Unlearn.AI

Unlearn.AI works with biopharma sponsors and clinical research teams at three points in a trial's lifecycle: before the protocol is finalized, while the study is running, and after readout. In design, teams model the patients they expect to enroll and pressure-test sample size, eligibility criteria, endpoints, and comparators before locking the protocol. The company's PROCOVA method — described as EMA-qualified and FDA-supported — applies twin-based covariate adjustment to reduce sample size in randomized controlled trials while holding statistical power, or to increase power without adding patients.

Mid-study, Unlearn supports pre-specified twin-based analyses that tighten interim estimates without enrolling anyone new, and lets teams re-run design scenarios when new evidence — a competitor readout, a fresh publication, or blinded data drifting from assumptions — changes the picture. After a study reads out, digital twins are used to contextualize population, endpoints, and variability, improving sensitivity across endpoints and helping teams re-evaluate historical trial data with regulatory-aligned methods.

Unlearn publishes concrete case outcomes: a 20% sample-size reduction built into a Phase 1/2 design with roughly $2M in savings, a 33% smaller control arm and four-plus months of enrollment time saved in a re-analysis of completed Phase 3 trials from a top-5 global sponsor, and a completed Phase 3 ALS trial re-analyzed to keep full power with 92 fewer participants. The company partners with sponsors including Trace Neuroscience, QurAlis, Acumen Pharmaceuticals, SOLA Biosciences, VectorY Therapeutics, and CHDI Foundation, and states it works with top global pharma companies. Unlearn does not publish list pricing; engagements are scoped through direct contact.