Oura Launches In-House AI Model for Women’s Reproductive Health

Purpose-built reproductive health AI trained on clinical data, not generic models, signals that wearable companies are moving beyond aggregation into domain-specific medicine.

Photo: Petr Urbánek / Unsplash

Oura just announced its first AI model built entirely in-house — not layered on top of ChatGPT or Gemini, but trained from scratch on clinical research and nearly a decade of real biometric data from its own members. It’s designed specifically for women’s health. And it’s live now in beta for anyone with an Oura Ring who wants to try it.

That might sound like a software update. It’s a little bigger than that.

What Is Oura and Why Does It Matter?

If you’ve never heard of Oura, here’s the quick version: it’s a titanium ring you wear on your finger that tracks what your body is doing while you sleep, exercise, and go about your day. Heart rate, heart rate variability, skin temperature, blood oxygen, respiratory rate — all from your finger, 24/7.

The company started in Oulu, Finland in 2013. Four engineers who believed the key to better health wasn’t counting steps — it was understanding sleep. While Fitbit was racing to be the best pedometer, Oura was asking a different question: what happens to your body when you’re not doing anything?

The first ring launched on Kickstarter in 2015. Stanford researchers bought two of them and published a study confirming it was the most accurate consumer wearable for sleep tracking at the time. That validation gave Oura something most fitness gadgets never get: scientific credibility before mainstream fame.

Since then, three more generations. The Oura Ring 4, released in October 2024, finally moved all the sensors flush inside the band — no bumps, just a smooth ring that happens to be running 50+ health metrics in the background while you live your life.

The Numbers Tell a Wild Story

Oura’s growth over the last two years has been hard to keep up with.

Revenue doubled two years in a row. In 2025, the company sold more rings in a single year than it had in the previous ten years combined. A $900 million Series E in October 2025 valued the company at $11 billion — backed by Fidelity, ICONIQ, and others. CEO Tom Hale has said revenue could hit $2 billion in 2026.

The ring costs $299–$499 depending on the finish. The app is $6 a month. That hardware plus subscription model is the same playbook that made Apple Watch a billion-dollar business unit. Except Oura doesn’t have a watch face.

The US military signed a $96 million contract with Oura in 2024 to use rings for fatigue tracking. Oura is also now the official wearable of Team USA heading into the LA28 Olympics. Those are not the partnerships of a niche biohacker gadget.

Why Women, and Why Now

Here’s something that surprised people when Oura said it out loud last October: its fastest-growing customer segment isn’t gym rats or biohackers. It’s women in their early twenties.

That’s not an accident. It’s been building for years.

Women’s health has been badly served by the medical system for a long time. For most of the 20th century, clinical trials defaulted to male subjects. Drug dosing guidelines, device calibration, health benchmarks — all built around a “70kg reference man.” Women were treated as a variation on the default, not a category with their own physiology.

Most fitness wearables made the same mistake. Cycle tracking was a checkbox feature. Hormonal patterns were ignored. The ring was designed for the gym, not the biology that actually runs 28 days on a cycle with measurable temperature shifts, HRV changes, and sleep pattern variations that a generic algorithm can’t interpret correctly.

Oura built differently. Temperature-based cycle tracking. Fertile window predictions. Pregnancy insights. Perimenopause signals. These aren’t add-ons — they’re built into the core product and developed with OB/GYNs. Women’s retention at the 12-month mark is running in the high 80s, which is unusually strong for a consumer hardware subscription.

The market followed the product. Women now make up the majority of Oura’s user base.

The New AI Model

The announcement Tuesday takes that logic one step further.

Oura built a large language model from scratch — no OpenAI, no Google, no third-party backbone. It was trained on peer-reviewed clinical research, reviewed by Oura’s in-house clinicians and board-certified OB/GYNs, and integrated directly with the biometric data the ring collects.

When you ask Oura Advisor a women’s health question now, it doesn’t go to a generic AI and come back with a Wikipedia summary. It routes to this model, which pulls from its clinical knowledge base and then cross-references your specific data — your temperature trends over the last three months, your HRV across different cycle phases, your sleep patterns, your stress signals. The answer is built around your body.

Tanvi Jayaraman, Oura’s clinical lead for health AI, put it plainly: most AI models train on broad datasets built around broad populations. That means women asking about cycle irregularities, perimenopause symptoms, or hormonal changes get answers calibrated to everyone — which often means calibrated to men. Oura’s model is built to know the difference.

The model is also intentionally designed to not be dismissive. That’s a specific product decision, not marketing language. Women’s health concerns have historically been minimized in clinical settings. The team wanted the AI to feel like the opposite of that.

One thing it isn’t: a doctor. Oura is explicit — this doesn’t diagnose, doesn’t prescribe, and shouldn’t replace a clinical appointment. The goal, as Oura’s clinical director of women’s health Dr. Chris Curry described it, is to help a woman “walk into her appointment more informed and confident.” Think of it as prep work, not a substitute.

The model lives entirely on Oura’s own servers. Conversations aren’t shared or sold, and they’re not used to train anyone else’s AI.

It’s available right now in Oura Labs — the company’s opt-in beta hub inside the app — for anyone who wants early access.

The Bigger Picture

Oura is not the only company in this space. Samsung launched the Galaxy Ring last year. Startups like Ultrahuman and RingConn compete on price. Apple has ring patents and is rumored to be years away from a product.

But none of them have what Oura has: ten years of real biometric data from millions of real users, and a clinical team that has been building health features with that data the whole time. You can’t fast-follow that. You have to start when they started.

The custom AI model is the latest layer on top of something that took a decade to build. That’s the moat — not the ring, not the app, but the combination of longitudinal data and clinical depth that makes the product actually useful for the people it was built for.

What to Watch the Rest of This Week

Oura Labs adoption: Oura will likely share early opt-in data from the beta. Strong numbers would signal real demand before any wider rollout.

Competitor response: Clue, Natural Cycles, and Maven Clinic all operate in adjacent women’s health AI territory. Watch for any counter-announcements in the coming days.

FDA digital health rules: Oura noted in a February 10 post that new FDA guidance is opening doors for them. How regulators classify AI health advisors inside consumer devices matters for everyone in this space.

IPO chatter: Oura is profitable, valued at $11B, and pulling $1B in annual revenue. Tom Hale has cited Stripe and SpaceX as reasons to stay private — but he hasn’t ruled it out. Any banker roadshow signals are worth watching.