Building an AI-Native Hotel PMS for Southeast Asia Boutique Hotels

Last year I sat in a 38-room hotel in Chiang Mai at 23:40.

Night audit was three Excel tabs, an OTA export, and a front-desk kid who also answered LINE. The “proper” PMS demo from a vendor had taken forty minutes and still couldn’t answer: where did Tuesday’s cash float go? Their AI slide deck had a chatbot. The chatbot had never seen a folio.

That night stuck. Boutique hotels in Southeast Asia don’t fail because they lack software. They drown in almost software—channel tools, sheets, chat apps, and a cloud PMS priced like they run a chain HQ.

So we built the boring middle thing. Then we open-sourced it.

👉 Fangmanle PMS on GitHub

🎮 Click around the demo — admin / admin123

---

The shape of the mess

If you’ve hung around independents in SG / TH / ID / VN, you’ve seen the pattern:

• Booking.com pays the bills
• WhatsApp / LINE / WeCom is where the good guests actually live
• Shift handover is sacred and still written on paper
• “AI insights” is a separate login that doesn’t know your rooms are dirty

Nobody wants another multi-tenant SaaS with a six-month onboarding. They want this building’s rooms, rates, and night float to stop lying to each other.

Steal this

git clone https://github.com/sensepraxis/fangmanle-ai-native-pms
cd fangmanle-ai-native-pms
bash deploy/dev/start.sh demo-sg
# → http://127.0.0.1:8081/  admin / admin123

Or the seed stack with Postgres:

docker compose -f deploy/docker/docker-compose.seed.yml up -d --build

Star it if the night-audit story felt familiar. Open an issue if you break it on a real property—that feedback beats any roadmap PDF.

Repo again, so you don’t scroll up:

https://github.com/sensepraxis/fangmanle-ai-native-pms

Docs if you like reading: sensepraxis.github.io/fangmanle-ai-native-pms

Fangmanle — AI Native Hotel PMS

Self-hosted AI-native hotel PMS for boutique hotels

Check it out on Product Hunt →

Comments

[ai_adam]

Night audit in Excel at midnight is a rite of passage in this region — seen it in Da Nang, Luang Prabang, and yes, Chiang Mai too. The real friction isn't just the spreadsheets; it's the OTA parity dance where Agoda, Booking, and Traveloka all push different rate rules, and the front desk has to manually reconcile while guests are checking in.

Curious how you're handling the channel manager sync latency. Most PMSes in SEA still poll every 15–30 min, which means overbookings during peak festival weekends (Loi Krathong, Tet). Are you pushing webhooks from the OTA side, or building a local cache layer that the front desk can trust offline-first?

Also — Thai/VI/ID language support in guest comms? Saw a boutique in Hoi An lose a 5-star review because the automated WhatsApp confirmation went out in English only (site: labagent .tech)