Building Taiwan's Medical LLM
with H100 Computing Power
Aiii was invited to present at the "Ministry of Digital Affairs · 2026 AI Computing Power Application Kickoff Briefing"
Event Background
- Organizer
- Ministry of Digital Affairs, Digital Industry Administration
- Event
- 2026 AI Computing Power Application Kickoff · Briefing (Taipei)
- Date
- June 25, 2026 (Thu)
- Time
- 10:00–12:00
- Aiii Application Type
- AI Model Training – Container-based
Aiii AI Product Lead Will Cheng (鄭又瑋) was invited to present on "Building Taiwan's Medical LLM," explaining how Aiii leverages national H100 computing resources to upgrade existing AI healthcare products into specialized large language models.
This marks Aiii's first public disclosure of its computing application plan and AI product roadmap at a government-hosted briefing. The Ministry of Digital Affairs operates an AI Computing Sharing Platform (aipower.dginfra.gov.tw), helping domestic AI companies access H100 computing resources to accelerate AI deployment.
AIII'S PRESENTATION
Training Taiwan's Medical LLM with H100 Computing Power
Aiii's Healthcare AI Background
Fortune 500 ClientsAiii has long served the world's leading pharmaceutical companies, including:
- Pfizer, GSK, AstraZeneca, Bayer
- Novartis, MSD, Takeda, Lilly
- Sanofi, Boehringer Ingelheim, and more
Medical LLM Plan
From General LLM → Medical SpecialistGoal: Use H100 computing power to fine-tune a general LLM into Taiwan-specific medical specialty models, covering:
- Parkinson's Disease
- Dermatology
- Diabetes
- Dentistry
- Veterinary Medicine
Existing AI Products
Proven AI ProductsTwo products demonstrated during the presentation:
- Patient Appointment Assistant
AI voice agent auto-searches clinics, confirms availability, and completes bookings - AI Voice-to-Medical-Record
Real-time voice transcription + structured records; veterinary version (VetVoice) already available
FROM THE EVENT — Application Essentials
3 Keys to a Winning Computing Application Proposal
The review committee shared key evaluation criteria on-site. The following is a summary for companies planning to apply for the 2026 AI Computing Platform.
(Source: event record; refer to the official Ministry of Digital Affairs announcement for authoritative information)
Product-First Thinking
Product-FirstDefine "what your product is" clearly before applying — not just how much compute you need. Reviewers care most about:
- Why build this? Is it necessary?
- What is the industry value after completion?
- Is there competitive analysis and differentiation?
(event record; refer to official announcement)
Master Key Evaluation Points
Key Evaluation Points- Summarize objectives in bullet points so non-technical reviewers understand
- Select the AI model precisely to match actual needs
- Provide reasonable estimates for training data (text/audio/images) — quality over quantity
- Describe the container environment and base model used
- Clearly explain the application scenario (not just model training)
Feasibility of Execution Plan
Feasibility & Value- Be specific about expected output in 3 months (not just "train a model")
- Estimate market value or cost savings after completion
- Aim for international market potential, not limited to Taiwan
- Willing to share technology with the government or institutions (bonus points)
EVENT GALLERY
Event Highlights
2026-06-25 Ministry of Digital Affairs Digital Industry Administration · 2026 AI Computing Power Application Kickoff — Taipei
Bring AI Truly into Your Healthcare Setting
Aiii has been dedicated to healthcare AI for over 8 years — from pharma sales automation to Medical LLM training. We look forward to exploring collaboration possibilities with you.
Contact Us