Back to Details

zero-to-launch

Compare original and translation side by side

🇺🇸

Original

English
🇨🇳

Translation

Chinese

Zero to Launch: Idea → Prototype Playbook

从0到发布：想法→原型手册

When This Skill Activates

技能触发场景

Claude uses this skill when:
  • User asks to "build", "create", or "prototype" a new feature
  • Starting from a product idea or user need
  • Planning MVP or initial scope
  • Making "what to build first" decisions
  • Guiding from concept to code
当出现以下情况时，Claude会启用本技能：
  • 用户提出“构建”“创建”或“制作原型”新功能的需求
  • 从产品想法或用户需求起步
  • 规划MVP或初始功能范围
  • 制定“优先构建什么”的决策
  • 引导从概念阶段推进至代码实现

Core Frameworks

核心框架

1. OpenAI's AI-First Product Development (Source: Kevin Weil, CPO of OpenAI)

1. OpenAI的AI优先产品开发（来源：OpenAI首席产品官Kevin Weil）

The Model Improvement Mindset:
"The AI models you're using today is the worst AI model you will ever use for the rest of your life. Every two months, computers can do something they've never been able to do before."
Use when: Building any product that could benefit from AI capabilities
How:
  1. Design for Future Models: Build assuming models will get 10x better in 2 months
  2. Edge Cases Today = Core Cases Tomorrow: If it barely works now, it'll sing soon
  3. Evals as Product Specs: Write test cases that measure quality, not just functionality
  4. Hybrid Approach: Combine AI + traditional code based on task suitability
Example:
User request: "Build a search feature"

Apply AI-First Thinking:
✅ Could AI understand intent better than keyword matching?
✅ Design for streaming results (models will get faster)
✅ Add eval: "Does search return relevant results for ambiguous queries?"
✅ Hybrid: Use AI for intent, traditional for exact matches
模型迭代思维：
“你如今使用的AI模型，会是你余生中用到的最差的AI模型。每过两个月，计算机就能实现此前无法完成的新功能。”
适用场景： 构建任何可借助AI能力赋能的产品
实施方法：
  1. 为未来模型设计： 以“两个月后模型性能提升10倍”为假设进行构建
  2. 今日边缘场景=明日核心场景： 如今勉强可用的功能，未来会变得极为出色
  3. 以评估用例作为产品规格： 编写衡量质量而非仅功能的测试用例
  4. 混合方案： 根据任务适配性，结合AI与传统代码实现
示例：
User request: "Build a search feature"

Apply AI-First Thinking:
✅ Could AI understand intent better than keyword matching?
✅ Design for streaming results (models will get faster)
✅ Add eval: "Does search return relevant results for ambiguous queries?"
✅ Hybrid: Use AI for intent, traditional for exact matches

2. Figma's Simplicity Forcing Function (Source: Dylan Field, CEO of Figma)

2. Figma的简洁性约束方法（来源：Figma首席执行官Dylan Field）

The Core Question:
"The hardest thing is saying no. We operationalize simplicity by constantly asking: What's the ONE thing that matters here?"
Use when: Feature scope is unclear or growing too large
How:
  1. Identify the Core Job: What's the ONE outcome users need?
  2. Remove Until It Breaks: Strip features until core value disappears
  3. Craft Quality Threshold: Details matter when they create moats
  4. Progressive Disclosure: Hide complexity, reveal when needed
Example:
User request: "Build a dashboard with 15 metrics"

Apply Simplicity Test:
❌ 15 metrics = paralysis
✅ What's the ONE metric that drives action?
✅ Show that metric prominently
✅ Hide other 14 behind "View Details"
✅ Polish the main metric display (craft quality)
核心问题：
“最难的事是学会说不。我们通过不断问自己‘这里最重要的一件事是什么？’来践行简洁性原则。”
适用场景： 功能范围不明确或过度膨胀时
实施方法：
  1. 明确核心目标： 用户需要实现的核心成果是什么？
  2. 做减法直到核心价值受损： 不断剔除功能，直到核心价值无法实现
  3. 设定质量阈值： 能构建竞争壁垒的细节才值得投入
  4. 渐进式展示： 隐藏复杂功能，在需要时再展示
示例：
User request: "Build a dashboard with 15 metrics"

