From Copilots to Bounded Autonomy
Agentic AI in Healthcare
Four modules on moving from one-shot AI assistance to systems that pursue multi-step goals, use tools, maintain state and take bounded actions inside real healthcare workflows. You learn what 'agentic' means operationally, how to find workflows worth delegating, how tools, permissions and oversight actually constrain what a system can do, how to evaluate a trajectory rather than a paragraph, and how to choose the lowest sufficient autonomy for the value you want. Prior AI foundations help but are not required, and no coding is involved.
From Copilots to Agents
What 'agentic' means operationally, how it differs from a generative assistant and from deterministic automation, the agent loop, and how to choose the lowest autonomy a healthcare workflow actually needs.
Anatomy, Tools & Healthcare Integrations
Two chapters. First the parts you are actually buying or building: model and orchestrator, instructions and policies, governed knowledge, memory versus workflow state, identity and permissions, tracing, and when a second agent helps. Then how agents act through systems: tool contracts and schemas, read versus write, FHIR and EHR integration at leader level, provenance, least privilege, idempotency and safe failure.
Finding the Work, Autonomy & Oversight
Two chapters. First decomposing real workflows into tasks, decisions, actions, handoffs and exceptions, then judging structure, reversibility, consequence and value to find bounded candidates. Then designing oversight that holds: approval gates set by action risk, permission scopes, escalation and abstention, stop conditions and kill paths, and how to avoid rubber-stamp review.
Evaluation, Security & Implementation
Three chapters. Evaluating a trajectory rather than a paragraph: task success, tool selection, action accuracy, recovery, latency, cost and drift. The threat surface that appears once a system can act: prompt injection, excessive agency, exfiltration, memory poisoning, the safety case, audit trail and regulatory routing by intended use. And getting to a running service: sandboxed starts, shadow mode, pilot gates, operating model, sourcing, TCO, incidents and decommissioning.