# Prototype Pipeline

## Purpose

The prototype pipeline defines how ideas move from loose capture to structured learning and possible promotion.

## Stage 0 — Raw Capture

Capture ideas without judging them immediately.

Location:

- `inbox/`
- rough notes
- conversations
- sketches

Done when the idea is saved somewhere and no longer needs to be held in memory.

## Stage 1 — Triage

Decide whether an idea deserves a prototype card.

Questions:

- What is the idea?
- Who might care?
- What problem, desire, curiosity, or opportunity does it address?
- What is the smallest way to test it?
- Is there an obvious reason to park it?

Outcomes:

- Create prototype card.
- Park.
- Merge with another idea.
- Reject.

## Stage 2 — Prototype Card

Turn the idea into a structured prototype candidate.

Required fields:

- Name.
- One-line pitch.
- Target user or audience.
- Learning question.
- Smallest useful test.
- Expected signal.
- Promotion target if successful.

Location:

- `prototypes/`

## Stage 3 — Experiment

Test the idea with minimal cost.

Possible experiment types:

- Concept note.
- Landing page.
- Clickable mockup.
- CLI/demo script.
- Wizard-of-Oz process.
- Manual concierge test.
- Closed conversation.
- Zulip discussion.
- Private beta.
- Paid pre-order test, only when legally and commercially safe.

## Stage 4 — Signal Review

Evaluate what was learned.

| Signal Type | Examples |
|---|---|
| Interest | People ask for more, subscribe, volunteer, comment |
| Usefulness | People use it to solve something real |
| Retention | People return after first exposure |
| Referral | People tell others |
| Payment | People pay, pre-order, or ask for pricing |
| Contribution | People offer work, content, code, or feedback |
| Strategic Fit | It strengthens Helix, Coulomb, Sloppers, Plenitude, or Binky |

## Stage 5 — Decision

Possible decisions:

- Park.
- Iterate.
- Promote.
- Reject.
- Merge into another initiative.

Promotion requires an explicit decision record in `DECISIONS.md`.