI tools to ask a question like that you know like what investors have invested like in space related companies that would be like several hours of writing SQL and so like unless it was really important you just wouldn't bother.","startTime":528.56,"endTime":550,"durationSeconds":21,"level":"C1","overallScore":9.2,"rationale":"행동 변화 통찰과 자연스런 표현 풍부."},{"segmentIndex":49,"text":"It's just another example of the you know this instance of Gvon's paradox that you get when you remove the amount of back and forth uh between different teams in order to get a thing done right if in order to answer ask some kind of complex question about YC I have to go and knock on you know the data science team's door and wait for them to get it through you know their backlog uh I'm just going to ask far fewer questions.","startTime":550,"endTime":576.48,"durationSeconds":26,"level":"C1","overallScore":9.2,"rationale":"마찰 감소의 원리를 설득력 있게 설명."},{"segmentIndex":7,"text":"Interestingly, uh I now realize YC itself is a context engineering uh sort of process in that like people we're frequently teaching people you have perfect context about what's going on in your brain, but great communication is replicating that same context in someone else's brain.","startTime":1245.2,"endTime":1264.32,"durationSeconds":19,"level":"C1","overallScore":8.8,"rationale":"강한 통찰과 추상적 설명이 공존."},{"segmentIndex":16,"text":"and now lived in the context of that meeting transcript and handing that back to the agent and saying given you know what you've learned by reading through this context improve the two sentence description skill and they got noticeably better after that like this thing is now better than I am I would I thing is now better than I am I would argue at writing those >> this is how super intelligence happens inside organizations I mean this two sentence pitch thing sounds like something kind of small but uh embedded in it is actually something very powerful I'm sure you guys have heard um Jack Dorsey talk about what he's doing with block.","startTime":1363.6,"endTime":1397.52,"durationSeconds":34,"level":"C1","overallScore":8.8,"rationale":"조직 학습 메커니즘 통찰이 강함."},{"segmentIndex":19,"text":"like you can look at the operation of any organization as uh the aggregate of you know I mean the two sentence pitch at YC is that's sort of one of like thousands of things that I would argue we do for founders but you know we just walked through a very concrete way where someone wrote a prompt used it used a bunch more other people used it uh a bunch of artifacts came off of that around literally like the transcript of using it becomes a thing that can be used to metaprompt and improve in an automated fashion on a daily basis the operation of that one skill and then suddenly that one skill you just said it that skill is now better than any of us individually than be you know when before we actually had access to that and so this is like a particular like needle pin prick in the fabric of like how any organization does things and then how do you build super intelligence inside a company you do that on everything you do.","startTime":1412.96,"endTime":1472.799,"durationSeconds":60,"level":"C1","overallScore":8.8,"rationale":"구체 예시로 일반 원리까지 확장함."},{"segmentIndex":24,"text":"But the real idea underneath was just kind of that the potential for AI is to shift control of software from the developer to the user, right?","startTime":1983.76,"endTime":2012.96,"durationSeconds":29,"level":"C1","overallScore":8.8,"rationale":"핵심 통찰이 매우 선명하고 구조도 좋음."},{"segmentIndex":27,"text":"I think the conclusion that this essay points to is that as we get better at building AI native software, it's going to look a lot more like the agent wrapping software deterministic tools rather than deterministic software wrapping AI, right?","startTime":2034.48,"endTime":2053.04,"durationSeconds":19,"level":"C1","overallScore":9,"rationale":"추상적 통찰과 고급 구조가 매우 풍부함."