{"success":true,"count":5,"items":[{"videoId":"Rumft-rsEu4","chunkIndex":0,"totalChunks":5,"title":"Knowing What Your Customers Want, All the Time: Listen Labs' Alfred Wahlforss — Part 1 of 5","thumbnail":"https://i.ytimg.com/vi/Rumft-rsEu4/maxresdefault.jpg","duration":2521,"uploader":"Sequoia Capital","youtubeUrl":"https://www.youtube.com/watch?v=Rumft-rsEu4","keywords":["customer-research","ai","market-research","voice-interviews","startups","product-management","advertising","user-feedback","saas","automation"],"normalizedKeywords":["비즈니스·전략","프로덕트","마케팅"],"targetAudience":[{"who":"창업자","why":"고객을 더 빠르고 싸게 이해하는 리서치 체계를 배울 수 있음"},{"who":"프로덕트 매니저","why":"설문보다 인터뷰와 행동 데이터 결합의 실무적 기준을 얻을 수 있음"},{"who":"마케터","why":"광고 소재와 메시지를 고객 반응 기반으로 검증하는 방법을 볼 수 있음"}],"normalizedAudience":["창업자·스타트업","프로덕트 매니저·기획자","마케터·그로스"],"summary":"이 영상은 Listen Labs의 Alfred Wahlforss가 AI를 활용해 고객 리서치를 어떻게 대규모로 자동화하는지 설명하는 인터뷰다. 핵심은 설문처럼 얕고 불안정한 응답에 의존하기보다, AI 에이전트가 사람과 직접 대화하며 인터뷰를 수행하고, 그 결과를 분석해 인사이트와 추천을 내놓는 방식이 더 유용하다는 주장이다. 더 나아가 표정·음성·영상까지 함께 보며 실제 감정과 행동에 가까운 신호를 읽고, 모든 데이터 포인트를 추적 가능하게 만들어 AI의 환각 문제도 줄이려 한다.","insights":["고객을 많이 만나는 것보다, 맞는 사람을 먼저 찾는 게 중요하다.","설문은 일관성이 낮고, 사고를 요구하는 인터뷰가 더 안정적이다.","말보다 표정·음성·시선이 실제 의도를 더 잘 드러낸다.","리서치는 추측을 줄이는 도구지, A/B 테스트를 대체하진 않는다.","AI 리서치의 가치는 규모보다 '추적 가능성'에서 커진다."],"keyClips":[{"clipId":"Rumft-rsEu4:c0:1-5","startSegmentIndex":1,"endSegmentIndex":5,"startTime":2.79,"endTime":52.79,"durationSeconds":50,"preview":"정확한 사람 찾기","mustSee":false},{"clipId":"Rumft-rsEu4:c0:11-17","startSegmentIndex":11,"endSegmentIndex":17,"startTime":84.799,"endTime":142.56,"durationSeconds":57.8,"preview":"AI 리서치의 원리","mustSee":true},{"clipId":"Rumft-rsEu4:c0:19-24","startSegmentIndex":19,"endSegmentIndex":24,"startTime":145.68,"endTime":186.8,"durationSeconds":41.1,"preview":"실제 고객 사례","mustSee":false},{"clipId":"Rumft-rsEu4:c0:30-41","startSegmentIndex":30,"endSegmentIndex":41,"startTime":213.28,"endTime":313.039,"durationSeconds":99.8,"preview":"설문 한계와 보완","mustSee":true},{"clipId":"Rumft-rsEu4:c0:45-57","startSegmentIndex":45,"endSegmentIndex":57,"startTime":331.68,"endTime":432.639,"durationSeconds":101,"preview":"표정까지 읽는 인터뷰","mustSee":false},{"clipId":"Rumft-rsEu4:c0:59-65","startSegmentIndex":59,"endSegmentIndex":65,"startTime":436.88,"endTime":478.96,"durationSeconds":42.1,"preview":"창업 계기와 전환","mustSee":false},{"clipId":"Rumft-rsEu4:c0:70-79","startSegmentIndex":70,"endSegmentIndex":79,"startTime":497.039,"endTime":579.76,"durationSeconds":82.7,"preview":"옛 방식의 비용","mustSee":false}],"curatedSegments":[{"segmentIndex":17,"text":"Put it another way, as we get closer to AGI, it will be easier to build things, but the hard part will know what to build and that's what we're building at Listen.","startTime":132.239,"endTime":142.56,"durationSeconds":10,"level":"C1","overallScore":8.8,"rationale":"AGI 시대의 핵심 문제를 압축한다."},{"segmentIndex":37,"text":"I'm confident that in the future you'll still need to have human input because even if you have a per perfect rational being uh like AGI, humans are still irrational totally and they will still want kind of be chaotic in their nature where they all of a sudden get obsessed with a new product, you know, a new Tik Tok trend that shows up and you have to change your entire marketing strategy towards that.","startTime":2142.8,"endTime":2171.44,"durationSeconds":29,"level":"C1","overallScore":8.8,"rationale":"인간 비합리성이 계속 중요하단 통찰."},{"segmentIndex":42,"text":"But if companies are about serving people because that's what that's why we're all working is to help someone else in some way and intelligence gets better and you kind of have what the human wants here and the intelligence is approaching that asmtote then the delta in that asmtote which is what is in a human's mind that isn't in the AI's mind only becomes more important.","startTime":2213.04,"endTime":2233.68,"durationSeconds":21,"level":"C2","overallScore":9,"rationale":"인간 욕구와 AI 한계의 핵심 통찰."},{"segmentIndex":49,"text":"We also have um the network affects the data mode that as we do more interviews you gets better simulation and then the product is very sticky because you have all these interviews in your platform and you don't want to lose that you want to track things over time >> but even like the simplest things I think in terms of product advantages like one of the first things that Brian Sher said one of our partner >> was that founders want to build something that's complex X, but customers want something that's stupid simple and it just works. >> You don't want to lead the witness.","startTime":2269.599,"endTime":2282,"durationSeconds":12,"level":"C1","overallScore":9.2,"rationale":"데이터 해자와 단순성 원칙이 풍부함."},{"segmentIndex":32,"text":"And so I guess I come from the school of thoughts where like actual just telemetry in the real world matters so much more than asking people about what they would do.","startTime":231.519,"endTime":240.48,"durationSeconds":9,"level":"C1","overallScore":7.8,"rationale":"행동 데이터 우선 관점을 선명히 제시."},{"segmentIndex":37,"text":"Um but we did the same thing with listen when you have actually have to think and you have to really reason through your answer and then you're much more consistent um with at least how you answer the same question.","startTime":268.24,"endTime":281.919,"durationSeconds":14,"level":"C1","overallScore":8,"rationale":"대화형 응답의 장점을 논리적으로 설명."},{"segmentIndex":53,"text":"um versus when you might have very high scores on a survey question like that, but when someone also reacts very enthusiastically, it's going to be like perform >> much higher.","startTime":388.56,"endTime":401.36,"durationSeconds":13,"level":"C1","overallScore":8,"rationale":"정서 반응이 성과와 연결됨을 보여준다."},{"segmentIndex":74,"text":"It's actually really hard to know like how do you ask questions to your point that get to um how someone actually will behave.","startTime":529.36,"endTime":537.92,"durationSeconds":9,"level":"C1","overallScore":7.8,"rationale":"좋은 질문 설계의 본질을 짚는다."},{"segmentIndex":75,"text":"You can't just ask like how much are you willing to pay for this? Um there's different methodologies that work better than others to finding the audience.","startTime":537.92,"endTime":545.44,"durationSeconds":8,"level":"C1","overallScore":8,"rationale":"시장조사 질문법의 핵심 교훈이다."