---
tags: [project, idea, active]
created: 2026-09-13
status: 🟡 Concept — pre-validate, not committed
type: side-project / product idea
---

# Plan Maker 2000 — decision-room app

**One-line pitch:** a shareable, ephemeral "decision room" you boot from a group chat when a chat is stuck — options get enumerated, a hectic probability machine underneath resolves it, everyone moves on.

**Origin moment (the problem it solves):** in a WhatsApp chat with your partner and you can't decide on something → move the convo to this. Also framed as "dinner decision maker" / "magic 8 ball". Language used: *"normal chat vibe on top but hectic probability machine underneath."*

## Core insight (the wedge)
Decisions don't have **state** in iMessage/WhatsApp — options, vetoes, constraints, who's paying, "is it actually decided?" all live in scrollback purgatory. A **decision-native surface** (tiny, social, multiplayer decision matrix) is the real product shape. The wedge insight from Kelpie's critique: **build the decision-room, not the chat-app** — a one-off room you boot on demand, not inbox #14.

## Shape
- Shareable link spins up a decision "room"; invites the relevant people; dies in ~24h
- *A moment, not an inbox* — no expectation of network effects or daily DAUs
- Chill, chat-like UI on top; the **reveal** is the payoff

## Feature sketch
- Option enumeration / fill-in (structured, not scrollback)
- Expiry pressure: "must decide by 8pm or suffer default preset"
- **Veto chips:** everyone gets N vetoes, spendable exactly once
- Confidence-weighted / probability voting; bet on outcomes
- Dice roll as the ritual tiebreaker ("the dice have spoken")
- Animated probability-resolution reveal (meme-able, screenshot-able)
- **Asset memory (retention engine):** recurring decisions, saved preferences, ledger of past resolutions ("we always fight about Friday dinner")

## The Jev question
- Jev as final decider = risky: "the AI picked" is a **worse** hard feeling than "the dice picked" (dice: nobody's fault; AI: someone's pick got validated)
- Better role: **AI as negotiator** — propose ranked options from chat history, surface what each side cares about, force clarity ("is the no because of $ or Tuesday?")… then hand the final call to dice/veto. *AI prepares the gunfight, dice fire it.*

## Validation next step
Ship the smallest room (dice UI + enumeration + expiry + veto chips + reveal) to **one couple**; watch whether a WhatsApp group chain lights up on a Saturday night. Market signal = spontaneous re-share.

## Open questions
- Name: "Plan Maker 2000"? Dinner Decision Maker? Magic 8 Ball homage? (brandable room-name matters for share links)
- Where does "move the convo" handoff live: paste-in import from WhatsApp, or manual entry?
- Monetisation shape: free rooms + paid recurring/premium ledger? TBD later — don't block the wedge.
- Jev integration: real-time model access? cost per room?

## Log
- 2026-09-13 Idea raised in chat (WhatsApp-chat hook; dinner/magic-8-ball framing); feedback loop done; direction locked: **decision-room, not chat app**. Note filed in Projects/.


## Handover notes for engineering (prototype brief context) — 13 Sep 2026
Context beyond the feature sketch that will otherwise cause rework:

1. **NO money anywhere.** The concept mixes 'probability', 'voting' and 'betting' language — do NOT gamify with stakes/rewards/payouts. AU gambling regulation applies the moment value changes hands (even credits in some framings). Veto chips = spendable-once counters only, no purchases.
2. **Identity is name-only.** No auth, no accounts, no profiles. A room link + a typed name is the entire identity system. Consequence: no cross-room user history in v1 (asset memory becomes per-device/localStorage until there's an account reason).
3. **'Chat vibe' is a skin, not infrastructure.** Do not build messaging: no DMs, no message history, no notification backend, no thread persistence. It's a sequence of structured state screens (options → chips → resolution → reveal) dressed casually. If a screen looks like a conversation, it's the *layout* borrowing chat tropes, not a chat engine.
4. **State machine is the real spec.** A room = exactly ONE decision with terminal states: expired (default preset applies) OR resolved (dice/veto outcome). Nothing reopens, nothing continues. The ephemerality is the product — resist persistent rooms.
5. **WhatsApp handoff = paste-in.** No API integration (WhatsApp Business API is heavy and off-brand). The flow is: user copies a few messages / types the question and options manually. Design the entry form to make that fast (3 options max suggested, free-text allowed).
6. **Mobile web first.** Share-sheet targets are phones (iOS/Android browser). No native app, no app store, no push. Expiry reminders are in-room only.
7. **The reveal is the product moment.** Budget real animation effort here (probability resolve → ritual line 'the dice have spoken'). Screenshot-ability is a success metric — capture-safe layout.
8. **Jev/LLM negotiator is a stretch goal, not v1.** Dice + enumeration + veto ships first. AI negotiator needs chat-history context that paste-in only partially provides.
9. **Tone guardrail:** Australian-English, irreverent but warm. The 'hard feeling' risk (-bloc policing the method) is the failure state to avoid — veto chips exist FOR this.
10. **Success metric for prototype:** an unsolicited share (someone forwarding the room link after losing the decision). If nobody re-shares, the wedge is wrong, not the UI.