AgentSkillsCN

Empirical Bottoms-Up Development

基于实证的自下而上研发模式

SKILL.md
--- frontmatter
name: "Empirical Bottoms-Up Development"
name_zh: "Empirical Bottoms-Up Development（实证式自下而上开发）"
category: "strategy-planning"
source: "Lenny's Podcast"
guest: "Alexander Embiricos"

Empirical Bottoms-Up Development

Empirical Bottoms-Up Development（实证式自下而上开发）

概述 / Overview

一种优先采用快速原型（'vibe coding'）和内部“吃狗粮”策略，而非拘泥于僵化长期规划的产品开发理念。它深刻认识到：在 AI 时代，技术能力的更迭速度极快，传统的“先瞄准后开枪”式路线图已不再适用。

来源 / Source

  • •嘉宾: Alexander Embiricos
  • •职位: Product Lead for Codex
  • •公司: OpenAI

核心步骤 / Core Steps

  1. •Hypothesis / Vibe Coding (Prototype)
  2. •Aggressive Dogfooding (Internal Use)
  3. •Empirical Learning (Feedback Analysis)
  4. •Pivot or Scale (Rapid Iteration)

核心原则 / Core Principles

  • •Ready, Fire, Aim: Build prototypes immediately to test feasibility rather than spending months on specs.
  • •Aggressive Dogfooding: Use the tool internally to run the company (e.g., building Atlas or Sora app) to find real friction points.
  • •Fuzzy Long-Term, Sharp Short-Term: Have a vague 1-year vision but hyper-tactical weekly execution based on what works.
  • •Vibe Coding: Designers and PMs write throwaway prototypes using AI to validate ideas before engineering commits.

适用场景 / When to Use

基于快速演进的基础设施（如 LLMs）构建产品，且用户行为呈现涌现性及不可预测性时。

常见错误 / Common Mistakes

在“预备、瞄准”（即打磨战略 PPT）上消耗过多时间，而在“开火”（即上线发布与验证）上投入不足；或者生搬硬套，误以为传统的 PM 框架能直接复用于 AI 领域。

实战案例 / Real-World Example

Sora Android App 由一个微型团队历时 18 天构建完成，通过 Codex 移植 iOS 端逻辑，并在 10 天后正式对外发布。

金句 / Quote

"We can have really good conversations about what's happening in low months or weeks... But there's this awkward middle ground... it's much more important for us to be very humble and learn a lot more empirically."