Apply Simplicity Test:
❌ 15 metrics = paralysis
✅ What's the ONE metric that drives action?
✅ Show that metric prominently
✅ Hide other 14 behind "View Details"
✅ Polish the main metric display (craft quality)

3. Airbnb's Complete Experience Design (Source: Brian Chesky, CEO of Airbnb)

3. Airbnb的完整体验设计（来源：Airbnb首席执行官Brian Chesky）

The One Roadmap Philosophy:
"We shifted to one company roadmap. Leaders are in the details. If you don't know the details, how do you know people are doing a good job?"
Use when: Building features that touch multiple parts of the product
How:
  1. Design Complete Flow: Map entire user journey before coding
  2. Consider All States: Loading, error, empty, success states
  3. Cross-Functional From Start: Design, eng, marketing together
  4. Story Over Features: How would you talk about this to customers?
Example:
User request: "Build user onboarding"

Apply Complete Experience:
✅ Map full journey: Signup → Setup → First Value → Habit
✅ Design all states: Loading screens, errors, empty states
✅ Include: Welcome email, in-app guidance, success celebration
✅ Story: "Get your first [outcome] in under 2 minutes"
单一路线图理念：
“我们转向了公司统一路线图。领导者要深入细节。如果你不了解细节，怎么判断工作质量？”
适用场景： 构建涉及产品多个环节的功能时
实施方法：
  1. 设计完整流程： 编码前先梳理完整用户旅程
  2. 考虑所有状态： 加载、错误、空数据、成功状态
  3. 从一开始就跨职能协作： 设计、开发、营销团队共同参与
  4. 以故事而非功能为核心： 你会如何向客户介绍这个功能？
示例：
User request: "Build user onboarding"

Apply Complete Experience:
✅ Map full journey: Signup → Setup → First Value → Habit
✅ Design all states: Loading screens, errors, empty states
✅ Include: Welcome email, in-app guidance, success celebration
✅ Story: "Get your first [outcome] in under 2 minutes"

Decision Tree: What to Build First

决策树：优先构建什么

START: New Feature Idea
│
├─ Can AI 10x this? ─────────────────┐
│  YES: Apply AI-First Framework     │
│  NO: Continue                       ↓
│                                Use OpenAI Patterns
├─ What's the ONE core job? ─────────┤
│  Apply Simplicity Test              │
│  Define: Must-have vs nice-to-have  │
│                                     │
├─ Map complete experience ───────────┤
│  All states, full journey           │
│  Cross-functional considerations    │
│                                     │
└─ BUILD MVP ←───────────────────────┘
   Start with core job
   Add details that create moats
   Ship to small group first
START: New Feature Idea
│
├─ Can AI 10x this? ─────────────────┐
│  YES: Apply AI-First Framework     │
│  NO: Continue                       ↓
│                                Use OpenAI Patterns
├─ What's the ONE core job? ─────────┤
│  Apply Simplicity Test              │
│  Define: Must-have vs nice-to-have  │
│                                     │
├─ Map complete experience ───────────┤
│  All states, full journey           │
│  Cross-functional considerations    │
│                                     │
└─ BUILD MVP ←───────────────────────┘
   Start with core job
   Add details that create moats
   Ship to small group first

Action Templates

行动模板

Template 1: MVP Scope Definition

模板1：MVP范围定义

markdown
undefined
markdown
undefined

Feature: [Name]

Feature: [Name]

The ONE Job (Figma Simplicity Test)

The ONE Job (Figma Simplicity Test)

What outcome must this deliver?
  • Core job: [describe]
  • Success = when user can [achieve outcome]
What outcome must this deliver?
  • Core job: [describe]
  • Success = when user can [achieve outcome]

AI-First Considerations (OpenAI Thinking)

AI-First Considerations (OpenAI Thinking)

  • Could AI help? [yes/no + how]
  • Designed for future models? [yes/no]
  • Evals needed: [list test cases]
  • Could AI help? [yes/no + how]
  • Designed for future models? [yes/no]
  • Evals needed: [list test cases]

Complete Experience (Airbnb Approach)

Complete Experience (Airbnb Approach)

