Serif
Serif is health middleware that delivers individualized causal intelligence to longevity and health platforms. We sit inside existing apps and EHR portals that already own the data (wearable dashboards, longevity clinics, provider networks) and run individualized Bayesian causal inference on their users’ longitudinal health data. Where LLMs give population-average answers, Serif treats each user as a sample size of one: continuously updating posterior distributions, simulating counterfactuals, and returning confidence intervals instead of recommendations. Our goal is to help platforms move from dashboards to genuine health outcome ownership, using non-commoditizable intelligence that compounds with every new data point.
Points of Differentiation
- Individualized Bayesian causal inference: each user is a sample size of one with their own posterior, not population-average LLM outputs.
- Counterfactual simulation: models simulate interventions before they happen with quantified uncertainty.
- Middleware, not platform: plugs into existing health apps and EHR portals via API, making their intelligence layer non-commoditizable.
- Compounding intelligence and transfer learning: models update continuously with new data, creating a moat that deepens over time.
Supplementary Images
Team
Sam Cialek, CAS ‘11
