SyncTrack

Inspiration

Yasmeen has been able to maximize her own productivity by organizing her calendar in-line with her own hormonal cycles. After practicing this for a month and finding a spike in her productivity, she thought that many women can also benefit from doing the same.

What it does

The calendar tracks hormonal phases for a woman according to her period (and how she feels at points during her period), while also allowing the user to enter tasks to our agent that they need to accomplish. The agent will then organize the calendar in a way to maximize efficiency for the female user to complete their tasks during the best-fit hormonal periods throughout the month.

How we built it

Frontend Stack Next.js 15 (App Router) → Modern React framework with server-side rendering ├── TypeScript → Type safety and better developer experience ├── TailwindCSS → Utility-first styling with custom cycle phase colors ├── FullCalendar → Interactive calendar with day/time grid views ├── React Hot Toast → Non-intrusive notifications └── Date-fns → Date manipulation and formatting

AI & Backend Integration AWS Bedrock Agent (Claude 3.5 Sonnet) → Core AI reasoning engine ├── Knowledge Base → RAG with menstrual cycle research papers ├── Action Groups → Lambda functions for calendar operations ├── Session Management → Maintains context across interactions └── Structured Output → JSON responses for calendar integration

Data Flow Architecture User Input → Next.js API Route → Bedrock Agent → Knowledge Base + Action Groups → AI Response → Calendar Update

Challenges we ran into

The biggest challenge we found was creating an accurate knowledge base for our agent, as well as dealing with edge cases (such as if the female user is out of rhythm with what is considered a normal hormonal phase). We solved both cases through allowing user input to have appropriate weighting for the agent to better serve the end-user.

Accomplishments that we're proud of

We're proud that we were able to get this to work, and build it from scratch in such a short period of time. We were a bit worried about whether or not we would be able to get the agent to be accurate within this small window of time, but we're happy with the results, and it seems like an actual viable solution for Yasmeen and other women. We hope this can help other women maximize their productivity while also dealing with the challenges that women-hood comes with.

What we learned

We learned how to train a knowledge base in a way that's actually useful to the end user, and also utilize user input for more accurate and appropriate agentic responses. We also learned how to work together in a style where we both benefit from each other's strengths, and work through challenges in a way where we work together.

What's next for SyncTrack

Both Yasmeen and Milad will put it in front of their female friends to see if it can help them, and listen to users about what works well and what does not. We would also love to build other agentic projects that can really help advance the lives of others.

Built With

• aws-bedrock
• base
• claude-3.5-sonnet
• engine
• lambda
• next.js
• rag
• tailwindcss
• typescript