montrado.com — Plan

Wedge: the definitive Indonesian-language heritage portal for Montrado, Republik Lanfang, and the 19th-century Hakka kongsi gold-mining confederation of West Borneo. Nothing comparable exists in Bahasa Indonesia — Wikipedia entries are stubs, academic PDFs are paywalled, English Google results recycle five shallow paragraphs. Whoever owns the canonical Indonesian explanation of Lanfang/kongsi/Mandor wins this corner of the SEO map permanently.

1. Why this product wins

• Genuine information gap. "Republik Lanfang" gets ~12k Indonesian searches/month with no satisfying landing page. English coverage is thin too — most readers leave Wikipedia confused about who Low Lan Pak was, where Montrado is on a map, what a kongsi actually did.
• Authority shape, not commerce shape. Heritage portals compound. Every new entry (kongsi, tokoh, peristiwa, tempat) becomes an interlinked node that lifts the others. Domain age starts working for us from day 1.
• Cultural moment. Tionghoa diaspora identity revival (Lunar holiday now national, Cap Go Meh Singkawang going viral on TikTok every February), plus Indonesia's renewed Belt-and-Road interest in Hakka history. Demand curve is going up.
• Defensible by primary sources. We translate Dutch colonial archives, Mandarin chronicles, oral histories from Singkawang/Mempawah. Competitors can't AI-rewrite our way out of source work.

2. Money stack

Heritage portals don't monetize like SaaS. The three layers, ordered by yield-per-effort:

1. Affiliate (books + tours) — Tokopedia/Shopee links to Lanfang books (Heidhues, Yuan Bingling, Mary Somers Heidhues); commission on Singkawang heritage tours via partners (5-15% typical).
2. Premium downloadables (Rp 15-49k via Mayar QRIS) — high-res Lanfang map poster, kongsi timeline infographic, family-tree chart of the 14 kongsi, downloadable EPUB of translated chapters.
3. Donations + grants — Kitabisa for digitization fund, Yayasan Nabil / Tionghoa Indonesia foundations for grant outreach (Year 2+).

Unit economics: ~Rp 0/visitor cost (CF free tier sustains 100k MRR), affiliate clicks convert ~1% at Rp 30k avg commission → Rp 300 RPM. Premium downloads ~0.3% at Rp 25k → Rp 750 RPM. Goal: 50k pageviews/mo by month 6, ≈ Rp 50jt/mo blended.

3. Acquisition triple-stack

• Programmatic SEO — generated entries: /kongsi/<name> × 14 historical kongsi, /tokoh/<name> × ~30 figures, /tempat/<lokasi> × 12 sites, /peristiwa/<year> × ~20 events, /istilah/<term> × 50 Hakka/mining glossary terms. = ~125 indexable pages from a small seed dataset.
• TikTok shorts — 60-second explainer reels ("kenapa Republik Lanfang dianggap republik pertama di dunia?", "Pembantaian Mandor 1944"). Indonesian Gen Z is starved for accessible Tionghoa history content.
• Wikipedia + Wikidata back-references — citations from our well-sourced pages back to Wikipedia entries seed organic referrals for years.

4. AI automation map

• Long-form article drafts → DeepSeek v3.2 via OpenRouter (cheap bulk), then editorial pass by human.
• Source translation (Dutch/Mandarin → ID) → Gemini 2.0 Flash Lite (handles tri-script well, free tier).
• Map illustrations + historical scene art → kie.ai z-image at Rp ~70/img.
• Logo + identity → kie.ai gpt-image-2 (one-time).
• Audio (Bahasa narration of articles for Spotify/podcast distribution) → ElevenLabs v3, only once per article, cached in R2.

5. Cloudflare architecture

Phase 1 (this build): Worker + Assets only — static SSR landing, 8 sub-pages, no D1 yet. Phase 2 (later, when content is ready):

• Worker: montrado-com (this) + montrado-com-api (later, for search + premium downloads)
• D1: montrado-db — tables: kongsi, tokoh, tempat, peristiwa, istilah, articles, downloads, donations
• R2: montrado-assets — heroes, map posters, premium downloadables (gated by signed URLs)
• KV: CACHE for edge-cached article HTML, RATE for Mayar webhook idempotency
• Cron: weekly IndexNow ping, monthly sitemap regen
• Queue: none needed yet
• No DO needed — heritage portal is read-heavy.