  • User journey: [list steps]
  • States to design: [loading, error, empty, success]
  • Story to tell: "[how we'll talk about this]"
  • User journey: [list steps]
  • States to design: [loading, error, empty, success]
  • Story to tell: "[how we'll talk about this]"

MVP Scope

MVP Scope

Must Have (Week 1):
  • [feature]
  • [feature]
Should Have (Week 2):
  • [feature]
Nice to Have (Later):
  • [feature]
undefined
Must Have (Week 1):
  • [feature]
  • [feature]
Should Have (Week 2):
  • [feature]
Nice to Have (Later):
  • [feature]
undefined

Template 2: Build Decision Framework

模板2：构建决策框架

When deciding what to build:
QuestionFrameworkAction
Is this AI-suitable?OpenAIBuild with AI if: repetitive, pattern-matching, improves over time
What's essential?FigmaStrip to ONE core job, build that first
Is experience complete?AirbnbDesign all states before building any
Can we ship in 1 week?AllIf no, scope is too big - simplify
在决定构建内容时：
问题框架行动
该功能适合AI赋能吗？OpenAI若满足以下条件则采用AI构建：重复性工作、模式匹配类任务、可随时间迭代优化
核心需求是什么？Figma聚焦单一核心目标，优先构建该部分
体验是否完整？Airbnb构建前先设计所有状态
能否在1周内交付？所有框架若不能，说明范围过大 - 需简化

Quick Reference Card

快速参考卡

🏗️ Zero to Launch Checklist

🏗️ 从0到发布检查清单

Before You Code:
  • Defined the ONE core job (Figma Test)
  • Considered AI-first approach (OpenAI Lens)
  • Mapped complete user experience (Airbnb Standard)
  • Identified must-have vs nice-to-have
  • Designed all states (loading, error, empty, success)
During Build:
  • Building for future model improvements (if AI)
  • Maintaining simplicity (saying no to scope creep)
  • Including cross-functional perspectives
  • Crafting details that matter
Before Ship:
  • Core job works end-to-end
  • All states handled
  • Story ready (how to talk about it)
  • Ship to small group first
编码前：
  • 明确单一核心目标（Figma简洁性测试）
  • 考虑AI优先方案（OpenAI思维）
  • 梳理完整用户体验（Airbnb标准）
  • 区分必备功能与锦上添花功能
  • 设计所有状态（加载、错误、空数据、成功）
构建中：
  • 为未来模型迭代预留空间（若涉及AI）
  • 保持简洁性（拒绝范围蔓延）
  • 纳入跨职能视角
  • 打磨关键细节
发布前：
  • 核心目标端到端可用
  • 所有状态均已处理
  • 准备好对外话术
  • 先向小范围用户发布

Real-World Examples from Episodes

来自案例的真实世界示例

Example 1: OpenAI's ChatGPT Features (Kevin Weil)

示例1：OpenAI的ChatGPT功能（Kevin Weil）

Challenge: Users wanted ChatGPT to remember context across conversations
AI-First Approach:
  • Built knowing models would improve memory capabilities
  • Started with basic context, designed for future sophistication
  • Created evals: "Does it remember key facts across sessions?"
  • Hybrid: Explicit memory + AI interpretation
Result: Feature that gets better as models improve
挑战： 用户希望ChatGPT能跨对话记忆上下文
AI优先方案：
  • 基于“模型记忆能力会持续提升”的假设进行构建
  • 从基础上下文功能起步，为未来的复杂功能预留设计空间
  • 制定评估标准：“是否能跨会话记忆关键信息？”
  • 混合方案：显式记忆 + AI解读
结果： 随模型迭代持续优化的功能

Example 2: Figma's Feature Development (Dylan Field)

示例2：Figma的功能开发（Dylan Field）

Challenge: Users requested 50+ features
Simplicity Test Applied:
  • Asked: "What's the ONE thing designers need most?"
  • Answer: Collaboration in real-time
  • Shipped: Multiplayer editing (core job)
  • Deferred: 45+ other requests
Result: Killer feature that defined the product
挑战： 用户提出50+功能需求
应用简洁性测试：
  • 提问：“设计师最核心的需求是什么？”
  • 答案：实时协作
  • 交付：多人实时编辑功能（核心目标）
  • 推迟：45+其他需求
结果： 定义产品核心竞争力的杀手级功能