},{"segmentIndex":54,"text":"like they were complaining about like I had made you know this monolithic writer chat interface and it was like full of bugs because I was like reimplementing things that openclaw and telegram already do and now they just use open andclaw telegram and my retrieval system with like all the same data that I extracted it out and with our MCP and it's working great like basically you know Gary's list 2.0 know the next rewrite thankfully is not half a million lines of uh Rails code that is like insane to actually you it's rigid it's uh takes a long time it like takes like 10 times long you know even though it was 1/100th the amount of time to do it like by hand you don't have to do it by hand like that half a million lines of code in Rails is easily like 10,000 lines of like TypeScript and like maybe 2,000 lines of markdown and all of that is way more dynamic like you could just say like actually for the second paragraph uh I really like including a biography of like the politician we're focusing on and it's like I don't have to code that in Rails.","startTime":2239.44,"endTime":2302.96,"durationSeconds":64,"level":"C1","overallScore":9,"rationale":"구식 개발과 신방식 대비가 매우 선명함."},{"segmentIndex":57,"text":"Like this is actually the dawn of just in time software and I can see it right now.","startTime":2321.599,"endTime":2327.119,"durationSeconds":6,"level":"B2","overallScore":8.6,"rationale":"핵심 개념 선언이라 통찰 밀도가 높음."},{"segmentIndex":35,"text":"And that's what I hope that we are building with things like Gbrain and you know Hermes agent and openclaw like the ability to run your own software to change your own prompts to test all of it to have your own private repo that like you know is only yours um to be able to choose which model to use and maybe it's an openweight model like to me that's sort of the white pill for AI is uh we could have corporate control, no control of your own prompts and like literally the AI happens to you.","startTime":2635.44,"endTime":2669.92,"durationSeconds":34,"level":"C1","overallScore":9,"rationale":"핵심 비전과 표현 자원이 매우 풍부함."},{"segmentIndex":1,"text":"How do you build super intelligence inside a company? >> Part of the key thing is not to just use AI as a copilot.","startTime":1.59,"endTime":7.12,"durationSeconds":6,"level":"B2","overallScore":7.8,"rationale":"핵심 원칙 제시와 유용한 표현 포함."},{"segmentIndex":3,"text":"And you need to start recording all the artifacts. >> It's like a shared organizational brain. It's like the closest thing to us being able to like connect our brains.","startTime":11.679,"endTime":19.6,"durationSeconds":8,"level":"B2","overallScore":8,"rationale":"비유가 강하고 학습 표현도 좋음."},{"segmentIndex":11,"text":"And it's been I think a very powerful symbiotic relationship for us to actually be adopting these tools and like transforming our own organization which was started way pre- AI into a super AI native organization ourselves and Pete has really been leading the charge for that.","startTime":98.15899999999999,"endTime":115.2,"durationSeconds":17,"level":"C1","overallScore":8,"rationale":"조직 변화 통찰과 고급 표현 풍부."},{"segmentIndex":21,"text":"The finance team would describe to our software engineers how you know this complicated financial workflow worked and then the software engineers would go and build some purpose-built software where there was a deterministic workflow encapsulating everything that they had been told and then hand it back to the finance team and so on.","startTime":226.239,"endTime":244.879,"durationSeconds":19,"level":"C1","overallScore":8.2,"rationale":"비효율 프로세스 설명이 통찰적임."},{"segmentIndex":27,"text":"And so the original impetus was why don't we try to build some tools at YC that we could use to run agents that would give the finance team control over their own software, right?","startTime":276.72,"endTime":288.56,"durationSeconds":12,"level":"B2","overallScore":8,"rationale":"핵심 아이디어와 실용적 구조가 강함."},{"segmentIndex":34,"text":"I remember I built those tools and I felt a little bit like I was breaking the rules because initially we started with very limited tools that had very narrowly scoped domains and I kept getting frustrated because they weren't powerful enough to do the things that I wanted and so I was like what if we just gave the thing like access complete access to the production database where it could just like trample on anything and I sort of like surreptitiously pushed it out maybe late at Right.","startTime":389.6,"endTime":419.759,"durationSeconds":30,"level":"C1","overallScore":8.4,"rationale":"실험 서사와 고급 구어 표현이 많음."