},{"segmentIndex":13,"text":"We've had people really open up. It's a very therapeutic experience because it's a non-judgmental entity that's really interested in you.","startTime":669.36,"endTime":678.48,"durationSeconds":9,"level":"B2","overallScore":7.8,"rationale":"AI 대화의 심리적 효과를 잘 설명함."},{"segmentIndex":14,"text":"And we can also have sensitive conversations like interviewing kids um how they react to different products and so I think that's another advantage as well that people can be brutally honest talking to AI.","startTime":678.48,"endTime":686.079,"durationSeconds":8,"level":"B2","overallScore":7.8,"rationale":"민감한 인터뷰의 장점을 구체화함."},{"segmentIndex":22,"text":"Every company is driven by a power law and customer segmentation.","startTime":723.279,"endTime":727.44,"durationSeconds":4,"level":"C1","overallScore":7.6,"rationale":"사업 구조를 일반 원리로 요약한다."},{"segmentIndex":26,"text":"So if you can find that segment, um, the research is so much more actionable.","startTime":749.36,"endTime":755.36,"durationSeconds":6,"level":"B2","overallScore":7.6,"rationale":"세그먼트 선정이 실행력 높임을 말함."},{"segmentIndex":45,"text":"But then the really interesting part is how do you talk to prospective customers, people who also may not be kind of current power law users and how do you compare those two?","startTime":888.72,"endTime":900.399,"durationSeconds":12,"level":"C1","overallScore":8,"rationale":"기존 고객과 잠재 고객 비교의 핵심 통찰."},{"segmentIndex":58,"text":"I think traditional services and being able to then implement these changes is still extremely valuable, but a lot of margins are going to drop and you have to make sure you kind of unbundle a lot of your services to maybe allow for AI agents to help solve some of the problems that you would go to traditional consulting firms before.","startTime":985.44,"endTime":1012.399,"durationSeconds":27,"level":"C1","overallScore":8.2,"rationale":"AI가 서비스 구조를 재편한단 통찰."},{"segmentIndex":76,"text":"And I think what's really exciting is also simulation, which is something we're building now, where you're able to unlock the 99% of use cases where you would never have time to talk to real people.","startTime":1121.039,"endTime":1134.32,"durationSeconds":13,"level":"C1","overallScore":8,"rationale":"시뮬레이션의 확장성을 설득력 있게 설명."},{"segmentIndex":82,"text":"But if they could do it at any time, like in an app on their phone as part of the normal homepage app and give feedback on their EHR or something in the operating room or something along those lines, that seems like a life-saving use case for listen over time.","startTime":1164.559,"endTime":1181.44,"durationSeconds":17,"level":"C1","overallScore":8,"rationale":"구체적 현장 활용상과 효과가 선명함."},{"segmentIndex":12,"text":"So you can essentially try to feed in as much information as possible on a single individual and then in some cases we're able to get 95% accuracy to predict how they will answer certain questions.","startTime":1279.12,"endTime":1290.24,"durationSeconds":11,"level":"B2","overallScore":7.6,"rationale":"구체 수치로 예측 성능을 제시함."},{"segmentIndex":13,"text":"Now the problem becomes things are changing all the time and chaos theory tells us it's really hard to predict the future.","startTime":1290.24,"endTime":1298.32,"durationSeconds":8,"level":"B2","overallScore":7.6,"rationale":"변화성과 예측 한계를 원리로 설명."},{"segmentIndex":18,"text":"So you can imagine a future where you can ask a question and listen like how do software engineers think about cloud code and then listen will say well I already talked to a thousand software engineers this week let me predict how they're going to answer that question but the tricky part is knowing what things can you answer and what can't you answer um because >> and how do you do that?","startTime":1327.6,"endTime":1351.28,"durationSeconds":24,"level":"C1","overallScore":7.8,"rationale":"제품 비전과 한계 지점을 함께 제시."}],"generatedAt":"2026-06-25T00:05:03.748Z","keyClipsTotalSec":934},{"videoId":"Rumft-rsEu4","chunkIndex":1,"totalChunks":5,"title":"Knowing What Your Customers Want, All the Time: Listen Labs' Alfred Wahlforss — Part 2 of 5","thumbnail":"https://i.ytimg.com/vi/Rumft-rsEu4/maxresdefault.jpg","duration":2521,"uploader":"Sequoia Capital","youtubeUrl":"https://www.youtube.com/watch?v=Rumft-rsEu4","keywords":["customer-research","ai-interviews","consumer-insights","market-segmentation","user-research","crm","consulting","business-strategy","product-management","ai-automation"],"normalizedKeywords":["비즈니스·전략","프로덕트","기술 트렌드"],"targetAudience":[{"who":"창업자","why":"고객 이해를 빠르게 넓히는 AI 리서치 활용법을 배울 수 있음"},{"who":"프로덕트 매니저","why":"고객 세분화와 인터뷰 자동화가 의사결정에 주는 효과가 큼"},{"who":"리서처","why":"비동기 인터뷰와 대상자 선별 방식의 구조적 장단점을 이해할 수 있음"},{"who":"투자자","why":"AI가 리서치·컨설팅 시장의 수익 구조를 어떻게 바꾸는지 볼 수 있음"}],"normalizedAudience":["창업자·스타트업","프로덕트 매니저·기획자","투자자·VC"],"summary":"이 영상은 AI 기반 고객 인터뷰 플랫폼이 왜 기존 리서치보다 빠르고 싸며, 때로는 더 솔직한 답을 얻을 수 있는지를 설명한다. 핵심은 고객과 이야기하는 장벽을 낮추면 의사결정의 질이 올라가고, 특히 비동기 방식과 AI의 비판단적 특성이 민감한 주제나 접근이 어려운 집단(예: 아이들, 특정 세그먼트)에서 큰 장점을 만든다는 점이다.\n\n또한 단순히 '고객을 더 빨리 만나는 도구'를 넘어, 매우 구체적인 대상자 풀을 축적하고 검색할 수 있는 데이터베이스가 중요하다고 강조한다. 브랜드의 CRM, 외부 모집, 규제·스팸 문제를 우회하면서도 잠재 고객과 기존 고객을 비교할 수 있고, 나아가 컨설팅과 전통적 리서치의 역할도 AI로 재편될 수 있다는 전망을 제시한다.","insights":["고객과의 대화 장벽을 낮출수록 의사결정 품질이 올라간다.","비동기 인터뷰는 속도·비용·응답 솔직함을 동시에 개선한다.","진짜 가치는 인터뷰 수가 아니라 '맞는 사람'을 찾는 능력이다.","CRM보다 중요한 것은 쓸 수 있는 대상자 데이터베이스다.","AI는 리서치를 대체하기보다 리서치의 빈도와 범위를 폭발시킨다."],"keyClips":[{"clipId":"Rumft-rsEu4:c1:1-15","startSegmentIndex":1,"endSegmentIndex":15,"startTime":601.59,"endTime":693.04,"durationSeconds":91.4,"preview":"빠른 인터뷰의 힘","mustSee":false},{"clipId":"Rumft-rsEu4:c1:16-32","startSegmentIndex":16,"endSegmentIndex":32,"startTime":693.04,"endTime":816.24,"durationSeconds":123.2,"preview":"세그먼트가 핵심","mustSee":true},{"clipId":"Rumft-rsEu4:c1:33-49","startSegmentIndex":33,"endSegmentIndex":49,"startTime":816.24,"endTime":935.44,"durationSeconds":119.2,"preview":"데이터베이스의 가치","mustSee":false},{"clipId":"Rumft-rsEu4:c1:50-68","startSegmentIndex":50,"endSegmentIndex":68,"startTime":935.44,"endTime":1073.557,"durationSeconds":138.1,"preview":"컨설팅의 재편","mustSee":false},{"clipId":"Rumft-rsEu4:c1:69-84","startSegmentIndex":69,"endSegmentIndex":84,"startTime":1073.557,"endTime":1200.88,"durationSeconds":127.3,"preview":"가격과 미래","mustSee":false}],"curatedSegments":[{"segmentIndex":17,"text":"Put it another way, as we get closer to AGI, it will be easier to build things, but the hard part will know what to build and that's what we're building at Listen.","startTime":132.239,"endTime":142.56,"durationSeconds":10,"level":"C1","overallScore":8.8,"rationale":"AGI 시대의 핵심 문제를 압축한다."