← All Work

Work → Case Studies → Wellness Buddy Thailand — LLM-Powered Wellness Recommender

Wellness Buddy Thailand — LLM-Powered Wellness Recommender

I built Wellness Buddy for a simple reason: Thailand's wellness market is worth billions, but travelers still struggle to choose with confidence.

AI Product Developer & Researcher2024–Present

Overview

I built Wellness Buddy for a simple reason: Thailand's wellness market is worth billions, but travelers still struggle to choose with confidence.

Role: AI Product Developer & Researcher

Timeline: 2024–Present

Tools: Python · NLP · LLMs · Next.js

The Problem

Big Market. Confused Buyer.

Thailand is one of the world's wellness hotspots. The market is worth billions. But for most people, choosing a program still felt like guesswork.

I kept seeing the same pattern: travelers were not choosing the best option. They were choosing the safest-looking one. The one that felt easy to justify.

Meanwhile, 430+ wellness programs were described inconsistently, and 6,400+ reviews sat untouched. The signal was there. The product gap was obvious.

Research

Listen to Behavior, Not Hype

  • I analyzed 6,400+ traveler reviews from Google Maps and TripAdvisor across 165 wellness centers
  • I mapped language patterns into 6 wellness dimensions: physical, mental, spiritual, emotional, social, environmental
  • I found a recurring behavior: travelers describe goals in human terms, but providers market in category terms
  • I published the findings in IEEE ICSEC 2024

Solution & Build

From Research to Real Product

  • I built an NLP pipeline to classify reviews and extract sentiment by wellness dimension
  • I trained an intent-matching model with 90% classification accuracy
  • I added an LLM layer to explain recommendations in plain language people can trust
  • I published extended research in IEICE Transactions 2025
  • I shipped wellnessbuddythailand.com as a live production product

Outcome

Live product serving international wellness travelers in Thailand

2 peer-reviewed publications tied directly to shipped product decisions

End-to-end research-to-product execution completed as MSc thesis

Closed a real decision gap in a multi-billion-dollar tourism category

Learnings

I learned that people do not buy the most accurate recommendation. They buy the one they understand in 10 seconds. Accuracy builds the engine. Trust drives the click.