Vaccine Genie is a product and platform that converts fragmented, multilingual vaccination records into reliable digital profiles through human-AI collaboration. AI performs extraction, translation, and structuring, while users provide final review and approval to maintain trust and safety.
Vaccination records are often needed for immigration, school enrollment, employment, and travel. However, many exist only on paper and in multiple languages, which makes verification slow and error-prone. Digitization was necessary, but fully automated medical data processing could lead to harmful mistakes.
After uploading a document, users did not understand:
what information was extracted,
what needed their attention,
or how confident the system was.
From interviews and market research, I learned:
Users care more about what AI did than how advanced it is.
Trust increases when the original document remains visible.
Unexplained errors feel more threatening than expected inaccuracies.
Users want guidance on what to check first.
Transparency and responsibility are more important than speed.
The core problem was not data capture, but the lack of a clear verification experience that defined AI confidence and human responsibility.
Redesigned the information architecture to boost verification efficiency and confidence.
The new flow prioritizes review context over raw data, using AI extraction with confidence scoring to organize and pre-fill key data. This approach significantly reduces manual input and accelerates record finalization and storage.
I conducted three rounds of usability testing to evaluate how well the core features were understood and used. Additional user testing will be required to support future AI evaluation and refinement.
I redesigned the review experience to make collaboration explicit and actionable.
Each iteration focused on reducing ambiguity around AI outputs while preserving user agency in high-stakes decisions.
Automation vs. Control
We limited full automation to ensure users retained final authority over sensitive records.Speed vs. Accuracy
Faster approvals were enabled for high-confidence fields, while uncertainty intentionally introduced friction.Seamless Sharing vs. Privacy Protection (HIPAA Compliance)
Prioritizing HIPAA-compliant consent and visibility added minor friction, but ensured safe, trustworthy health data sharing.
The final model positions AI as an organizer and accelerator, while keeping humans accountable for all critical decisions.
The outcome was not faster automation, but a safer and more guided health record management experience.
I extended Vaccine Genie to the web, enabling users to upload a wide range of file formats and automatically extract structured information. This approach preserved automation benefits while maintaining design consistency across platforms, ensuring that AI-assisted data intake felt predictable, trustworthy, and scalable as the product expanded beyond mobile.