},{"segmentIndex":37,"text":"I'm confident that in the future you'll still need to have human input because even if you have a per perfect rational being uh like AGI, humans are still irrational totally and they will still want kind of be chaotic in their nature where they all of a sudden get obsessed with a new product, you know, a new Tik Tok trend that shows up and you have to change your entire marketing strategy towards that.","startTime":2142.8,"endTime":2171.44,"durationSeconds":29,"level":"C1","overallScore":8.8,"rationale":"인간 비합리성이 계속 중요하단 통찰."},{"segmentIndex":42,"text":"But if companies are about serving people because that's what that's why we're all working is to help someone else in some way and intelligence gets better and you kind of have what the human wants here and the intelligence is approaching that asmtote then the delta in that asmtote which is what is in a human's mind that isn't in the AI's mind only becomes more important.","startTime":2213.04,"endTime":2233.68,"durationSeconds":21,"level":"C2","overallScore":9,"rationale":"인간 욕구와 AI 한계의 핵심 통찰."},{"segmentIndex":49,"text":"We also have um the network affects the data mode that as we do more interviews you gets better simulation and then the product is very sticky because you have all these interviews in your platform and you don't want to lose that you want to track things over time >> but even like the simplest things I think in terms of product advantages like one of the first things that Brian Sher said one of our partner >> was that founders want to build something that's complex X, but customers want something that's stupid simple and it just works. >> You don't want to lead the witness.","startTime":2269.599,"endTime":2282,"durationSeconds":12,"level":"C1","overallScore":9.2,"rationale":"데이터 해자와 단순성 원칙이 풍부함."},{"segmentIndex":32,"text":"And so I guess I come from the school of thoughts where like actual just telemetry in the real world matters so much more than asking people about what they would do.","startTime":231.519,"endTime":240.48,"durationSeconds":9,"level":"C1","overallScore":7.8,"rationale":"행동 데이터 우선 관점을 선명히 제시."},{"segmentIndex":37,"text":"Um but we did the same thing with listen when you have actually have to think and you have to really reason through your answer and then you're much more consistent um with at least how you answer the same question.","startTime":268.24,"endTime":281.919,"durationSeconds":14,"level":"C1","overallScore":8,"rationale":"대화형 응답의 장점을 논리적으로 설명."},{"segmentIndex":53,"text":"um versus when you might have very high scores on a survey question like that, but when someone also reacts very enthusiastically, it's going to be like perform >> much higher.","startTime":388.56,"endTime":401.36,"durationSeconds":13,"level":"C1","overallScore":8,"rationale":"정서 반응이 성과와 연결됨을 보여준다."},{"segmentIndex":74,"text":"It's actually really hard to know like how do you ask questions to your point that get to um how someone actually will behave.","startTime":529.36,"endTime":537.92,"durationSeconds":9,"level":"C1","overallScore":7.8,"rationale":"좋은 질문 설계의 본질을 짚는다."},{"segmentIndex":75,"text":"You can't just ask like how much are you willing to pay for this? Um there's different methodologies that work better than others to finding the audience.","startTime":537.92,"endTime":545.44,"durationSeconds":8,"level":"C1","overallScore":8,"rationale":"시장조사 질문법의 핵심 교훈이다."},{"segmentIndex":13,"text":"We've had people really open up. It's a very therapeutic experience because it's a non-judgmental entity that's really interested in you.","startTime":669.36,"endTime":678.48,"durationSeconds":9,"level":"B2","overallScore":7.8,"rationale":"AI 대화의 심리적 효과를 잘 설명함."},{"segmentIndex":14,"text":"And we can also have sensitive conversations like interviewing kids um how they react to different products and so I think that's another advantage as well that people can be brutally honest talking to AI.","startTime":678.48,"endTime":686.079,"durationSeconds":8,"level":"B2","overallScore":7.8,"rationale":"민감한 인터뷰의 장점을 구체화함."},{"segmentIndex":22,"text":"Every company is driven by a power law and customer segmentation.","startTime":723.279,"endTime":727.44,"durationSeconds":4,"level":"C1","overallScore":7.6,"rationale":"사업 구조를 일반 원리로 요약한다."},{"segmentIndex":26,"text":"So if you can find that segment, um, the research is so much more actionable.","startTime":749.36,"endTime":755.36,"durationSeconds":6,"level":"B2","overallScore":7.6,"rationale":"세그먼트 선정이 실행력 높임을 말함."},{"segmentIndex":45,"text":"But then the really interesting part is how do you talk to prospective customers, people who also may not be kind of current power law users and how do you compare those two?","startTime":888.72,"endTime":900.399,"durationSeconds":12,"level":"C1","overallScore":8,"rationale":"기존 고객과 잠재 고객 비교의 핵심 통찰."},{"segmentIndex":58,"text":"I think traditional services and being able to then implement these changes is still extremely valuable, but a lot of margins are going to drop and you have to make sure you kind of unbundle a lot of your services to maybe allow for AI agents to help solve some of the problems that you would go to traditional consulting firms before.","startTime":985.44,"endTime":1012.399,"durationSeconds":27,"level":"C1","overallScore":8.2,"rationale":"AI가 서비스 구조를 재편한단 통찰."},{"segmentIndex":76,"text":"And I think what's really exciting is also simulation, which is something we're building now, where you're able to unlock the 99% of use cases where you would never have time to talk to real people.","startTime":1121.039,"endTime":1134.32,"durationSeconds":13,"level":"C1","overallScore":8,"rationale":"시뮬레이션의 확장성을 설득력 있게 설명."},{"segmentIndex":82,"text":"But if they could do it at any time, like in an app on their phone as part of the normal homepage app and give feedback on their EHR or something in the operating room or something along those lines, that seems like a life-saving use case for listen over time.","startTime":1164.559,"endTime":1181.44,"durationSeconds":17,"level":"C1","overallScore":8,"rationale":"구체적 현장 활용상과 효과가 선명함."