}],"generatedAt":"2026-06-25T00:29:51.538Z","keyClipsTotalSec":1574},{"videoId":"B246K_G7mHU","chunkIndex":4,"totalChunks":5,"title":"Inside YC's AI Playbook — Part 5 of 5","thumbnail":"https://i.ytimg.com/vi/B246K_G7mHU/maxresdefault.jpg","duration":2790,"uploader":"Y Combinator","youtubeUrl":"https://www.youtube.com/watch?v=B246K_G7mHU","keywords":["ai","personal-computing","open-source","software","automation","decentralization","platform-policy","startup","productivity","future-of-work"],"normalizedKeywords":["비즈니스·전략","엔지니어링","기술 트렌드"],"targetAudience":[{"who":"창업자","why":"AI 제품의 개방성, 통제권, 배포 방향에 대한 전략적 관점을 얻을 수 있음"},{"who":"엔지니어","why":"에이전트로 소프트웨어를 확장하고 개인화하는 기술적 흐름을 이해하는 데 유용함"},{"who":"지식노동자","why":"AI가 업무를 대체하기보다 개인 역량을 증폭시키는 방식으로 쓰일 수 있음을 보여줌"}],"normalizedAudience":["창업자·스타트업","엔지니어·개발자","지식노동자 일반"],"summary":"이 영상은 AI가 앞으로 어떤 방식으로 사회에 배치될지를 두고, 중앙집중형 통제와 개인이 직접 제어하는 분산형 미래를 대비시킨다. 화자는 Gmail처럼 프롬프트조차 바꿀 수 없는 폐쇄적 AI 환경을 경계하며, 개인 컴퓨터 혁명처럼 누구나 실험하고 수정할 수 있는 '개인 AI'가 더 바람직한 방향이라고 주장한다.\n\n또한 오픈소스 하니스와 에이전트 기반 도구를 예로 들며, 소프트웨어가 점점 '자기 확장적'이고 '사용자 맞춤형'이 되는 흐름을 강조한다. 핵심 메시지는 AI가 사람을 대체하는 기술이 아니라, 사람이 자기 목적에 맞게 프로그래밍하고 통제할 수 있을 때 진짜 가치가 생긴다는 것이다. 결국 이 미래는 기술의 문제만이 아니라, 기업과 사용자 모두가 어떤 선택을 하느냐의 문제로 정리된다.","insights":["AI의 핵심 쟁점은 성능보다 통제권이다.","개인화된 도구가 혁신을 만드는 힘이 더 크다.","AI가 중앙집중되면 사용자는 기술의 소비자로만 남는다.","진짜 생산성은 반복 노동 제거보다 자기화에서 나온다.","기업도 기본값은 폐쇄형이므로 의도적 개방 설계가 필요하다."],"keyClips":[{"clipId":"B246K_G7mHU:c4:1-8","startSegmentIndex":1,"endSegmentIndex":8,"startTime":2400.55,"endTime":2445.04,"durationSeconds":44.5,"preview":"자기확장 소프트웨어","mustSee":false},{"clipId":"B246K_G7mHU:c4:9-18","startSegmentIndex":9,"endSegmentIndex":18,"startTime":2445.04,"endTime":2520.56,"durationSeconds":75.5,"preview":"AI의 중앙집중 위험","mustSee":true},{"clipId":"B246K_G7mHU:c4:19-28","startSegmentIndex":19,"endSegmentIndex":28,"startTime":2520.56,"endTime":2594.56,"durationSeconds":74,"preview":"개인PC의 역사 반복","mustSee":false},{"clipId":"B246K_G7mHU:c4:29-36","startSegmentIndex":29,"endSegmentIndex":36,"startTime":2594.56,"endTime":2681.359,"durationSeconds":86.8,"preview":"진짜 개인 AI","mustSee":true},{"clipId":"B246K_G7mHU:c4:37-44","startSegmentIndex":37,"endSegmentIndex":44,"startTime":2681.359,"endTime":2763.359,"durationSeconds":82,"preview":"AI는 사람을 증폭","mustSee":true}],"curatedSegments":[{"segmentIndex":4,"text":"If you frame this as a way for everyone in an organization to get better at what they do using the like collective skill and instinct of the people they work with, it's incredibly powerful. Today we have a real treat.","startTime":19.6,"endTime":41.99,"durationSeconds":22,"level":"C1","overallScore":9,"rationale":"조직 학습 통찰과 구조 표현 풍부."},{"segmentIndex":28,"text":"like remove the software engineers from this crazy loop where they have to sort of understand these complicated workflows and give the finance team the tools that they could use to encode their own workflows not as you know not as Ruby uh but as English with prompts right >> I mean what's interesting is like uh we all funded companies like maybe even like two or three years ago when LLMs were out but like agent coding wasn't a thing yet and so the first thing actually was not a coding it was LLMs for writing SQL queries.","startTime":288.56,"endTime":318.72,"durationSeconds":30,"level":"C1","overallScore":9.2,"rationale":"핵심 전환 논리와 표현 밀도 매우 높음."},{"segmentIndex":31,"text":"I was really surprised too to be honest and so that we started with this kind of purpose-built thing for finance and then rewrote it to even more of a general agent loop right and it and it's this is now you see these all over the place now but um I the first kind of magical moment that I had was we had this agent loop and we had a tool registry a shared tool registry for kind of YC specific tools and the first tool that really was an unlock for me was I think a tool looking back that you actually built Jared.","startTime":344.88,"endTime":376.08,"durationSeconds":31,"level":"C1","overallScore":9.2,"rationale":"마법 같은 전환점 서사와 표현 풍부."},{"segmentIndex":45,"text":"It just turns out when all of that context is in one place with a little bit of additional uh information about how the schema is laid out, an agent can go and ask any or answer arbitrary questions about our business.","startTime":509.52,"endTime":521.279,"durationSeconds":12,"level":"C1","overallScore":9,"rationale":"맥락 집약의 힘을 잘 설명한 문장."