Context before output.
Good support starts by showing what is known, what is missing, and where each piece of information came from.
About Stan
Built around the way psychiatrists think.
Psychiatric care depends on context accumulated over time. Stan is designed to help hold that context together without taking control away from the clinician.
Our position
Intelligence should make responsibility clearer.
The physician remains the decision-maker. Stan can organize, surface, question, and draft—but it cannot diagnose, prescribe, or approve its own work.
That boundary should be visible in the interface, preserved in the data model, and testable in the release process. A disclaimer alone is not enough.
Product principles
Fewer promises. Stronger boundaries.
Review before record.
AI content remains a draft. It cannot quietly become clinical documentation or alter a care plan.
Safety before speed.
Privacy, tenant isolation, crisis handling, evidence, and traceability are release gates—not features to add later.
How we are building
Privacy before intelligence.
Private by default
Patient information stays scoped to authorized workspaces, with explicit controls for every read, write, export, and AI call.
Evidence in view
Generated clinical claims must retain their source, version, limitations, and uncertainty.
Useful without AI
Core patient records and clinical workflows remain available when an AI provider is unavailable.
Release with proof
Safety evaluations, tenant-isolation tests, restore drills, and professional review come before real-patient use.
Early access