DoctorsInKorea.com:
AI Healthcare Discovery, Built From First Principles

The Challenge

A Fragmented, Language-Locked Market

Korea's healthcare system is genuinely world-class, but discovering the right doctor — especially for international patients and residents navigating NHIS and private insurance — was fragmented across word-of-mouth, expensive concierge services, and clinic websites with no real discovery layer. Clinics, meanwhile, had no structured way to run patient acquisition or growth marketing across Korea's two genuinely distinct search ecosystems, Naver and Google — most platforms entering this market only optimize for one. Medical equipment procurement was still running on email and phone RFQs, with no digital marketplace layer at all.

The Build

One Platform, Four Sides of the Market

Hayya Med AI built DoctorsInKorea.com as a genuinely multi-sided platform rather than a single-purpose directory — each side of the market gets a purpose-built surface, all on shared, RLS-isolated infrastructure:

• ✓An AI Health Navigator that guides patients from symptoms to the right specialty and insurance-matched provider in natural conversation.
• ✓An 11-persona AI growth toolkit for clinics — a named AI receptionist handling WhatsApp/KakaoTalk inquiries, an AI SEO engine, content and marketing automation, analytics, and more — each persona scoped to one job rather than one generic chatbot trying to do everything.
• ✓A B2B procurement marketplace where suppliers list catalogs (an AI engine parses a raw PDF into structured, SEO-ready product pages automatically) and buyers run RFQs with AI-assisted quotation comparison.
• ✓A medical tourism agency portal, letting referral partners route international patients into partner clinics with commission tracking.

The Hard Engineering Problem

Real-Time AI Translation, Solved Properly

The platform's video consultation product lets a patient and clinic speak live, in different languages, with AI translating in real time — a genuinely hard infrastructure problem, not just a chatbot feature. WebRTC media transport runs on managed cloud infrastructure, but the AI translation worker itself needs a full network stack that fully serverless platforms don't provide for real-time media — a constraint we identified and solved by running the translation agent on infrastructure built for persistent, low-latency media processing rather than forcing it onto the wrong tool. Getting this right meant the difference between a demo that works in a controlled test and a feature that works for two real people on a real call.

Going to Market in Two Search Engines at Once

SEO Built for Naver, Not Just Google

Korea is one of the only major markets where the dominant search engine isn't Google — most foreign platforms entering Korea optimize only for Google and simply miss the majority of local search volume. DoctorsInKorea.com's technical SEO was built and verified against both engines simultaneously: structured per-vertical sitemaps, reciprocal hreflang across the bilingual page pairs, and metadata written to satisfy Naver's own (stricter) content-length validation rules alongside Google's. Authenticated portal routes are explicitly excluded from crawling to protect crawl budget for real public content — a discipline many marketplace platforms get wrong.

Compliance as Architecture

Built Around Korea's Real Regulatory Framework

Rather than adapting a generic healthcare platform to Korea, the licensing and accreditation logic was built around Korea's actual regulatory bodies — the Ministry of Health and Welfare for base licensing, the Korean Medical Association for specialist credentials, and the Korea Institute for Healthcare Accreditation for hospital accreditation status. Every patient- and clinic-scoped table is protected by database-enforced row-level security, and AI features are built to operate on internal IDs and conversation content rather than raw patient identity — a real architectural commitment to PIPA-aligned data handling, not a policy statement layered on top afterward.