},{"segmentIndex":12,"text":"So you can essentially try to feed in as much information as possible on a single individual and then in some cases we're able to get 95% accuracy to predict how they will answer certain questions.","startTime":1279.12,"endTime":1290.24,"durationSeconds":11,"level":"B2","overallScore":7.6,"rationale":"구체 수치로 예측 성능을 제시함."},{"segmentIndex":13,"text":"Now the problem becomes things are changing all the time and chaos theory tells us it's really hard to predict the future.","startTime":1290.24,"endTime":1298.32,"durationSeconds":8,"level":"B2","overallScore":7.6,"rationale":"변화성과 예측 한계를 원리로 설명."},{"segmentIndex":18,"text":"So you can imagine a future where you can ask a question and listen like how do software engineers think about cloud code and then listen will say well I already talked to a thousand software engineers this week let me predict how they're going to answer that question but the tricky part is knowing what things can you answer and what can't you answer um because >> and how do you do that?","startTime":1327.6,"endTime":1351.28,"durationSeconds":24,"level":"C1","overallScore":7.8,"rationale":"제품 비전과 한계 지점을 함께 제시."}],"generatedAt":"2026-06-25T00:05:24.968Z","keyClipsTotalSec":934},{"videoId":"Rumft-rsEu4","chunkIndex":2,"totalChunks":5,"title":"Knowing What Your Customers Want, All the Time: Listen Labs' Alfred Wahlforss — Part 3 of 5","thumbnail":"https://i.ytimg.com/vi/Rumft-rsEu4/maxresdefault.jpg","duration":2521,"uploader":"Sequoia Capital","youtubeUrl":"https://www.youtube.com/watch?v=Rumft-rsEu4","keywords":["market-research","generative-ai","customer-insights","synthetic-data","llm","product-intelligence","user-research","startup"],"normalizedKeywords":["비즈니스·전략","프로덕트","기술 트렌드"],"targetAudience":[{"who":"창업자","why":"고객 이해를 제품 기능과 의사결정으로 연결하는 방법을 배울 수 있음"},{"who":"프로덕트 매니저","why":"고객 인터뷰를 대규모로 추상화해 실무 의사결정에 쓰는 방식을 다룸"},{"who":"리서처","why":"인터뷰 기반 시뮬레이션과 검증 방식의 장단점을 이해하는 데 유용함"}],"normalizedAudience":["창업자·스타트업","프로덕트 매니저·기획자","리서처·학자"],"summary":"이 영상은 Listen Labs가 고객 인터뷰 데이터를 활용해 특정 집단의 응답을 시뮬레이션하는 방식을 설명한다. 핵심 아이디어는 일반적인 LLM 프롬프트보다 실제 인터뷰 데이터를 바탕으로 한 'augmented responses'가 더 정확하게 특정 니치의 선호와 판단을 예측할 수 있다는 점이다. 화자는 메시지 테스트, 제목 선택, 제품 의사결정처럼 작은 선택부터 고객 이해를 자동화하는 가능성을 보여주며, 동시에 어떤 질문은 예측 가능하고 어떤 질문은 불가능한지 경계를 명확히 해야 한다고 강조한다.\n\n또한 단순히 프리트레인 데이터나 합성 데이터만으로는 부족하고, 인터뷰처럼 맥락과 행동 질문이 가능한 데이터가 필요하다고 주장한다. 미래에는 이런 시스템이 코딩 에이전트나 다른 에이전트 안에 들어가 '인간 API'처럼 사용자 선호를 호출하는 형태로 확장될 수 있다고 본다. 다만 여러 에이전트가 서로 토론하며 합성하는 방식에는 복잡성 증가와 예측 불가능성에 대한 우려도 함께 제시한다.","insights":["고객 시뮬레이션의 핵심은 평균이 아니라 니치다.","프리트레인보다 인터뷰 데이터가 개인 선호를 더 잘 드러낸다.","예측 성능은 질문을 얼마나 잘 설계하느냐에 달려 있다.","정답을 맞히는 것보다 못 맞히는 범위를 아는 게 중요하다.","고객 이해는 앞으로 에이전트가 호출하는 '인간 API'가 된다."],"keyClips":[{"clipId":"Rumft-rsEu4:c2:3-7","startSegmentIndex":3,"endSegmentIndex":7,"startTime":1207.919,"endTime":1257.36,"durationSeconds":49.4,"preview":"시장조사 3.0 논쟁","mustSee":true},{"clipId":"Rumft-rsEu4:c2:8-21","startSegmentIndex":8,"endSegmentIndex":21,"startTime":1257.36,"endTime":1388.88,"durationSeconds":131.5,"preview":"인터뷰가 모델을 만든다","mustSee":true},{"clipId":"Rumft-rsEu4:c2:23-26","startSegmentIndex":23,"endSegmentIndex":26,"startTime":1393.679,"endTime":1490.96,"durationSeconds":97.3,"preview":"메시지 테스트의 효용","mustSee":false},{"clipId":"Rumft-rsEu4:c2:30-38","startSegmentIndex":30,"endSegmentIndex":38,"startTime":1541.84,"endTime":1629.44,"durationSeconds":87.6,"preview":"왜 인터뷰가 최고인가","mustSee":false},{"clipId":"Rumft-rsEu4:c2:39-47","startSegmentIndex":39,"endSegmentIndex":47,"startTime":1629.44,"endTime":1713.039,"durationSeconds":83.6,"preview":"고객 API의 미래","mustSee":true},{"clipId":"Rumft-rsEu4:c2:49-57","startSegmentIndex":49,"endSegmentIndex":57,"startTime":1717.279,"endTime":1792.96,"durationSeconds":75.7,"preview":"에이전트 합성의 한계","mustSee":false}],"curatedSegments":[{"segmentIndex":17,"text":"Put it another way, as we get closer to AGI, it will be easier to build things, but the hard part will know what to build and that's what we're building at Listen.","startTime":132.239,"endTime":142.56,"durationSeconds":10,"level":"C1","overallScore":8.8,"rationale":"AGI 시대의 핵심 문제를 압축한다."},{"segmentIndex":37,"text":"I'm confident that in the future you'll still need to have human input because even if you have a per perfect rational being uh like AGI, humans are still irrational totally and they will still want kind of be chaotic in their nature where they all of a sudden get obsessed with a new product, you know, a new Tik Tok trend that shows up and you have to change your entire marketing strategy towards that.","startTime":2142.8,"endTime":2171.44,"durationSeconds":29,"level":"C1","overallScore":8.8,"rationale":"인간 비합리성이 계속 중요하단 통찰."},{"segmentIndex":42,"text":"But if companies are about serving people because that's what that's why we're all working is to help someone else in some way and intelligence gets better and you kind of have what the human wants here and the intelligence is approaching that asmtote then the delta in that asmtote which is what is in a human's mind that isn't in the AI's mind only becomes more important.","startTime":2213.04,"endTime":2233.68,"durationSeconds":21,"level":"C2","overallScore":9,"rationale":"인간 욕구와 AI 한계의 핵심 통찰."},{"segmentIndex":49,"text":"We also have um the network affects the data mode that as we do more interviews you gets better simulation and then the product is very sticky because you have all these interviews in your platform and you don't want to lose that you want to track things over time >> but even like the simplest things I think in terms of product advantages like one of the first things that Brian Sher said one of our partner >> was that founders want to build something that's complex X, but customers want something that's stupid simple and it just works. >> You don't want to lead the witness.","startTime":2269.599,"endTime":2282,"durationSeconds":12,"level":"C1","overallScore":9.2,"rationale":"데이터 해자와 단순성 원칙이 풍부함."},{"segmentIndex":32,"text":"And so I guess I come from the school of thoughts where like actual just telemetry in the real world matters so much more than asking people about what they would do.","startTime":231.519,"endTime":240.48,"durationSeconds":9,"level":"C1","overallScore":7.8,"rationale":"행동 데이터 우선 관점을 선명히 제시."