← Thoughts
# Vision A fast, mobile 1‑v‑1 battler where **players invent the meta** by describing creatures and watching them come alive. --- # Three Existential Risks 1. **Creation must feel magical** 2. **Gameplay must stay balanced** 3. **Progression must stay engaging** --- ## 1 · Creation Must Feel Magical <!-- .element: class="text-red-600" --> **Checkpoint:** 80 % of testers rate new creature ≥ 4/5 for “delight”. -- ### Key Tasks - Prompt gallery A/B harness - On‑device LLM latency < 18 s (p95) - Image‑mod pipeline + fallback silhouettes --- ## 2 · Gameplay Must Stay Balanced <!-- .element: class="text-red-600" --> **Checkpoint:** Sim win‑rate spread ≤ 55 / 45 across 1 k decks. -- ### Key Tasks - Dice → mana engine - Arena & trait rules - Headless sim farm + telemetry --- ## 3 · Progression Must Stay Engaging <!-- .element: class="text-red-600" --> **Checkpoint:** D7 retention ≥ 20 % in closed beta. -- ### Key Tasks - CCC / UC currency loop - Upgrade decision tree UX - Early cosmetic hooks --- # 4‑Month Timeline (high‑level) | Month | Primary Outcomes | | ----- | ------------------------------------------------------------- | | **1** | Playable core loop, on‑device model POC, battle HUD wireframe | | **2** | Friends‑&‑family alpha; arena effects; prompt AB testing live | | **3** | Closed‑beta; progression, economy, first balance pass | | **4** | Soft‑launch build; performance polish; app‑store submission | --- ## 16‑Week Gantt (by Work‑Stream) ```mermaid gantt title 2025‑07‑07 → 2025‑10‑27 (16 weeks) dateFormat YYYY‑MM‑DD axisFormat %b‑%d %% ---------- AI --------------- section AI & Generation On‑device LLM POC :ai1, 2025-07-07, 4w Fine‑tune + Moderation :ai2, after ai1, 4w OTA Pipeline :ai3, after ai2, 4w %% ---------- Gameplay --------- section Gameplay & Systems Dice → Mana Engine :gp1, 2025-07-07, 3w Traits & Arena Rules :gp2, after gp1, 5w Balance Pass #1 :gp3, after gp2, 4w %% ---------- UI --------------- section UI & Player XP Battle HUD v1 :ui1, 2025-07-07, 3w Creature Creation Flow :ui2, after ui1, 3w Inventory + Results :ui3, after ui2, 4w %% ---------- Backend ---------- section Backend & Ops Matchmaking MVP :be1, 2025-07-07, 4w Persistence v1 :be2, after be1, 4w Scale & Anti‑Cheat :be3, after be2, 4w %% ---------- Marketing ------- section Marketing & Launch Community Boot‑up :mk1, 2025-07-07, 2w Creator Outreach Wave 1 :mk2, 2025-08-25, 4w Soft‑Launch Campaign :mk3, 2025-09-22, 4w --- # Immediate Next Steps (Sprint 0 → July 21) 1. **Lock V1 scope** — PM + Leads 2. **Dice API & state machine** — Gameplay Eng 3. **Prompt A/B harness live** — ML Eng 4. **Risk register seeded** — All 5. **Device perf test (iPhone 12/14, Pixel 8)** — QA --- # Ground Rules for Prioritisation - **Kill risks first:** tasks that reduce any existential risk outrank polish. - **Dual track:** weekly R&D drops; prod build every two weeks. - **Evidence > opinion:** merge only with metrics (latency, win‑spread, delight score). --- # Open Questions / Decisions Needed - Final art outsource budget? - Matchmaker scale‑test hardware (10 × Android)? - Legal review for user‑generated likenesses? ```