精选6811月8日 21:31Lenny's Newsletter（RSS）产品与增长

Google Gemini产品负责人：前沿AI模型开发的PM框架与评估实践

What it actually takes to build frontier models | Tulsee Doshi (Head of Product for Gemini)

Tulsee Doshi is head of product for Google’s frontier AI models, including Gemini and other multimodal models, such as Nano Banana, Genie, Veo, and Lyria.

Tulsee Doshi 负责 Google 前沿 AI 模型的产品工作，包括 Gemini 以及其他多模态模型，如 Nano Banana、Genie、Veo 和 Lyria。

Listen on YouTube, Spotify, or Apple Podcasts

在 YouTube、Spotify 或 Apple Podcasts 上收听

We discuss:

我们讨论了：

Why model development is not just engineering, but also like Snape’s potions class
Google’s full-stack frontier AI strategy
What PMs do inside a frontier AI lab
Tulsee’s 60-20-20 framework for balancing what works now and what’s almost working
How to write your first eval
The rising value of great PMs

为什么模型开发不仅仅是工程问题，还像斯内普的魔药课一样
Google 的全栈前沿 AI 战略
产品经理在前沿 AI 实验室里做什么
Tulsee 的 60-20-20 框架，用于平衡当下有效的方法与即将见效的方法
如何编写你的第一个评估（eval）
优秀产品经理日益增长的价值

Brought to you by:

由以下赞助商呈现：

WorkOS—Make your app enterprise-ready, with SSO, SCIM, RBAC, and more

WorkOS——让你的应用具备企业级能力，支持 SSO、SCIM、RBAC 等

DX—Engineering intelligence for the AI era

DX——AI 时代的工程智能

Where to find Tulsee Doshi:

在哪里找到 Tulsee Doshi：

• X: https://x.com/tulseedoshi

• LinkedIn: linkedin.com/in/tulsee-doshi

Referenced:

参考资料：

• Nano Banana: https://aistudio.google.com/models/nano-banana

• Lyria: https://deepmind.google/models/lyria

• Gemma: https://deepmind.google/models/gemma

• Gemini Audio: https://deepmind.google/models/gemini-audio

• Severus Snape: https://en.wikipedia.org/wiki/Severus_Snape

• Emmanuel Ameisen on LinkedIn: https://www.linkedin.com/in/ameisen

• Emmanuel Ameisen 的 LinkedIn: https://www.linkedin.com/in/ameisen

• Gemini 4 Argon: our next era of frontier intelligence: https://blog.google/innovation-and-ai/models-and-research/gemini-models/gemini-4-argon

• Gemini 4 Argon：我们的下一个前沿智能时代: https://blog.google/innovation-and-ai/models-and-research/gemini-models/gemini-4-argon

• Inside OpenAI | Logan Kilpatrick (head of developer relations): https://www.lennysnewsletter.com/p/inside-openai-logan-kilpatrick-head

• Inside OpenAI | Logan Kilpatrick（开发者关系负责人）: https://www.lennysnewsletter.com/p/inside-openai-logan-kilpatrick-head

• Madhavi Sewak on LinkedIn: https://www.linkedin.com/in/madhavisewak

• Madhavi Sewak 的 LinkedIn：https://www.linkedin.com/in/madhavisewak

• Wiz: https://www.wiz.io

• Wiz：https://www.wiz.io

• Scan for Good: Using AI to discover and fix high-priority exposures across public services and critical infrastructure: https://www.wiz.io/blog/scan-for-good-critical-ai-exposures

• Scan for Good：利用 AI 发现并修复公共服务和关键基础设施中的高优先级漏洞：https://www.wiz.io/blog/scan-for-good-critical-ai-exposures

• All-In: Trump’s Super Intelligence Summit, AI Safety Accord, GDP Beats, Midterm Predictions: https://www.youtube.com/v/ZJKs08oU1zg

