Principle
Vibe prototyping is deliberate ignorance in service of speed. You temporarily accept shallow data, fake auth, and happy-path flows so you can learn shape before you pay the tax of ownership.
The skill is not prompting — it is knowing what you are allowed not to know yet, and writing that down before the demo seduces you.
The decision
DEC_005
Hypothesis (1 sentence)
│
▼
Time-boxed vibe build (hours–days)
│
▼
Demo to reality ──► falsified? ──yes──► kill or pivot
│ │
│ no
│ ▼
└────────────► sharpen hypothesis / next layerRules of the vibe session
- Name the falsifier first — what observation would change your mind?
- One workflow, not a product — cut nav, cut settings, cut “admin.”
- Fake data on purpose — label it fake in the UI or README so no one confuses seed rows with truth.
- Record the checklist — steps a user takes, not a Loom of you clicking fast.
- End with a retro — what surprised you, what the tool fought, what ticket you’d open in a real repo.
Workflow
- Copy
templates/spec-stub.md— outcome, non-goals, human-owned decisions. - Build the smallest path that tests the falsifier.
- Show one external person (not your co-founder) — watch where they hesitate.
- Log signal: behavior, not praise.
- Either iterate in vibe, run a validation test (Chapter 04), or graduate toward ownership (Chapter 07).
Tooling
Lovable, Bolt, v0, Figma Make — pick by iteration speed for your falsifier, not by Twitter hype. Swap tools freely; keep the hypothesis stable.
Common mistakes
- Polishing UI before the workflow is understood.
- Treating generated CRUD as a data model — it’s a sketch.
- Adding “just one more screen” instead of running the falsifier.
- No time box — vibe sessions that last weeks are stealth product denial.
Artifacts
- Session log: hypothesis · falsifier · time box · signal · next action.
templates/migration-checklist.md— when you approach the leave-prototype gate.
Further reading
- Chapter 07 — When to leave the prototype
- Chapter 00 — Principles (prototype to learn, code to own)