Lilith

What inspired me

I'm 31, and I've had a complicated relationship with my cycle since the very beginning. As a teenager, I had intense pain every month. I told everyone around me — and every time, I got the same responses: "Thank god you're not pregnant." "This happens to everybody." "That's just your period."

Nobody believed how bad it was until I nearly fainted in front of the school nurse. Even when I finally saw a doctor, nothing looked alarming on paper. I was told to take analgesics before my period started. No explanation. No follow-up. At my early 20s, I was diagnosed with PCOS and given medication — which caused severe GI episodes that almost sent me to the ER. For two years, I lived with that pain. My doctor kept dismissing it. When I pushed hard enough, she suggested switching treatments.

The birth control pill helped some symptoms. But my periods still didn't seem normal, and no one could explain why. The mood swings stayed. Nobody connected the dots. In 2024, I finally found a health professional who didn't gaslight my needs. I got treated for depression and anxiety, and was diagnosed with ADHD.

Then, after recent medication changes, I started struggling again. Half the month I felt like myself — creating, laughing, functioning. The other half was something else entirely: depressed, anxious, unable to function, losing interest in everything I loved. I thought: what am I doing wrong?

The pattern wasn't new. I recognized it from years of fighting with an ex-boyfriend before my period, from the times I couldn't explain my own reactions. But the severity this time was alarming. I also noticed my mood stabilizer and stimulants seemed to stop working at certain points in my cycle.

So I started observing. Journaling. Tracking patterns. Using AI to find insights. What I discovered: what I'd been experiencing for years was a severe form of PMS — and I wasn't alone. The number of women experiencing the same thing, dismissed the same way, was staggering. That's why I built Lilith.

How I built it

I already had a system: I wrote daily notes in Google Keep, then shared them with AI for pattern analysis. It worked, but it was manual, fragmented, and not designed for this. I had an idea. So I started prompting in Claude — what began as a journal form became an app mockup in React. Using that skeleton, I used Goose by Block to implement the full calculation logic, AI integration, and bug fixes. The UI evolved from a wireframe into something that felt like it was made for women who've never been taken seriously by their health system. Stack: React 19 · Gemini 2.0 Flash Lite via OpenRouter · localStorage (all data stays on your device) · Vercel

Challenges I faced

I'm not a developer. Communicating exactly what changes I needed — and in what order — was harder than I expected. When I ran out of API credits mid-build, I had to debug and fix things myself with AI assistance. That was frustrating and also, somehow, one of the best learning experiences I've had. I also discovered this hackathon late — late February — which meant I had almost no runway. But the idea of building something that genuinely matters pushed me through.

What I learned

The biggest lesson: prompt structure is everything. The clearer and more specific my instructions to AI, the better the output. I need to get much better at this. I also learned that building something personal doesn't make it less universal — it makes it more honest. Lilith is built around my needs, but every woman I've talked to about it has said "this is exactly what I needed too."

What's next

Lilith is working, but she's early. The prompt architecture needs refinement. More real-world data is needed to surface patterns with confidence. I want a diverse group of women to test it and tell me where it breaks — especially women whose cycles don't look like mine. The long-term vision: a tool that gives every woman the kind of informed, non-judgmental health conversation I waited 31 years to have.

Built With

• claude
• css3
• goose
• javascript
• llm
• react
• wsl