Meet Sophie
Your behavioral health operations assistant. Sophie answers from current workspace and module data, surfaces what needs attention, and proposes next actions — every one of which you approve. Today she is deterministic: no large language model and no external AI calls.
Grounded. Recommend-only. Context-aware.
Grounded in workspace data
Sophie only answers from current workspace/module records she is permitted to see. She does not invent or generalize. If the answer is not in that data, she says so.
Recommend-only
Sophie never auto-submits, auto-files, or auto-acts. Every suggestion requires your explicit click. She is an advisor, not an actor — by design, not by accident.
Context-aware
Sophie knows which module you are in, what is currently pending, and what the rules are for your organization. She answers questions relevant to where you are, not generic ones.
Answers that name their source.
Sophie is live in the current platform release. These are the things she can do today.
Important: Sophie is recommend-only
Sophie is an operations-support assistant. All outputs are recommendations — Sophie does not make clinical determinations, does not diagnose, and does not prescribe. No action is taken without explicit human approval. Sophie does not store protected health information (PHI). Outputs should be reviewed by a qualified clinician or operations professional before any action is taken.
Sophie is live in the platform.
See a demonstration with the synthetic dataset — no PHI, no signup required to view.