Case study · f. 03r
HealthMate
A family health platform that reads uploaded lab reports and medical PDFs and explains them in plain language, tied back to the values in the report.
- Client
- Independent product
- Year
- 2025
- Timeline
- MVP
- Stack
- MongoDB, Express, React, Node.js, LLM APIs
- Family profilesMulti-member
- AccessRole-based
- SummariesTied to report values
The problem
Lab reports are written for doctors. Families get a PDF full of ranges and abbreviations and no idea which numbers matter.
Health data is sensitive, so each family member's reports had to stay private even inside the same family account.
The approach
- 01
Document-first pipeline
Uploaded PDFs are parsed into structured values before any AI step, so summaries refer to real numbers instead of guesses.
- 02
Grounded summaries
Each summary points to the value and range it is talking about. The platform explains, it does not diagnose.
- 03
Privacy by role
Multi-member family accounts with role-based access, so a parent can manage a child's reports without seeing everyone else's.
The outcome
Families get a plain-language view of their reports and can always check it against the original values.
Next case · f. 01rRobotics AI Tutor4.2s → 1.2s response time →