},{"segmentIndex":37,"text":"Um but we did the same thing with listen when you have actually have to think and you have to really reason through your answer and then you're much more consistent um with at least how you answer the same question.","startTime":268.24,"endTime":281.919,"durationSeconds":14,"level":"C1","overallScore":8,"rationale":"대화형 응답의 장점을 논리적으로 설명."},{"segmentIndex":53,"text":"um versus when you might have very high scores on a survey question like that, but when someone also reacts very enthusiastically, it's going to be like perform >> much higher.","startTime":388.56,"endTime":401.36,"durationSeconds":13,"level":"C1","overallScore":8,"rationale":"정서 반응이 성과와 연결됨을 보여준다."},{"segmentIndex":74,"text":"It's actually really hard to know like how do you ask questions to your point that get to um how someone actually will behave.","startTime":529.36,"endTime":537.92,"durationSeconds":9,"level":"C1","overallScore":7.8,"rationale":"좋은 질문 설계의 본질을 짚는다."},{"segmentIndex":75,"text":"You can't just ask like how much are you willing to pay for this? Um there's different methodologies that work better than others to finding the audience.","startTime":537.92,"endTime":545.44,"durationSeconds":8,"level":"C1","overallScore":8,"rationale":"시장조사 질문법의 핵심 교훈이다."},{"segmentIndex":13,"text":"We've had people really open up. It's a very therapeutic experience because it's a non-judgmental entity that's really interested in you.","startTime":669.36,"endTime":678.48,"durationSeconds":9,"level":"B2","overallScore":7.8,"rationale":"AI 대화의 심리적 효과를 잘 설명함."},{"segmentIndex":14,"text":"And we can also have sensitive conversations like interviewing kids um how they react to different products and so I think that's another advantage as well that people can be brutally honest talking to AI.","startTime":678.48,"endTime":686.079,"durationSeconds":8,"level":"B2","overallScore":7.8,"rationale":"민감한 인터뷰의 장점을 구체화함."},{"segmentIndex":22,"text":"Every company is driven by a power law and customer segmentation.","startTime":723.279,"endTime":727.44,"durationSeconds":4,"level":"C1","overallScore":7.6,"rationale":"사업 구조를 일반 원리로 요약한다."},{"segmentIndex":26,"text":"So if you can find that segment, um, the research is so much more actionable.","startTime":749.36,"endTime":755.36,"durationSeconds":6,"level":"B2","overallScore":7.6,"rationale":"세그먼트 선정이 실행력 높임을 말함."},{"segmentIndex":45,"text":"But then the really interesting part is how do you talk to prospective customers, people who also may not be kind of current power law users and how do you compare those two?","startTime":888.72,"endTime":900.399,"durationSeconds":12,"level":"C1","overallScore":8,"rationale":"기존 고객과 잠재 고객 비교의 핵심 통찰."},{"segmentIndex":58,"text":"I think traditional services and being able to then implement these changes is still extremely valuable, but a lot of margins are going to drop and you have to make sure you kind of unbundle a lot of your services to maybe allow for AI agents to help solve some of the problems that you would go to traditional consulting firms before.","startTime":985.44,"endTime":1012.399,"durationSeconds":27,"level":"C1","overallScore":8.2,"rationale":"AI가 서비스 구조를 재편한단 통찰."},{"segmentIndex":76,"text":"And I think what's really exciting is also simulation, which is something we're building now, where you're able to unlock the 99% of use cases where you would never have time to talk to real people.","startTime":1121.039,"endTime":1134.32,"durationSeconds":13,"level":"C1","overallScore":8,"rationale":"시뮬레이션의 확장성을 설득력 있게 설명."},{"segmentIndex":82,"text":"But if they could do it at any time, like in an app on their phone as part of the normal homepage app and give feedback on their EHR or something in the operating room or something along those lines, that seems like a life-saving use case for listen over time.","startTime":1164.559,"endTime":1181.44,"durationSeconds":17,"level":"C1","overallScore":8,"rationale":"구체적 현장 활용상과 효과가 선명함."},{"segmentIndex":12,"text":"So you can essentially try to feed in as much information as possible on a single individual and then in some cases we're able to get 95% accuracy to predict how they will answer certain questions.","startTime":1279.12,"endTime":1290.24,"durationSeconds":11,"level":"B2","overallScore":7.6,"rationale":"구체 수치로 예측 성능을 제시함."},{"segmentIndex":13,"text":"Now the problem becomes things are changing all the time and chaos theory tells us it's really hard to predict the future.","startTime":1290.24,"endTime":1298.32,"durationSeconds":8,"level":"B2","overallScore":7.6,"rationale":"변화성과 예측 한계를 원리로 설명."},{"segmentIndex":18,"text":"So you can imagine a future where you can ask a question and listen like how do software engineers think about cloud code and then listen will say well I already talked to a thousand software engineers this week let me predict how they're going to answer that question but the tricky part is knowing what things can you answer and what can't you answer um because >> and how do you do that?","startTime":1327.6,"endTime":1351.28,"durationSeconds":24,"level":"C1","overallScore":7.8,"rationale":"제품 비전과 한계 지점을 함께 제시."}],"generatedAt":"2026-06-25T00:05:55.198Z","keyClipsTotalSec":934},{"videoId":"Rumft-rsEu4","chunkIndex":3,"totalChunks":5,"title":"Knowing What Your Customers Want, All the Time: Listen Labs' Alfred Wahlforss — Part 4 of 5","thumbnail":"https://i.ytimg.com/vi/Rumft-rsEu4/maxresdefault.jpg","duration":2521,"uploader":"Sequoia Capital","youtubeUrl":"https://www.youtube.com/watch?v=Rumft-rsEu4","keywords":["market-research","customer-interviews","ai","vertical-ai","product-strategy","simulation","startup","enterprise-software","human-behavior"],"normalizedKeywords":["비즈니스·전략","프로덕트","기술 트렌드"],"targetAudience":[{"who":"창업자","why":"고객 이해를 제품·전략의 핵심 루프로 보는 관점을 얻을 수 있음"},{"who":"프로덕트 매니저","why":"리서치와 실험을 통해 무엇을 만들지 결정하는 방식이 정리됨"},{"who":"리서처","why":"AI 시대에 인터뷰와 시뮬레이션이 어떻게 바뀌는지 볼 수 있음"}],"normalizedAudience":["창업자·스타트업","프로덕트 매니저·기획자","리서처·학자"],"summary":"이 영상은 Listen Labs가 말하는 차세대 시장조사/고객 리서치의 방향을 다룬다. AI 기반 시뮬레이션이 단순히 인터뷰를 대체하는 것이 아니라, 작은 의사결정에서는 빠른 대안이 되고 큰 의사결정에서는 여전히 인간 인터뷰가 필요하다는 균형 잡힌 관점을 제시한다. 동시에 vertical AI 회사의 경쟁력은 더 좋은 데이터와 더 어려운 문제를 반복해 풀며 전용 평가 지표를 계속 끌어올리는 데 있다고 강조한다.