zero-to-launch（零启动）

Zero to Launch: Idea → Prototype Playbook

When This Skill Activates

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

Core Frameworks

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

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

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

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)

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

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"

Decision Tree: What to Build 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

Template 1: MVP Scope Definition

# Feature: [Name]

## The ONE Job (Figma Simplicity Test)
What outcome must this deliver?
- Core job: [describe]
- Success = when user can [achieve outcome]

## AI-First Considerations (OpenAI Thinking)
- Could AI help? [yes/no + how]
- Designed for future models? [yes/no]
- Evals needed: [list test cases]

## Complete Experience (Airbnb Approach)
- User journey: [list steps]
- States to design: [loading, error, empty, success]
- Story to tell: "[how we'll talk about this]"

## MVP Scope
**Must Have (Week 1):**
- [feature]
- [feature]

**Should Have (Week 2):**
- [feature]

**Nice to Have (Later):**
- [feature]

Template 2: Build Decision Framework

When deciding what to build:

Quick Reference Card

🏗️ Zero to Launch Checklist

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

Real-World Examples from Episodes

Example 1: OpenAI's ChatGPT Features (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

Example 2: Figma's Feature Development (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

Example 3: Airbnb's Product Redesign (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

Common Pitfalls to Avoid

❌ Mistake 1: Building Without the AI Lens

Problem: Missing 10x opportunities by defaulting to traditional approaches Fix: Always ask: "Could AI make this 10x better?"

❌ Mistake 2: Scope Creep

Problem: "Just one more feature" → bloated MVP that never ships Fix: Ruthlessly apply Figma simplicity test - ONE core job

❌ Mistake 3: Incomplete States

Problem: Shipping without error/empty/loading states Fix: Use Airbnb complete experience checklist

❌ Mistake 4: Feature Factory Mentality

Problem: Building what's requested vs solving jobs Fix: Start with user job, not feature request

Related Skills

• 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

Key Quotes from Episodes

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."

Further Learning

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