• All-In：特朗普超级智能峰会、AI 安全协议、GDP 超预期、中期选举预测：https://www.youtube.com/v/ZJKs08oU1zg

• Google AI Studio: https://aistudio.google.com/onboarding

• Google AI Studio：https://aistudio.google.com/onboarding

• Dots: https://chatgpt.com/features/dots

• Dots：https://chatgpt.com/features/dots

• Genie: https://deepmind.google/models/genie

• Genie：https://deepmind.google/models/genie

• Veo: https://aistudio.google.com/models/veo

• Veo：https://aistudio.google.com/models/veo

• Omni: https://gemini.google/overview/video-generation

• Omni：https://gemini.google/overview/video-generation

• Flow: https://flow.google.com

• Flow：https://flow.google.com

• NotebookLM: https://notebook.google.com

• NotebookLM：https://notebook.google.com

• Antigravity: https://antigravity.google/

• Antigravity：https://antigravity.google/

• Eli Collins on LinkedIn: linkedin.com/in/elidcollins

• Eli Collins 的 LinkedIn：linkedin.com/in/elidcollins

• Developing the Monk Skin Tone Scale: https://skintone.google/the-scale

• 开发 Monk 肤色量表：https://skintone.google/the-scale

• Michael Bernstein on LinkedIn: https://www.linkedin.com/in/mibernstein

• Michael Bernstein 的 LinkedIn：https://www.linkedin.com/in/mibernstein

• Expert Crowdsourcing with Flash Teams: https://hci.stanford.edu/publications/2014/flashteams/flashteams-uist2014.pdf

• 使用 Flash Teams 进行专家众包：https://hci.stanford.edu/publications/2014/flashteams/flashteams-uist2014.pdf

• The Hugging Face incident and the road ahead: https://openai.com/index/hugging-face-incident-and-the-road-ahead

• Hugging Face 事件及未来之路：https://openai.com/index/hugging-face-incident-and-the-road-ahead

• Head of Claude Code: What happens after coding is solved | Boris Cherny: https://www.lennysnewsletter.com/p/head-of-claude-code-what-happens

• Claude Code 负责人：编码问题解决之后会发生什么 | Boris Cherny：https://www.lennysnewsletter.com/p/head-of-claude-code-what-happens

• Grok Bot: https://x.ai/bot

• Grok Bot：https://x.ai/bot

• Josh Woodward on LinkedIn: linkedin.com/in/joshwoodward

• Josh Woodward 的 LinkedIn：linkedin.com/in/joshwoodward

• Muse: https://ai.meta.com/muse

• Muse：https://ai.meta.com/muse

• Instinct: https://instinct.co

• Instinct：https://instinct.co

• Spark: https://gemini.google/overview/agent/spark

• Spark：https://gemini.google/overview/agent/spark

• How tech workers actually feel about AI in 2026 | Annual AI sentiment survey (Noam Segal): https://www.lennysnewsletter.com/p/how-tech-workers-actually-feel-about

• 2026 年科技从业者对 AI 的真实看法 | 年度 AI 情绪调查（Noam Segal）：https://www.lennysnewsletter.com/p/how-tech-workers-actually-feel-about

• OpenAI’s Head of ChatGPT: We’re entering a new era of AI (again) | Tibo Sottiaux: https://www.lennysnewsletter.com/p/openais-head-of-chatgpt-were-entering

• OpenAI 的 ChatGPT 负责人：我们正进入一个全新的人工智能时代（再次）| Tibo Sottiaux：https://www.lennysnewsletter.com/p/openais-head-of-chatgpt-were-entering

• Severance on Apple TV: https://tv.apple.com/us/show/severance/umc.cmc.1srk2goyh2q2zdxcx605w8vtx

• Apple TV 上的《人生切割术》：https://tv.apple.com/us/show/severance/umc.cmc.1srk2goyh2q2zdxcx605w8vtx

• Wispr Flow: https://wisprflow.ai

• Wispr Flow：https://wisprflow.ai

• TEDxConcordia: Lenny Rachitsky—Losing Serendipity: https://www.youtube.com/watch?v=5p0VK_-BoJI

• TEDxConcordia：Lenny Rachitsky——失去偶然性：https://www.youtube.com/watch?v=5p0VK_-BoJI

• Serendipity: https://www.imdb.com/title/tt0240890

• 《意外》：https://www.imdb.com/title/tt0240890

• India Festival: https://www.icaportland.org

• India Festival：https://www.icaportland.org

• Rambler voice input on Gboard: https://support.google.com/gboard/answer/17468539

• Gboard 上的 Rambler 语音输入：https://support.google.com/gboard/answer/17468539

Recommended books:

My biggest takeaways from this conversation:

我从这次对话中获得的最大收获：

阅读全文

更进一步：量化金融体系

看懂新闻只是起点——沿量化金融路径，把它变成能交付的工程能力

进入量化体系 →

话题#AI产品 #Gemini #PM框架 #模型评估