Every person deserves to see the road first

Pilots train in flight simulators before they fly the real plane. Your health gets no simulator. Every choice about your care happens once, to the real you.

We started Double Time Labs so every person can see the road first. We are the frontier lab building world models for medicine. Our goal is a digital twin for every person.

1. Imagine a twin that lives a billion lives, so you can live your best one.

   Your twin would be a computer model of your health. With your permission, it would know your story: your conditions, your medicines, your past hospital stays. Each day it would read signals from your body, like heart rate and sleep, and learn what normal looks like for you. Say your heart rate at night is usually 62. A night at 70 looks fine on a chart. For you, it is worth a closer look.

   Then you could ask it questions. What if your sleep stays poor for another week? What if a nurse checks on you tonight, not next week? What if nothing changes at all? The twin would play out each path and show how your health might change.

   That is what we mean by a billion lives: more futures than any doctor or family could ever picture. Your twin walks those roads first, so you do not have to. Then the real you, with the people who care for you, can choose the best path.

2. A world model learns what happens next, and what happens if you act.

   AI labs now build world models for roads, robots and games, and early work has begun in health. Our world is the human body over time. We are building a world model that learns how health changes. Your twin would be that model, tuned to you.

   Many tools that watch patients stop at a risk score, like “this person may fall.” A score does not tell anyone what to do. We are building a twin that goes further. It would compare actions, suggest one, see what happened, and learn. We call this a world action model.

3. Start from the health record. Then add the body’s signals.

   Your history is already written down, so the record comes first. Watches, monitors and carers’ notes add to it.

   Published research supports this. One 2025 study looked at over 30,000 emergency visits for possible heart trouble. Read together, the record, a heart test (ECG) and vital signs did better at predicting who would need a hospital stay. They beat the record alone and the ECG alone. In another study, adding data from devices people wear, like watches, to the record improved predictions for all 10 health outcomes it tested. Our CEO’s own epilepsy research saw the same: the record plus the brain signal beat either one alone. That work is small, early and not yet published.

   Every reading comes from one real person. Imagine every reading, from a watch, a heart monitor or a lab test, checked against your twin. We believe the tests we already have could work better, because each result would be read with your whole story.

4. Why now: AI can learn from millions of health records, and more people wear sensors.

   A model from Epic, a health records company, has already learned from the records of about 118 million patients. In a survey of 8,000 Americans, 46% owned a smartwatch or ring. In 2015, only 13% owned any wearable. We believe the next step is to bring world models, records and sensors together, one person at a time.

5. The twin advises. A named person decides. Trust comes in steps.

   The twin will never diagnose, prescribe or change a dose. A twin can be wrong. So a named person makes every decision: your carer, nurse or doctor, together with you or your family. The twin shows. People choose. We are building it so that if a sensor fails, carers get a clear message and go on with their usual care.

   We plan to earn trust in three steps. First, the twin watches quietly. It makes predictions and writes them down, and nobody acts on them. Next, it advises: one action for one person, with a reason. Only then comes the full loop: a person acts, the twin sees what happened, and it learns.

6. Care homes first. Then every person.

   In US nursing homes, about 1 in 4 residents goes to hospital each year. Trouble can show in small changes first. A resident sleeps a little worse and drinks a little less. Each change looks normal on its own. Together, they can point to trouble.

   So we plan to start there, because the need is clear. Next come care at home, then hospitals. Each stage must earn the next. In time, we want a twin for every person, built with their permission from the records and devices they choose to share.

7. Who we are. Nothing is built yet.

   Our CEO is an ex-YC founder who won a Rhodes Scholarship to Oxford. They co-wrote a paper on a sleep model small enough to run on a tiny chip.

   Our CSO is an Associate Professor of AI for Digital Health at Oxford. One heart model they helped build learned from 2.3 million ECGs.

   Between us, we have worked on health records, heart and brain signals, and AI small enough for a chip. We have known each other for close to five years.

   We started in October 2026. This page is our plan, not a result. When we have results, we will say what we have shown, not what we hope.

8. Let your twin live the billion lives. You live the best one.

   Imagine every big choice in your care. Your twin has already tried the paths ahead. People you trust stand beside you. Every person deserves to see the road first. We are building it. There is a place for you here: if you give care, if you need care, or if you want to help build. Join us.