September 1, 2024
Kiddra: child health memory for parents
A calm place to keep vaccines, medications, visits, and notes—plus reminders and plain-language Q&A grounded in *your* records, with clear trust and safety boundaries (not medical advice).
Context
Parents juggle prescription photos, WhatsApp threads, and memory when a pediatric visit gets stressful. The problem isn’t “more apps”—it’s one trustworthy place that reflects what *your* child’s timeline actually contains. Kiddra is built for that: organization and recall, with grounded AI assistance, not a replacement for a clinician.
What we built
Kiddra is a memory for a child’s health story: vaccines, medications, visits, and notes in one system, with calendar and reminders so follow-ups and doses are harder to lose. Caregivers can capture (e.g. prescription photos), get structured summaries and schedules from what they save, and ask questions in plain language with answers anchored in their stored records—with explicit copy that this is not medical advice, only your history, organized (as stated on the public site). The product is private by design: no ads, no selling data, encrypted storage, and user control (including export and delete) framed clearly in the marketing and FAQ. Premium and family tiers (early access) scale history depth, AI usage, and multi-child / shared access for partners or caregivers.
Stack and architecture
I used Next.js for a web-first, mobile-friendly experience, Supabase for secure data and auth, and retrieval-style AI (RAG) so assistant answers are constrained to what the user saved—reducing the worst failure mode in health-adjacent products (confident fiction). A PWA posture supports installability and return visits without forcing an app store before the product is ready for store listings.
Decisions and tradeoffs
In child health, trust and calm are the product. That meant slower feature velocity where copy, consent, and grounded responses needed to come first. The FAQ and positioning make hard lines explicit: no diagnosis, no replacing your doctor, and when AI is uncertain it should read that way. That’s a product decision, not just a legal disclaimer—reducing anxiety is part of the UX spec.
Outcomes and learnings
Shipping Kiddra deepened my respect for domain-sensitive AI: retrieval and careful prompting beat raw chat for this use case. The next phase is about families in production—reminders, calendar density, and multi-caregiver workflows—while keeping the safety and privacy story as obvious on day thirty as on day one.
Back to Work on the homepage, or see all case studies.