\n\n또한 회사의 본질을 '무엇을 만들지 결정하는 전략 루프'로 보고, 고객의 실제 목소리를 제품 개발과 실행의 입력값으로 삼는 구조를 설명한다. P&G, Tide Pods, M&M's 같은 사례를 통해, 잘한 시장조사는 사람들의 숨은 불편함과 세그먼트를 찾아 브랜드/제품 전략을 바꾸는 힘이 있음을 보여준다. 결론적으로 AI가 발전할수록 사람의 마음속에만 있는 비대칭 정보, 즉 인간적 맥락을 읽는 능력이 더 중요해진다는 메시지다.","insights":["AI가 커질수록 인간 입력의 가치는 오히려 커진다.","작은 의사결정은 시뮬레이션, 큰 결정은 실제 인터뷰가 유리하다.","vertical AI의 해자는 데이터와 평가 지표를 계속 올리는 데 있다.","고객이 원하는 건 복잡한 도구가 아니라 그냥 잘 작동하는 시스템이다.","좋은 시장조사는 숨은 불편과 새로운 세그먼트를 찾아낸다."],"keyClips":[{"clipId":"Rumft-rsEu4:c3:1-10","startSegmentIndex":1,"endSegmentIndex":10,"startTime":1800.95,"endTime":1910.72,"durationSeconds":109.8,"preview":"시뮬레이션의 경계","mustSee":true},{"clipId":"Rumft-rsEu4:c3:11-19","startSegmentIndex":11,"endSegmentIndex":19,"startTime":1910.72,"endTime":1972.32,"durationSeconds":61.6,"preview":"전략 루프의 재편","mustSee":false},{"clipId":"Rumft-rsEu4:c3:20-27","startSegmentIndex":20,"endSegmentIndex":27,"startTime":1972.32,"endTime":2030.799,"durationSeconds":58.5,"preview":"큰 조직의 속도","mustSee":false},{"clipId":"Rumft-rsEu4:c3:28-34","startSegmentIndex":28,"endSegmentIndex":34,"startTime":2030.799,"endTime":2132.96,"durationSeconds":102.2,"preview":"시장조사의 힘","mustSee":true},{"clipId":"Rumft-rsEu4:c3:35-42","startSegmentIndex":35,"endSegmentIndex":42,"startTime":2132.96,"endTime":2233.68,"durationSeconds":100.7,"preview":"인간성의 잔여값","mustSee":false},{"clipId":"Rumft-rsEu4:c3:43-57","startSegmentIndex":43,"endSegmentIndex":57,"startTime":2233.68,"endTime":2372.8,"durationSeconds":139.1,"preview":"버티컬AI의 해자","mustSee":true},{"clipId":"Rumft-rsEu4:c3:58-60","startSegmentIndex":58,"endSegmentIndex":60,"startTime":2372.8,"endTime":2402.56,"durationSeconds":29.8,"preview":"리서치의 미래","mustSee":false}],"curatedSegments":[{"segmentIndex":17,"text":"Put it another way, as we get closer to AGI, it will be easier to build things, but the hard part will know what to build and that's what we're building at Listen.","startTime":132.239,"endTime":142.56,"durationSeconds":10,"level":"C1","overallScore":8.8,"rationale":"AGI 시대의 핵심 문제를 압축한다."},{"segmentIndex":37,"text":"I'm confident that in the future you'll still need to have human input because even if you have a per perfect rational being uh like AGI, humans are still irrational totally and they will still want kind of be chaotic in their nature where they all of a sudden get obsessed with a new product, you know, a new Tik Tok trend that shows up and you have to change your entire marketing strategy towards that.","startTime":2142.8,"endTime":2171.44,"durationSeconds":29,"level":"C1","overallScore":8.8,"rationale":"인간 비합리성이 계속 중요하단 통찰."},{"segmentIndex":42,"text":"But if companies are about serving people because that's what that's why we're all working is to help someone else in some way and intelligence gets better and you kind of have what the human wants here and the intelligence is approaching that asmtote then the delta in that asmtote which is what is in a human's mind that isn't in the AI's mind only becomes more important.","startTime":2213.04,"endTime":2233.68,"durationSeconds":21,"level":"C2","overallScore":9,"rationale":"인간 욕구와 AI 한계의 핵심 통찰."},{"segmentIndex":49,"text":"We also have um the network affects the data mode that as we do more interviews you gets better simulation and then the product is very sticky because you have all these interviews in your platform and you don't want to lose that you want to track things over time >> but even like the simplest things I think in terms of product advantages like one of the first things that Brian Sher said one of our partner >> was that founders want to build something that's complex X, but customers want something that's stupid simple and it just works. >> You don't want to lead the witness.","startTime":2269.599,"endTime":2282,"durationSeconds":12,"level":"C1","overallScore":9.2,"rationale":"데이터 해자와 단순성 원칙이 풍부함."},{"segmentIndex":32,"text":"And so I guess I come from the school of thoughts where like actual just telemetry in the real world matters so much more than asking people about what they would do.","startTime":231.519,"endTime":240.48,"durationSeconds":9,"level":"C1","overallScore":7.8,"rationale":"행동 데이터 우선 관점을 선명히 제시."},{"segmentIndex":37,"text":"Um but we did the same thing with listen when you have actually have to think and you have to really reason through your answer and then you're much more consistent um with at least how you answer the same question.","startTime":268.24,"endTime":281.919,"durationSeconds":14,"level":"C1","overallScore":8,"rationale":"대화형 응답의 장점을 논리적으로 설명."},{"segmentIndex":53,"text":"um versus when you might have very high scores on a survey question like that, but when someone also reacts very enthusiastically, it's going to be like perform >> much higher.","startTime":388.56,"endTime":401.36,"durationSeconds":13,"level":"C1","overallScore":8,"rationale":"정서 반응이 성과와 연결됨을 보여준다."},{"segmentIndex":74,"text":"It's actually really hard to know like how do you ask questions to your point that get to um how someone actually will behave.","startTime":529.36,"endTime":537.92,"durationSeconds":9,"level":"C1","overallScore":7.8,"rationale":"좋은 질문 설계의 본질을 짚는다."},{"segmentIndex":75,"text":"You can't just ask like how much are you willing to pay for this? Um there's different methodologies that work better than others to finding the audience.","startTime":537.92,"endTime":545.44,"durationSeconds":8,"level":"C1","overallScore":8,"rationale":"시장조사 질문법의 핵심 교훈이다."},{"segmentIndex":13,"text":"We've had people really open up. It's a very therapeutic experience because it's a non-judgmental entity that's really interested in you.","startTime":669.36,"endTime":678.48,"durationSeconds":9,"level":"B2","overallScore":7.8,"rationale":"AI 대화의 심리적 효과를 잘 설명함."},{"segmentIndex":14,"text":"And we can also have sensitive conversations like interviewing kids um how they react to different products and so I think that's another advantage as well that people can be brutally honest talking to AI.","startTime":678.48,"endTime":686.079,"durationSeconds":8,"level":"B2","overallScore":7.8,"rationale":"민감한 인터뷰의 장점을 구체화함."},{"segmentIndex":22,"text":"Every company is driven by a power law and customer segmentation.","startTime":723.279,"endTime":727.44,"durationSeconds":4,"level":"C1","overallScore":7.6,"rationale":"사업 구조를 일반 원리로 요약한다."