What this agent receives
Explicitly consented notes, patient questions, appointment history and curated public mental-health references.
GOVERNED SPECIALIST OFFICE / CONCEPT
Psychiatry Support agent.
No. This conceptual Psychiatry Support office can organize patient-consented notes, explain general reference material and prepare questions for licensed clinicians; psychiatric assessment, prescribing and emergencies remain with qualified people.
Operational home: Family & Personal · Health. See deployment example →
REAL JOB / REVIEWABLE OUTPUTINPUT / LEAST PRIVILEGE OUTPUT / EVIDENCE POLICY / APPROVAL
Can a psychiatrist AI agent diagnose or prescribe medication?
No. This conceptual Psychiatry Support office can organize patient-consented notes, explain general reference material and prepare questions for licensed clinicians; psychiatric assessment, prescribing and emergencies remain with qualified people.
What it hands back
Private visit-preparation summaries, symptom timeline drafts, questions for the clinician and reminders.
What it cannot self-authorize
No diagnosis, medication change, crisis triage autonomy, sharing of intimate records or disclosure to other domains; emergency situations must use immediate human services.
Sample workflow
Patient opts in → assistant organizes notes locally → qualified clinician reviews information → patient decides what to share.
What should be measured?
Verify consent, secure access, deletion/retention controls and clinician review—not clinical efficacy claims. These are acceptance criteria for a proposed workflow, not published benchmark results.
Every office remains a conceptual integration pattern until tools, credentials, policies, data scope and tests are configured for a deployment. An agent's job title never grants host or cross-domain authority.
ROLE-BASED ACCESSREVIEWABLE HANDOFFDAEMON POLICYOPERATOR OVERSIGHT
How the XMAN daemon makes resource decisions → · How SIDE supports agent development →