Featured
What Is a Persistent Virtual Patient?
A persistent virtual patient keeps state, time, disclosure boundaries, history, and replayable branches. It is a world, not a disposable chat.
Patientic Blog
Ideas, systems, and simulation receipts for AI working in patient worlds.
AI Agents Need Environments, Not Just Context Windows
Strong agents need a governed world: inspectable state, permissions, actions, consequences, and history. A longer context window is not that world.
How to Evaluate AI Agents in a Patient World
Evaluate the path, not only the final answer: information gathering, uncertainty, permitted actions, evidence use, state discipline, and branch outcomes.
From MediTutor to Patientic: Why a Patient Cannot Be a Disposable Chat Session
MediTutor's early synthetic sessions were useful and resettable. Patientic is the design response: a patient world that keeps consequential history.
From Standardized Patients to Governed Patient Worlds
Standardized patients remain essential human practice. Versioned synthetic worlds are a complement for on-demand rehearsal, evaluation, and agent research.
How We Created a Patient World and Gave Grok 4.7 a Mission
One fictional patient world, a bounded Grok 4.7 mission, and eight authored branches. Synthetic simulation, not medical advice or a clinical prediction.
Why Every AI Patient Run Needs a Simulation Receipt
A simulation receipt records seeds, model and policy versions, the actor, and time. Without it, a patient run cannot be repeated or honestly compared.
Synthetic Patient Simulation: What It Is—and What It Is Not
The initial boundary is synthetic, educational, or permitted scenarios. No diagnosis, no treatment claims, and no identifiable real-patient data by default.
What a Thousand Patient Simulations Helped Us Notice About Medical AI Interaction
A thousand synthetic sessions did not prove how clinicians behave. They helped us notice what one resettable encounter cannot show next time.
What Branching Clinical Simulation Actually Means
Branching clinical simulation commits decisions, moves time, and keeps forks for comparison and replay. It explores a setup. It does not predict care.
Why Transcript Feedback Needs Longitudinal Patient State
A transcript and a session rubric show one encounter. Later feedback needs the state change, the evidence version, and a history you can replay.
Persistent patients, branching time, and a record of what an agent did.
Explore Patientic