},{"segmentIndex":26,"text":"So if you can find that segment, um, the research is so much more actionable.","startTime":749.36,"endTime":755.36,"durationSeconds":6,"level":"B2","overallScore":7.6,"rationale":"세그먼트 선정이 실행력 높임을 말함."},{"segmentIndex":45,"text":"But then the really interesting part is how do you talk to prospective customers, people who also may not be kind of current power law users and how do you compare those two?","startTime":888.72,"endTime":900.399,"durationSeconds":12,"level":"C1","overallScore":8,"rationale":"기존 고객과 잠재 고객 비교의 핵심 통찰."},{"segmentIndex":58,"text":"I think traditional services and being able to then implement these changes is still extremely valuable, but a lot of margins are going to drop and you have to make sure you kind of unbundle a lot of your services to maybe allow for AI agents to help solve some of the problems that you would go to traditional consulting firms before.","startTime":985.44,"endTime":1012.399,"durationSeconds":27,"level":"C1","overallScore":8.2,"rationale":"AI가 서비스 구조를 재편한단 통찰."},{"segmentIndex":76,"text":"And I think what's really exciting is also simulation, which is something we're building now, where you're able to unlock the 99% of use cases where you would never have time to talk to real people.","startTime":1121.039,"endTime":1134.32,"durationSeconds":13,"level":"C1","overallScore":8,"rationale":"시뮬레이션의 확장성을 설득력 있게 설명."},{"segmentIndex":82,"text":"But if they could do it at any time, like in an app on their phone as part of the normal homepage app and give feedback on their EHR or something in the operating room or something along those lines, that seems like a life-saving use case for listen over time.","startTime":1164.559,"endTime":1181.44,"durationSeconds":17,"level":"C1","overallScore":8,"rationale":"구체적 현장 활용상과 효과가 선명함."},{"segmentIndex":12,"text":"So you can essentially try to feed in as much information as possible on a single individual and then in some cases we're able to get 95% accuracy to predict how they will answer certain questions.","startTime":1279.12,"endTime":1290.24,"durationSeconds":11,"level":"B2","overallScore":7.6,"rationale":"구체 수치로 예측 성능을 제시함."},{"segmentIndex":13,"text":"Now the problem becomes things are changing all the time and chaos theory tells us it's really hard to predict the future.","startTime":1290.24,"endTime":1298.32,"durationSeconds":8,"level":"B2","overallScore":7.6,"rationale":"변화성과 예측 한계를 원리로 설명."},{"segmentIndex":18,"text":"So you can imagine a future where you can ask a question and listen like how do software engineers think about cloud code and then listen will say well I already talked to a thousand software engineers this week let me predict how they're going to answer that question but the tricky part is knowing what things can you answer and what can't you answer um because >> and how do you do that?","startTime":1327.6,"endTime":1351.28,"durationSeconds":24,"level":"C1","overallScore":7.8,"rationale":"제품 비전과 한계 지점을 함께 제시."}],"generatedAt":"2026-06-25T00:06:30.342Z","keyClipsTotalSec":934},{"videoId":"Rumft-rsEu4","chunkIndex":4,"totalChunks":5,"title":"Knowing What Your Customers Want, All the Time: Listen Labs' Alfred Wahlforss — Part 5 of 5","thumbnail":"https://i.ytimg.com/vi/Rumft-rsEu4/maxresdefault.jpg","duration":2521,"uploader":"Sequoia Capital","youtubeUrl":"https://www.youtube.com/watch?v=Rumft-rsEu4","keywords":["user-research","customer-feedback","ai-tools","product-ideation","market-research","simulation","claude","mcp","design-thinking","startup"],"normalizedKeywords":["프로덕트","비즈니스·전략","기술 트렌드"],"targetAudience":[{"who":"프로덕트 매니저","why":"고객 인터뷰와 아이디어 생성이 AI로 어떻게 확장되는지 볼 수 있음"},{"who":"창업자","why":"고객 반응을 빠르게 검증하고 제품 방향을 정하는 감각을 얻을 수 있음"},{"who":"리서처","why":"인터뷰 중 실시간 가설 생성과 시뮬레이션 활용 가능성을 이해할 수 있음"}],"normalizedAudience":["프로덕트 매니저·기획자","창업자·스타트업","리서처·학자"],"summary":"이 대화는 고객 인터뷰와 제품 아이데이션의 경계가 AI로 얼마나 흐려질 수 있는지를 다룬다. 화자는 오늘날에도 이미 AI로 개념 이미지를 만들어 인터뷰에 넣고 있지만, 더 나아가 시뮬레이션과 연동하면 인터뷰 중에도 아이디어를 생성하고 마케팅 방향까지 실시간으로 탐색할 수 있다고 본다.\n\n특히 Claude와 MCP를 연결해 'listen in a loop'처럼 반복적으로 돌리면, 단순히 솔루션 이미지를 보여주는 수준을 넘어 인터뷰 과정 자체가 생성적 탐색 엔진이 될 수 있다고 말한다. 핵심은 리서치가 더 이상 검증만 하는 활동이 아니라, 고객의 언어를 입력으로 즉시 해결책을 발명하는 프로세스로 진화할 수 있다는 점이다.","insights":["고객 인터뷰는 이제 검증뿐 아니라 아이디어 생성의 장이 된다.","AI는 질문에 답하는 도구를 넘어 솔루션을 함께 발명하는 파트너가 된다.","시뮬레이션을 붙이면 인터뷰는 더 빠른 탐색 엔진이 된다.","리서치의 가치가 '듣기'에서 '즉시 가설 생성'으로 확장된다."],"keyClips":[{"clipId":"Rumft-rsEu4:c4:1-4","startSegmentIndex":1,"endSegmentIndex":4,"startTime":2400,"endTime":2432.48,"durationSeconds":32.5,"preview":"리서치의 진화","mustSee":false},{"clipId":"Rumft-rsEu4:c4:5-7","startSegmentIndex":5,"endSegmentIndex":7,"startTime":2432.48,"endTime":2470,"durationSeconds":37.5,"preview":"실시간 공동발명","mustSee":true},{"clipId":"Rumft-rsEu4:c4:9-11","startSegmentIndex":9,"endSegmentIndex":11,"startTime":2473.2,"endTime":2499.119,"durationSeconds":25.9,"preview":"마무리와 인정","mustSee":false}],"curatedSegments":[{"segmentIndex":17,"text":"Put it another way, as we get closer to AGI, it will be easier to build things, but the hard part will know what to build and that's what we're building at Listen.","startTime":132.239,"endTime":142.56,"durationSeconds":10,"level":"C1","overallScore":8.8,"rationale":"AGI 시대의 핵심 문제를 압축한다."},{"segmentIndex":37,"text":"I'm confident that in the future you'll still need to have human input because even if you have a per perfect rational being uh like AGI, humans are still irrational totally and they will still want kind of be chaotic in their nature where they all of a sudden get obsessed with a new product, you know, a new Tik Tok trend that shows up and you have to change your entire marketing strategy towards that.","startTime":2142.8,"endTime":2171.44,"durationSeconds":29,"level":"C1","overallScore":8.8,"rationale":"인간 비합리성이 계속 중요하단 통찰."},{"segmentIndex":42,"text":"But if companies are about serving people because that's what that's why we're all working is to help someone else in some way and intelligence gets better and you kind of have what the human wants here and the intelligence is approaching that asmtote then the delta in that asmtote which is what is in a human's mind that isn't in the AI's mind only becomes more important.","startTime":2213.04,"endTime":2233.68,"durationSeconds":21,"level":"C2","overallScore":9,"rationale":"인간 욕구와 AI 한계의 핵심 통찰."