},{"segmentIndex":48,"text":"It dramatically increased the number of questions that we would ask and dramatically increased the scale and complexity of the questions that we would dare to ask where like you know in the old days back when we were the in the old days back when we were using like BI tools to ask a question like that you know like what investors have invested like in space related companies that would be like several hours of writing SQL and so like unless it was really important you just wouldn't bother.","startTime":528.56,"endTime":550,"durationSeconds":21,"level":"C1","overallScore":9.2,"rationale":"행동 변화 통찰과 자연스런 표현 풍부."},{"segmentIndex":49,"text":"It's just another example of the you know this instance of Gvon's paradox that you get when you remove the amount of back and forth uh between different teams in order to get a thing done right if in order to answer ask some kind of complex question about YC I have to go and knock on you know the data science team's door and wait for them to get it through you know their backlog uh I'm just going to ask far fewer questions.","startTime":550,"endTime":576.48,"durationSeconds":26,"level":"C1","overallScore":9.2,"rationale":"마찰 감소의 원리를 설득력 있게 설명."},{"segmentIndex":7,"text":"Interestingly, uh I now realize YC itself is a context engineering uh sort of process in that like people we're frequently teaching people you have perfect context about what's going on in your brain, but great communication is replicating that same context in someone else's brain.","startTime":1245.2,"endTime":1264.32,"durationSeconds":19,"level":"C1","overallScore":8.8,"rationale":"강한 통찰과 추상적 설명이 공존."},{"segmentIndex":16,"text":"and now lived in the context of that meeting transcript and handing that back to the agent and saying given you know what you've learned by reading through this context improve the two sentence description skill and they got noticeably better after that like this thing is now better than I am I would I thing is now better than I am I would argue at writing those >> this is how super intelligence happens inside organizations I mean this two sentence pitch thing sounds like something kind of small but uh embedded in it is actually something very powerful I'm sure you guys have heard um Jack Dorsey talk about what he's doing with block.","startTime":1363.6,"endTime":1397.52,"durationSeconds":34,"level":"C1","overallScore":8.8,"rationale":"조직 학습 메커니즘 통찰이 강함."},{"segmentIndex":19,"text":"like you can look at the operation of any organization as uh the aggregate of you know I mean the two sentence pitch at YC is that's sort of one of like thousands of things that I would argue we do for founders but you know we just walked through a very concrete way where someone wrote a prompt used it used a bunch more other people used it uh a bunch of artifacts came off of that around literally like the transcript of using it becomes a thing that can be used to metaprompt and improve in an automated fashion on a daily basis the operation of that one skill and then suddenly that one skill you just said it that skill is now better than any of us individually than be you know when before we actually had access to that and so this is like a particular like needle pin prick in the fabric of like how any organization does things and then how do you build super intelligence inside a company you do that on everything you do.","startTime":1412.96,"endTime":1472.799,"durationSeconds":60,"level":"C1","overallScore":8.8,"rationale":"구체 예시로 일반 원리까지 확장함."},{"segmentIndex":24,"text":"But the real idea underneath was just kind of that the potential for AI is to shift control of software from the developer to the user, right?","startTime":1983.76,"endTime":2012.96,"durationSeconds":29,"level":"C1","overallScore":8.8,"rationale":"핵심 통찰이 매우 선명하고 구조도 좋음."},{"segmentIndex":27,"text":"I think the conclusion that this essay points to is that as we get better at building AI native software, it's going to look a lot more like the agent wrapping software deterministic tools rather than deterministic software wrapping AI, right?","startTime":2034.48,"endTime":2053.04,"durationSeconds":19,"level":"C1","overallScore":9,"rationale":"추상적 통찰과 고급 구조가 매우 풍부함."