Grimoire: AI Horror Story Game
Endless slow-burn horror, on demand.
AI generates Flanagan-style branching horror games.
01痛点与机会
痛点
Fans lack new personalized horror stories.
为什么是现在
'mike flanagan' searches doubled; AI text matured.
02解决方案与产品
Browser game where AI writes branching horror.
- AI story generation in 30 seconds
- Branching choices, multiple endings
- Content warnings, adaptive scariness
- Text-to-speech audio included
A无人公司 · 零人工运营架构
End-to-end automated; no human in loop.
| 环节 | 全自动实现方式 |
|---|---|
| 获客 | GPT-4o writes SEO articles; Buffer posts social |
| 交付 | User prompt→GPT-4o story→web app; ElevenLabs TTS |
| 客服 | Intercom Fin chatbot; auto refunds |
| 收款 | Stripe subscriptions; auto dunning |
| 运维 | UptimeRobot; Vercel auto-scaling |
| 优化 | User ratings feed A/B tests via VWO |
人工监督（法律最低限度）： Part-time compliance officer for DMCA, monthly audit.
03市场分析
Search volume 500 shows early signal.
04商业模式与定价
Monthly
Unlimited stories
Annual
Unlimited, save 34%
Per-story
One-time
Var cost $0.55/mo/user; margin 94%.
05增长策略
- SEO: AI articles on horror keywords
- Social: auto-post snippets via Repurpose.io
- Referral: free month per referral
- Automated outreach to horror YouTubers
06竞争格局
| 竞争对手 | 我们的优势 |
|---|---|
| AI Dungeon | General, not horror-specialized |
| Choice of Games | Static, not generated |
| Chai | Chat-only, no branching game |
07财务预测（5 年）
| 年度 | 收入 | 付费用户 | EBITDA |
|---|---|---|---|
| Y1 | $24,000 | 200 | -$50,000 |
| Y2 | $72,000 | 600 | -$20,000 |
| Y3 | $144,000 | 1,200 | $10,000 |
| Y4 | $288,000 | 2,400 | $80,000 |
| Y5 | $432,000 | 3,600 | $160,000 |
1% conversion, $9.99/mo, 5% churn, SEO growth
E数据依据与计算
| 关键论断 | 出处 / 计算式 |
|---|---|
| 500 searches, +100% YoY | Keyword Planner data |
| 25.8M horror readers | Pew 10% of 258M US adults |
| 10k visits/mo by Y1 end | 100 kw × 1k × 2% CTR = 2k; 5 clusters |
| 1% visitor→paying | Recurly avg 1-3% for subs |
| $0.55 var cost/user/mo | 10 stories × 3k tok × $0.015 = $0.45 + host |
| LTV $90, CAC $30 | 9 mo × $9.99; paid social CPA $30 |
C合规与公序良俗
合法性
No use of Flanagan name/likeness; original content.
公序良俗
18+ gate, content warnings, no real-person harm.
数据隐私
Encrypted storage, GDPR/CCPA, no PII beyond email.
08风险与对策
| 风险 | 对策 |
|---|---|
| Copyright claims | Originality checker; no specific characters |
| LLM platform outage | Fallback Claude/Gemini |
| Content churn | Novelty via user seeds, daily prompts |
| Payment processor risk | Backup Stripe/Paddle |
09产品路线图
M0-3
MVP: GPT-4o, Stripe, basic web
M4-6
SEO engine, social auto-post
M7-12
TTS audio, referrals, A/B test
Y2
Scale to subgenres, API access
模型: deepseek/deepseek-v4-pro · 查看全部计划书