6. 90-day execution plan

• Day 1-7 (this skill run): scaffold, landing, 8 sub-pages, brand kit, deploy.
• Day 7-30: write the 14 kongsi entries + 12 place pages (manual + DeepSeek-assisted). Wire D1 + search.
• Day 30-60: flagship long-reads — "Sejarah Lengkap Republik Lanfang", "Pembantaian Mandor 1944", "Cap Go Meh Singkawang: dari Kongsi ke Tarian Tatung". Submit to GSC, IndexNow ping.
• Day 60-90: TikTok account launch (10 shorts), affiliate widget on every article, first Mayar premium download (Lanfang historical map A3 poster Rp 29k).

7. Moat

• Primary-source translation work — slow to copy.
• Inter-page link graph — every new entity entry strengthens 5 others.
• Brand association — once Google "knows" we are the Lanfang authority, dethroning is a 12+ month effort by any newcomer.
• Cultural trust — community engagement (Singkawang Tionghoa elders, Hakka cultural associations) compounds into endorsements/citations.

8. Mandatory standards (CLAUDE.md)

• 100% Cloudflare for production traffic.
• Hero images via kie.ai z-image, piped through xmoj.com for webp.
• Logo via kie.ai gpt-image-2 (NOT handdrawn/sketch).
• Mobile-native UX: bottom-tab nav, horizontal sliders, safe-area-inset, ≥44px tap targets.
• All "links" use .btn system, no plain underlined nav.
• Lucide-style inline SVG sprite, NEVER emoji as functional UI.
• AA contrast — use the -text darker variant for any bright accent on light bg.

9. Competitor synthesis

• id.wikipedia.org/wiki/Republik_Lanfang — stub, ~600 words, poorly sourced. We outflank by length, depth, and inline maps.
• en.wikipedia.org/wiki/Lanfang_Republic — better but English-only and confused about Montrado's role vs Mandor's.
• Academia.edu PDFs (Heidhues, Yuan Bingling) — gold standard but locked behind paywalls + login walls. We summarize + cite (fair use).
• Singkawang tourism blogs — focus on Cap Go Meh tourism, ignore mining heritage. Different audience.
• No direct competitor exists for Indonesian-language Hakka mining heritage authority. This is wide open.

10. Missing parts audit (this skill run is Phase 1)

• D1 schema + search API → Phase 2
• Mayar QRIS integration for premium downloads → Phase 2
• TikTok account + first 10 shorts script → Phase 3
• Email capture (Resend) — already covered by shared bijaksana.org noreply
• Affiliate tracking helper (UTM-tagged Tokopedia/Shopee links) → Phase 2
• Sitemap auto-regen cron → Phase 2

11. Domain naming context

• Montrado is a real town in Bengkayang Regency, West Kalimantan (~70 km from Singkawang).
• It was the capital of the Thaikong (Dagang) kongsi and the second-largest Chinese settlement in West Borneo during 1820s-1850s.
• The "Kongsi Wars" between Dutch forces and Thaikong/Samtiaokioe alliance (1850-1854) centered on Montrado.
• Choosing "montrado.com" over "lanfang.com" / "kongsi.com" is deliberate: harder for casual outsiders to copy, signals deep insider knowledge, and Montrado-as-geographic-anchor lets us own the broader West Kalimantan heritage map.

⚙ HARD CONSTRAINTS (enforced for all sites)

This domain MUST operate within these constraints — no exceptions:

• 100% Cloudflare serverless — Workers + D1 + R2 + KV + Workers AI + Vectorize. NEVER PM2, NEVER VPS, NEVER Docker in production path.
• 100% AI-automated — every customer interaction, every moderation decision, every transaction reconcile = AI. No manual queue, no live human chat support, no physical fulfillment.
• 1-operator solo — one person can run the entire operation from a phone. No team meetings, no shared inbox, no shift rotation.
• WhatsApp AI bot for all support (24/7, instant response, no SLA promises that need humans).
• Mayar QRIS for all Indonesian payments (subscription auto-renew, no manual invoicing).
• Indonesian UI primary — bahasa-first, English fallback only where unavoidable.
• Privacy — opt-in only, delete-on-request honored within 24h (cron-driven).
• No physical goods, no inventory — digital products + affiliate referrals only.

If the plan above describes any flow that violates these constraints, treat the plan as ASPIRATIONAL only and rework before building. The constraint trifecta wins.