},{"segmentIndex":49,"text":"We also have um the network affects the data mode that as we do more interviews you gets better simulation and then the product is very sticky because you have all these interviews in your platform and you don't want to lose that you want to track things over time >> but even like the simplest things I think in terms of product advantages like one of the first things that Brian Sher said one of our partner >> was that founders want to build something that's complex X, but customers want something that's stupid simple and it just works. >> You don't want to lead the witness.","startTime":2269.599,"endTime":2282,"durationSeconds":12,"level":"C1","overallScore":9.2,"rationale":"데이터 해자와 단순성 원칙이 풍부함."},{"segmentIndex":32,"text":"And so I guess I come from the school of thoughts where like actual just telemetry in the real world matters so much more than asking people about what they would do.","startTime":231.519,"endTime":240.48,"durationSeconds":9,"level":"C1","overallScore":7.8,"rationale":"행동 데이터 우선 관점을 선명히 제시."},{"segmentIndex":37,"text":"Um but we did the same thing with listen when you have actually have to think and you have to really reason through your answer and then you're much more consistent um with at least how you answer the same question.","startTime":268.24,"endTime":281.919,"durationSeconds":14,"level":"C1","overallScore":8,"rationale":"대화형 응답의 장점을 논리적으로 설명."},{"segmentIndex":53,"text":"um versus when you might have very high scores on a survey question like that, but when someone also reacts very enthusiastically, it's going to be like perform >> much higher.","startTime":388.56,"endTime":401.36,"durationSeconds":13,"level":"C1","overallScore":8,"rationale":"정서 반응이 성과와 연결됨을 보여준다."},{"segmentIndex":74,"text":"It's actually really hard to know like how do you ask questions to your point that get to um how someone actually will behave.","startTime":529.36,"endTime":537.92,"durationSeconds":9,"level":"C1","overallScore":7.8,"rationale":"좋은 질문 설계의 본질을 짚는다."},{"segmentIndex":75,"text":"You can't just ask like how much are you willing to pay for this? Um there's different methodologies that work better than others to finding the audience.","startTime":537.92,"endTime":545.44,"durationSeconds":8,"level":"C1","overallScore":8,"rationale":"시장조사 질문법의 핵심 교훈이다."},{"segmentIndex":13,"text":"We've had people really open up. It's a very therapeutic experience because it's a non-judgmental entity that's really interested in you.","startTime":669.36,"endTime":678.48,"durationSeconds":9,"level":"B2","overallScore":7.8,"rationale":"AI 대화의 심리적 효과를 잘 설명함."},{"segmentIndex":14,"text":"And we can also have sensitive conversations like interviewing kids um how they react to different products and so I think that's another advantage as well that people can be brutally honest talking to AI.","startTime":678.48,"endTime":686.079,"durationSeconds":8,"level":"B2","overallScore":7.8,"rationale":"민감한 인터뷰의 장점을 구체화함."},{"segmentIndex":22,"text":"Every company is driven by a power law and customer segmentation.","startTime":723.279,"endTime":727.44,"durationSeconds":4,"level":"C1","overallScore":7.6,"rationale":"사업 구조를 일반 원리로 요약한다."},{"segmentIndex":26,"text":"So if you can find that segment, um, the research is so much more actionable.","startTime":749.36,"endTime":755.36,"durationSeconds":6,"level":"B2","overallScore":7.6,"rationale":"세그먼트 선정이 실행력 높임을 말함."},{"segmentIndex":45,"text":"But then the really interesting part is how do you talk to prospective customers, people who also may not be kind of current power law users and how do you compare those two?","startTime":888.72,"endTime":900.399,"durationSeconds":12,"level":"C1","overallScore":8,"rationale":"기존 고객과 잠재 고객 비교의 핵심 통찰."},{"segmentIndex":58,"text":"I think traditional services and being able to then implement these changes is still extremely valuable, but a lot of margins are going to drop and you have to make sure you kind of unbundle a lot of your services to maybe allow for AI agents to help solve some of the problems that you would go to traditional consulting firms before.","startTime":985.44,"endTime":1012.399,"durationSeconds":27,"level":"C1","overallScore":8.2,"rationale":"AI가 서비스 구조를 재편한단 통찰."},{"segmentIndex":76,"text":"And I think what's really exciting is also simulation, which is something we're building now, where you're able to unlock the 99% of use cases where you would never have time to talk to real people.","startTime":1121.039,"endTime":1134.32,"durationSeconds":13,"level":"C1","overallScore":8,"rationale":"시뮬레이션의 확장성을 설득력 있게 설명."},{"segmentIndex":82,"text":"But if they could do it at any time, like in an app on their phone as part of the normal homepage app and give feedback on their EHR or something in the operating room or something along those lines, that seems like a life-saving use case for listen over time.","startTime":1164.559,"endTime":1181.44,"durationSeconds":17,"level":"C1","overallScore":8,"rationale":"구체적 현장 활용상과 효과가 선명함."},{"segmentIndex":12,"text":"So you can essentially try to feed in as much information as possible on a single individual and then in some cases we're able to get 95% accuracy to predict how they will answer certain questions.","startTime":1279.12,"endTime":1290.24,"durationSeconds":11,"level":"B2","overallScore":7.6,"rationale":"구체 수치로 예측 성능을 제시함."},{"segmentIndex":13,"text":"Now the problem becomes things are changing all the time and chaos theory tells us it's really hard to predict the future.","startTime":1290.24,"endTime":1298.32,"durationSeconds":8,"level":"B2","overallScore":7.6,"rationale":"변화성과 예측 한계를 원리로 설명."},{"segmentIndex":18,"text":"So you can imagine a future where you can ask a question and listen like how do software engineers think about cloud code and then listen will say well I already talked to a thousand software engineers this week let me predict how they're going to answer that question but the tricky part is knowing what things can you answer and what can't you answer um because >> and how do you do that?","startTime":1327.6,"endTime":1351.28,"durationSeconds":24,"level":"C1","overallScore":7.8,"rationale":"제품 비전과 한계 지점을 함께 제시."}],"generatedAt":"2026-06-25T00:06:55.472Z","keyClipsTotalSec":934}]}