Folia is a research-backed daily companion for women navigating PCOS — live in private beta. I'm the solo founder: I designed and built the whole product, and I'm now interviewing women with PCOS to sharpen what comes next.
I'm building it because PCOS care is fragmented by design: short clinic visits, sporadic dietitian advice, conflicting online protocols — and nothing that knows your actual day. PCOS is longitudinal, emotionally loaded, and deeply individual. Folia is the daily layer that the system never built.
The full story of why — fifteen years of PCOS, and the single blood test that finally gave me a root cause — is in 15 Years, One Number.
How it works
- Check in naturally — voice or text, whenever it fits. No rigid forms or streak pressure. AI extracts meals, mood, symptoms, and movement into a structured, editable diary.
- Learn with citations — a research chat that answers only from human-approved excerpts, with visible [n] citations. When the evidence isn't there, it says so.
Governed AI, not guesswork
Trust is the biggest barrier in PCOS care, so the research layer is deliberately constrained. The index is built from 35 research papers I found genuinely useful and curated by hand — each one parsed, chunked, and manually reviewed before it can appear in an answer. Retrieval runs over an approved-only pgvector index in Postgres, and every claim carries a citation you can open. Ask it something outside the research — why the sky is blue — and it tells you it has no research on that, instead of improvising.
What early users are teaching me
My first user interviews — five women with PCOS — reshaped the roadmap. Women who aren't actively thinking about fertility tend to defer dealing with PCOS: "I'll worry about it when I have to." What does capture interest is a number. So the next iteration leads with root-cause biomarker benchmarking: test the specific markers driving your PCOS (insulin response and hormone panels — not just weight), work on them through Folia's daily check-ins and cited research chat, then retest in about four months and watch the values move. The same loop that gave me my own root cause after fifteen years, productized.
Where it is today
- Working in beta: voice and text check-ins with AI extraction, a structured daily diary across 8 categories, cited research chat, the admin review pipeline, and magic-link sign-in
- Next: root-cause biomarker benchmarking (test → daily check-ins → retest in ~4 months), pattern insights over time, and clinician visit-prep summaries