},{"segmentIndex":54,"text":"like they were complaining about like I had made you know this monolithic writer chat interface and it was like full of bugs because I was like reimplementing things that openclaw and telegram already do and now they just use open andclaw telegram and my retrieval system with like all the same data that I extracted it out and with our MCP and it's working great like basically you know Gary's list 2.0 know the next rewrite thankfully is not half a million lines of uh Rails code that is like insane to actually you it's rigid it's uh takes a long time it like takes like 10 times long you know even though it was 1/100th the amount of time to do it like by hand you don't have to do it by hand like that half a million lines of code in Rails is easily like 10,000 lines of like TypeScript and like maybe 2,000 lines of markdown and all of that is way more dynamic like you could just say like actually for the second paragraph uh I really like including a biography of like the politician we're focusing on and it's like I don't have to code that in Rails.","startTime":2239.44,"endTime":2302.96,"durationSeconds":64,"level":"C1","overallScore":9,"rationale":"구식 개발과 신방식 대비가 매우 선명함."},{"segmentIndex":57,"text":"Like this is actually the dawn of just in time software and I can see it right now.","startTime":2321.599,"endTime":2327.119,"durationSeconds":6,"level":"B2","overallScore":8.6,"rationale":"핵심 개념 선언이라 통찰 밀도가 높음."},{"segmentIndex":35,"text":"And that's what I hope that we are building with things like Gbrain and you know Hermes agent and openclaw like the ability to run your own software to change your own prompts to test all of it to have your own private repo that like you know is only yours um to be able to choose which model to use and maybe it's an openweight model like to me that's sort of the white pill for AI is uh we could have corporate control, no control of your own prompts and like literally the AI happens to you.","startTime":2635.44,"endTime":2669.92,"durationSeconds":34,"level":"C1","overallScore":9,"rationale":"핵심 비전과 표현 자원이 매우 풍부함."},{"segmentIndex":1,"text":"How do you build super intelligence inside a company? >> Part of the key thing is not to just use AI as a copilot.","startTime":1.59,"endTime":7.12,"durationSeconds":6,"level":"B2","overallScore":7.8,"rationale":"핵심 원칙 제시와 유용한 표현 포함."},{"segmentIndex":3,"text":"And you need to start recording all the artifacts. >> It's like a shared organizational brain. It's like the closest thing to us being able to like connect our brains.","startTime":11.679,"endTime":19.6,"durationSeconds":8,"level":"B2","overallScore":8,"rationale":"비유가 강하고 학습 표현도 좋음."},{"segmentIndex":11,"text":"And it's been I think a very powerful symbiotic relationship for us to actually be adopting these tools and like transforming our own organization which was started way pre- AI into a super AI native organization ourselves and Pete has really been leading the charge for that.","startTime":98.15899999999999,"endTime":115.2,"durationSeconds":17,"level":"C1","overallScore":8,"rationale":"조직 변화 통찰과 고급 표현 풍부."},{"segmentIndex":21,"text":"The finance team would describe to our software engineers how you know this complicated financial workflow worked and then the software engineers would go and build some purpose-built software where there was a deterministic workflow encapsulating everything that they had been told and then hand it back to the finance team and so on.","startTime":226.239,"endTime":244.879,"durationSeconds":19,"level":"C1","overallScore":8.2,"rationale":"비효율 프로세스 설명이 통찰적임."},{"segmentIndex":27,"text":"And so the original impetus was why don't we try to build some tools at YC that we could use to run agents that would give the finance team control over their own software, right?","startTime":276.72,"endTime":288.56,"durationSeconds":12,"level":"B2","overallScore":8,"rationale":"핵심 아이디어와 실용적 구조가 강함."},{"segmentIndex":34,"text":"I remember I built those tools and I felt a little bit like I was breaking the rules because initially we started with very limited tools that had very narrowly scoped domains and I kept getting frustrated because they weren't powerful enough to do the things that I wanted and so I was like what if we just gave the thing like access complete access to the production database where it could just like trample on anything and I sort of like surreptitiously pushed it out maybe late at Right.","startTime":389.6,"endTime":419.759,"durationSeconds":30,"level":"C1","overallScore":8.4,"rationale":"실험 서사와 고급 구어 표현이 많음."}],"generatedAt":"2026-06-25T00:30:16.676Z","keyClipsTotalSec":1574}]}