- Популярные видео
- Авто
- Видео-блоги
- ДТП, аварии
- Для маленьких
- Еда, напитки
- Животные
- Закон и право
- Знаменитости
- Игры
- Искусство
- Комедии
- Красота, мода
- Кулинария, рецепты
- Люди
- Мото
- Музыка
- Мультфильмы
- Наука, технологии
- Новости
- Образование
- Политика
- Праздники
- Приколы
- Природа
- Происшествия
- Путешествия
- Развлечения
- Ржач
- Семья
- Сериалы
- Спорт
- Стиль жизни
- ТВ передачи
- Танцы
- Технологии
- Товары
- Ужасы
- Фильмы
- Шоу-бизнес
- Юмор
Why 95% of AI Agents Never Ship (and How to Be the 5%) | Agentic AI Execution Ep. 3
95% of AI agents never leave the demo stage. They look great in a prototype — and then collapse under real production pressure.
In this episode, Jyothi Nookula (ex-AWS, Meta, Netflix | 12 patents | 1,500+ PMs coached) breaks down exactly what separates the agents that ship from the ones that don't — and what it means for product managers building in the agentic era right now.
What you'll learn:
→ Why the MIT-cited 95% failure rate happens — and the specific production gaps killing your pipeline
→ How the PM role has fundamentally changed: you now own evals, prototypes, and golden datasets — not just PRDs
→ The "Ferrari vs. Camry" framework for knowing when AI is the wrong tool entirely
→ Why human-in-the-loop is a power design pattern, not a fallback
→ How to manage leadership expectations when your system is probabilistic but they're thinking deterministically
→ Practical tips for building with Claude, MCPs, and skills without blowing your context window
If you're a PM, TPM, or engineering leader trying to move from AI demo to production reality — this one's for you.
⏱️ Chapters
0:00 – The 95% Demo Gap: Why agents fail before they ship
02:20 – Hardening agents: Why this isn't like traditional dev
05:12 – What AI PMs actually own now (hint: it's not just the PRD)
08:30 – Evals deep dive: Happy paths, edge cases, and prompt injection
11:00 – Prototyping with Claude, MCPs, and keeping context lean
15:30 – The Ferrari & the Camry: When NOT to use AI
20:00 – Closing the leadership expectation gap
25:00 – Agentic pipelines vs. traditional pipelines
31:00 – Human-in-the-loop as a trust architecture
🔗 Resources
NextGen Product Manager → https://nextgenproductmanager.com/
Connect with Jyothi → https://www.linkedin.com/in/jyothinookula/
Connect with Leena → https://www.linkedin.com/in/leenaagarwal/
#AgenticAI #ProductManagement #AIEngineering #AIPM #MachineLearning #LLM #TechPodcast #ProductStrategy #AILeadership #ClaudeAI
Видео Why 95% of AI Agents Never Ship (and How to Be the 5%) | Agentic AI Execution Ep. 3 канала Agentic AI Execution
In this episode, Jyothi Nookula (ex-AWS, Meta, Netflix | 12 patents | 1,500+ PMs coached) breaks down exactly what separates the agents that ship from the ones that don't — and what it means for product managers building in the agentic era right now.
What you'll learn:
→ Why the MIT-cited 95% failure rate happens — and the specific production gaps killing your pipeline
→ How the PM role has fundamentally changed: you now own evals, prototypes, and golden datasets — not just PRDs
→ The "Ferrari vs. Camry" framework for knowing when AI is the wrong tool entirely
→ Why human-in-the-loop is a power design pattern, not a fallback
→ How to manage leadership expectations when your system is probabilistic but they're thinking deterministically
→ Practical tips for building with Claude, MCPs, and skills without blowing your context window
If you're a PM, TPM, or engineering leader trying to move from AI demo to production reality — this one's for you.
⏱️ Chapters
0:00 – The 95% Demo Gap: Why agents fail before they ship
02:20 – Hardening agents: Why this isn't like traditional dev
05:12 – What AI PMs actually own now (hint: it's not just the PRD)
08:30 – Evals deep dive: Happy paths, edge cases, and prompt injection
11:00 – Prototyping with Claude, MCPs, and keeping context lean
15:30 – The Ferrari & the Camry: When NOT to use AI
20:00 – Closing the leadership expectation gap
25:00 – Agentic pipelines vs. traditional pipelines
31:00 – Human-in-the-loop as a trust architecture
🔗 Resources
NextGen Product Manager → https://nextgenproductmanager.com/
Connect with Jyothi → https://www.linkedin.com/in/jyothinookula/
Connect with Leena → https://www.linkedin.com/in/leenaagarwal/
#AgenticAI #ProductManagement #AIEngineering #AIPM #MachineLearning #LLM #TechPodcast #ProductStrategy #AILeadership #ClaudeAI
Видео Why 95% of AI Agents Never Ship (and How to Be the 5%) | Agentic AI Execution Ep. 3 канала Agentic AI Execution
Комментарии отсутствуют
Информация о видео
27 апреля 2026 г. 17:00:42
00:27:28
Другие видео канала




















