Stop planning for 2027: how OpenAI builds product 90 days at a time | Tara Sesha and Nan Yu (OpenAI)

Name: Stop planning for 2027: how OpenAI builds product 90 days at a time | Tara Sesha and Nan Yu (OpenAI)
Uploaded: 2026-10-02T00:00:00.000Z
Duration: 1749 s
Channel: Lenny's Podcast
Description: Two OpenAI product leads defend imperfect launches, a two-to-three-month horizon, and quality bars defined by real usage. The binding constraint is organizational absorption, not model capability; forty agents collide with human attention; the new PM skill is writing evals from specific user cases.

LPLenny's Podcast@LennysPodcast

Full transcript

English

Summary：Two OpenAI product leads defend imperfect launches, a two-to-three-month horizon, and quality bars defined by real usage. The binding constraint is organizational absorption, not model capability; forty agents collide with human attention; the new PM skill is writing evals from specific user cases.

Watch on YouTube

Core points (3)

1Ship the imperfect toggle: real user evidence beats polish at billion-user scale.
2Plan two to three months ahead; the constraint is absorption, not capability.
3Bring specific user cases to research and learn to write evals — that is the PM job now.