# Unlearn.AI

> Unlearn.AI is a San Francisco biotech-AI company that builds disease-specific machine learning models to generate digital twins of clinical trial participants - longitudinal predictions of how each patient would have fared on placebo. Folded into a statistical method called PROCOVA and a trial design called TwinRCT, the approach lets sponsors shrink control arms by up to a third, so more patients receive the experimental drug while trials reach answers faster. Founded in 2017 by physicists Charles Fisher, Aaron Smith and Jon Walsh, the company has raised about $130M and earned a first-of-its-kind EMA qualification for an AI-based method to reduce sample size in pivotal trials.

- **Founded:** 2017
- **Headquarters:** San Francisco, California, United States
- **Founders:** Charles K. Fisher (Co-Founder (former CEO; biophysicist)), Aaron Smith (Co-Founder & Head of AI), Jon Walsh (Co-Founder & Chief Scientific Officer)
- **Team size:** ~71 employees
- **Products:** Digital Twin Generators (DTGs), TwinRCT (Twintelligent RCT), PROCOVA (Prognostic Covariate Adjustment)
- **Notable:** Received a first-of-its-kind EMA qualification opinion supporting an AI/machine-learning method (PROCOVA) to reduce sample size in pivotal trials, with the FDA concurring it does not deviate from current guidance., Built 13 disease-specific Digital Twin Generators trained on a proprietary dataset of 300,000+ patients and 1M+ patient interactions., Reported reductions in control-arm size of roughly 30-35%, and clinical trial timeline reductions cited at 25%+.

## Products & services

- **Digital Twin Generators (DTGs)** — Disease-specific machine learning models trained on large historical clinical datasets that produce a 'digital twin' for each participant - a comprehensive, longitudinal prediction of that patient's likely outcomes on a control/placebo treatment.
- **TwinRCT (Twintelligent RCT)** — A randomized controlled trial design that merges digital twins with new statistics to run trials with smaller control arms - reported to cut control-arm size by up to ~35% - without breaking randomization or introducing bias.
- **PROCOVA (Prognostic Covariate Adjustment)** — A statistical method for designing and analyzing RCTs using a prognostic score derived from each participant's digital twin; the subject of a first-of-its-kind EMA qualification opinion.

## Achievements

- Received a first-of-its-kind EMA qualification opinion supporting an AI/machine-learning method (PROCOVA) to reduce sample size in pivotal trials, with the FDA concurring it does not deviate from current guidance.
- Built 13 disease-specific Digital Twin Generators trained on a proprietary dataset of 300,000+ patients and 1M+ patient interactions.
- Reported reductions in control-arm size of roughly 30-35%, and clinical trial timeline reductions cited at 25%+.
- Raised approximately $130M in total venture funding across Series A-C.
- Reanalyzed an early-stage trial for drugmaker Remynd and surfaced an efficacy signal that would otherwise have been missed.

## Latest updates

- **2025-11** — Peer-reviewed study with AbbVie on AI-generated digital twins boosting clinical trial efficiency in Alzheimer's disease published in Alzheimer's & Dementia: TRCI.
- **2024-02** — Closed a $50M Series C led by Altimeter Capital to scale AI-powered digital twin technology for clinical research.
- **2024** — Steve Herne serving as Chief Executive Officer; board includes Mira Murati (former OpenAI CTO) and Ann Taylor (former AstraZeneca CMO).

## Links

- Website: https://www.unlearn.ai/
- LinkedIn: https://www.linkedin.com/company/unlearn-ai/
- Twitter/X: https://x.com/UnlearnAI
- YouTube: https://www.youtube.com/channel/UCJjWFXqy7P9yEfB8GyrZOKA
- Facebook: https://facebook.com/unlearnai/

---

Profile page: https://yespress.io/unlearn-ai
Published by YesPress — https://yespress.io
Last updated: 2026-06-20