SAHE is an early-stage AI heritage and hospitality engine for Sarawak, combining structured local knowledge and conversational interfaces to support tourism, cultural visibility, and community-facing services.
01 / About
SAHE draws from curated, community-informed data on Kuching's heritage sites, cultural practices, local hospitality, and visitor logistics — not generic scraped content. The knowledge layer is structured specifically for this geography and its communities.
Users interact through familiar messaging channels — currently WhatsApp and Telegram — without needing to install anything. The AI guide responds in natural language, offering context-aware answers rooted in local knowledge rather than generalized web results.
Beneath the conversational surface is a structured data layer — venues, cultural metadata, service information — stored in a format compatible with geospatial queries, multilingual output, and future integration with government or OTA systems.
02 / Architecture
Venue records, cultural assets, FAQ corpora, and local service data are stored in a structured Supabase-backed layer designed for multilingual retrieval, spatial organization, and future integration.
A RAG-augmented chatflow handles real-time visitor and merchant queries over WhatsApp and Telegram. The guide persona is tuned for local fluency — culturally grounded, multilingual-capable, and designed to avoid generic AI tone.
A lightweight Telegram-based intake flow allows merchant partners to submit venue information in natural language. An LLM parsing workflow normalizes submissions into structured JSON before writing to the data layer — reducing friction for non-technical contributors.
03 / Current Status
SAHE is at the pilot stage. The infrastructure is functional and under active development. The following reflects verified progress as of this build.
An initial set of structured Q&A entries covering Kuching logistics, cultural context, and visitor guidance has been seeded into the knowledge layer and prepared for conversational retrieval.
The primary data schema — covering venues, categories, multilingual fields, spatial coordinates, and embedding columns — is defined and active in a live Supabase environment.
The end-to-end flow from user query to structured response is functional in a controlled demo environment. Active iteration continues on retrieval accuracy, persona calibration, and edge case handling.
The system architecture — spanning ingestion, storage, retrieval, and conversational interaction — has been structured and documented for a scoped pilot and continued iteration.
04 / Why It Matters
Sarawak's cultural landscape — its ethnic diversity, its longhouses, its living traditions, its emerging food and heritage tourism — is poorly represented in mainstream travel platforms. Generic AI assistants have no grounded knowledge of these assets.
SAHE is not a chatbot layered on top of Wikipedia. It is a purpose-built knowledge infrastructure — one that treats local information as a first-class asset, and that puts conversational access in the hands of the communities and visitors who need it.
For government and cultural bodies, this means a sustainable way to make heritage data discoverable and useful. For local businesses, it means reaching visitors in their language, on their platforms, without requiring technical overhead.
Structured local knowledge designed for Sarawak's specific cultural and geographic context — not a generic model applied to a local dataset.
Conversational access over existing channels — reducing the gap between institutional heritage data and the visitor who wants to know what to see tomorrow morning.
An open infrastructure model that supports future integration with government platforms, OTAs, and regional tourism initiatives — built from the ground up in Sarawak.
05 / Contact
Pilot collaboration and demo inquiries welcome. We are actively seeking institutional partners, early merchant pilots, and stakeholder engagement in Sarawak.