Zyra — AI-Powered Women’s Health Companion

Inspiration

Women’s health is often fragmented, confusing, and reactive. From irregular cycles and PCOS to general reproductive health concerns, users are forced to navigate multiple disconnected tools — period trackers, generic search engines, and separate provider directories — none of which truly work together.

What stood out most was the lack of contextual, continuous support. Users don’t just need tracking; they need understanding, guidance, and a clear path to action — especially when deciding whether something is normal or requires medical attention.

Zyra was built to bridge that gap: a single, private, AI-powered companion that helps users track, understand, and act on their health.

What it does

Zyra is an AI-powered women’s health companion that combines:

Cycle tracking — log periods and understand patterns
Health logs — track symptoms, medications, and personal history
AI assistant — provides safe, educational guidance (not diagnosis)
Specialist discovery — find nearby gynecologists and relevant providers
Clinic intelligence (Tinyfish) — analyze provider websites to extract services, relevance (e.g., PCOS, fertility), and booking info
Saved specialists — build a personalized list of providers
Feedback system (Insforge) — users can request topics or report issues
Smart assistant memory (Redis) — improves response speed and context

Zyra doesn’t just track data — it helps users move from uncertainty → clarity → action.

How we built it

Zyra is built as a full-stack SaaS application:

Frontend: Next.js + TypeScript + Tailwind CSS
Authentication & Core Data: Supabase (Auth, Postgres, RLS)
AI Assistant: Groq (LLM inference)
Specialist Discovery: Google Places API

Sponsor Integrations:

Tinyfish — AI web agent for clinic website analysis
Insforge — backend for feedback/topic requests + deployment
Redis (Upstash) — caching + rate limiting for AI assistant
Deployment: Insforge (Vercel-backed infrastructure)

The architecture separates:

Persistent data → Supabase
Real-time intelligence → AI + Redis
External enrichment → Tinyfish

Challenges we ran into

Fragmented data flow - Integrating tracking, AI, and external provider data into a single experience required careful system design.
AI safety & boundaries - Ensuring the assistant guides without crossing into diagnosis was critical.
Third-party API orchestration - Combining multiple APIs while maintaining performance and reliability required robust error handling.
Deployment constraints - Large build artifacts and rate limits require optimization and cleanup strategies.
Privacy-first design - Handling sensitive health data requires strict access control and isolation.