Example 3: Airbnb's Product Redesign (Brian Chesky)

示例3：Airbnb的产品重设计（Brian Chesky）

Challenge: Rebuilding entire product experience
Complete Experience Approach:
  • One roadmap across all teams
  • Designed full booking journey before building any piece
  • Every feature needed a story (how to talk about it)
  • Leaders stayed in design details
Result: Coherent product, not disconnected features
挑战： 重构整个产品体验
完整体验方案：
  • 全团队采用统一路线图
  • 先设计完整预订流程，再开始构建各个模块
  • 每个功能都需要配套的对外话术
  • 领导者深度参与设计细节
结果： 连贯一致的产品，而非零散的功能集合

Common Pitfalls to Avoid

需避免的常见陷阱

❌ Mistake 1: Building Without the AI Lens

❌ 错误1：未采用AI视角构建

Problem: Missing 10x opportunities by defaulting to traditional approaches Fix: Always ask: "Could AI make this 10x better?"
问题： 因默认采用传统方案而错失10倍提升的机会 解决： 始终问自己：“AI能否让这个功能提升10倍？”

❌ Mistake 2: Scope Creep

❌ 错误2：范围蔓延

Problem: "Just one more feature" → bloated MVP that never ships Fix: Ruthlessly apply Figma simplicity test - ONE core job
问题： “再加一个功能”→ MVP臃肿，无法按时交付 解决： 严格应用Figma简洁性测试 - 聚焦单一核心目标

❌ Mistake 3: Incomplete States

❌ 错误3：状态缺失

Problem: Shipping without error/empty/loading states Fix: Use Airbnb complete experience checklist
问题： 发布时未处理错误/空数据/加载状态 解决： 使用Airbnb完整体验检查清单

❌ Mistake 4: Feature Factory Mentality

❌ 错误4：功能工厂思维

Problem: Building what's requested vs solving jobs Fix: Start with user job, not feature request
问题： 仅构建用户要求的功能，而非解决用户需求 解决： 从用户需求出发，而非功能请求

Related Skills

相关技能

  • strategic-build - For deciding if this is strategic vs tactical work
  • design-first-dev - For detailed craft and quality standards
  • ai-product-patterns - For deep AI implementation patterns
  • ship-decisions - For when to ship vs iterate more
  • jtbd-building - For understanding underlying user jobs
  • strategic-build - 用于判断工作属于战略级还是战术级
  • design-first-dev - 用于细节打磨与质量标准制定
  • ai-product-patterns - 用于深入的AI实现模式
  • ship-decisions - 用于判断何时发布 vs 何时迭代
  • jtbd-building - 用于理解用户底层需求

Key Quotes from Episodes

案例中的关键引用

Kevin Weil (OpenAI):
"Our general mindset is in two months, there's going to be a better model. If you're building and the product is right on the edge of capabilities, keep going. You're doing something right."
Dylan Field (Figma):
"AI makes design, craft, and quality the new moat for startups. The bar for quality is going to go way up."
Brian Chesky (Airbnb):
"Way too many founders apologize for how they want to run the company. What everyone really wants is clarity and to row in the same direction."
Kevin Weil（OpenAI）：
“我们的核心思维是，两个月后就会有更优秀的模型。如果你正在构建的产品刚好触及当前模型的能力边界，别停下，你做的是对的。”
Dylan Field（Figma）：
“AI让设计、工艺和质量成为初创企业的新竞争壁垒。质量门槛会大幅提高。”
Brian Chesky（Airbnb）：
“太多创始人对自己的公司运营方式感到抱歉。大家真正需要的是清晰的方向，以及齐心协力前行。”

Further Learning

拓展学习

For deeper dives on specific topics, see:
  • references/kevin-weil-openai.md - Full AI-first product philosophy
  • references/dylan-field-figma.md - Simplicity and craft standards
  • references/brian-chesky-airbnb.md - Complete experience design methodology
如需深入了解特定主题，请查看：
  • references/kevin-weil-openai.md - AI优先产品理念完整内容
  • references/dylan-field-figma.md - 简洁性与工艺标准
  • references/brian-chesky-airbnb.md - 完整体验设计方法论