EDENA, defined.
EDENA is the safety, ethics, and accountability layer that governs how AI moves from suggestion to action — classifying risk, gating consequential decisions before they execute, preserving accountable human judgment, and creating the evidence that AI acted within human-defined boundaries.
One framework, stated for two audiences.
EDENA — the Ethical Decision Engine for Nurse-led AI — can be read plainly or precisely. Both describe the same thing: the boundary between what AI proposes and what is allowed to happen.
EDENA is the safety-and-ethics layer for agentic AI. It ensures that AI systems propose, humans judge, and nurses steward the boundary between intelligence and action.
EDENA is a policy-and-risk execution layer for governed autonomy. It evaluates candidate AI actions before they execute, assigns risk tier, checks context, data sensitivity, reversibility, autonomy, and blast radius, routes required human review, logs the decision path, and activates containment when an AI system exceeds its authorized scope.
The complete, website-grade statement.
EDENA is the Ethical Decision Engine for Nurse-led AI: a runtime governance framework for agentic systems that classifies AI capabilities and actions by risk, gates consequential decisions before execution, preserves accountable human judgment, routes escalation to the right human authority, and creates auditable evidence that AI acted within human-defined ethical, clinical, operational, and safety boundaries.
Most AI governance frameworks answer one question: was this AI system developed responsibly? EDENA answers the more immediate, operational one: should this AI-generated action be allowed to happen right now? That is its distinctive contribution. EDENA is not a philosophy printed on a wall; it is a decision architecture that runs at the moment of action.
The strongest case for EDENA is also the simplest. AI development has moved from answer-generation to delegated autonomy, and delegated autonomy requires real-time governance, not retrospective ethics. Where earlier AI merely produced outputs, modern systems use tools, manage state, execute workflows, retain memory, and hand tasks between agents. EDENA governs the transition from output to action.
What EDENA is — and what it is not.
EDENA is precise about its own scope. It is a governance mechanism, not a clinical actor; an authority over action, not a participant in care.
A decision architecture
- A decision architecture — the policy-and-risk logic that sits between AI intent and action.
- A policy decision point that resolves each candidate action to a governed outcome.
- A risk classifier that tiers both the interaction and the underlying capability.
- An approval router that sends consequential decisions to the right named human.
- An evidence generator that records grounding, provenance, uncertainty, and rationale.
- An audit mechanism that makes the decision path reviewable in real time and after the fact.
Not a participant in care
- Not a chatbot — it does not converse or generate content for users.
- Not a diagnosis model — it makes no clinical determinations and renders no medical opinion.
- Not an ethics-committee replacement — it operationalizes oversight; it does not displace human deliberation.
- Not a passive dashboard — it acts at runtime to allow, review, escalate, contain, or block, rather than merely reporting after the fact.
The stewardship layer is nursing, by design.
EDENA's most distinctive contribution is nursing epistemology: whole-person, systems-aware judgment. The nurse sees not only the output but the patient, the family, the workflow, the unit culture, the burden, the environment, and the downstream consequences. Where AI optimizes a task, nursing stewards the environment in which the task occurs.
This is not an aspiration; it is professional obligation. The ANA's 2025 Code of Ethics — developed with more than 6,000 nurses and 49 experts — states in Provision 4.2 that nursing practice authority can be eroded by AI when it is integrated without careful consideration of harmful consequences, and that nurses remain accountable for the decisions and actions of their practice. Stewardship is the safeguard against that erosion.
It is also a question of trust. The authority to steward must be credible to patients, clinicians, and institutions alike — and no profession holds that standing more firmly.
Where EDENA sits: between AI intent and action.
Florence X is the reference architecture in which EDENA operates as the policy decision point. Florence X invokes EDENA rather than containing it — the governance layer is structurally separate from the system it governs.
In the Florence X pattern, the protection of sensitive data is part of the architecture, not an afterthought. Protected health information (PHI) remains on the edge. Cloud inference operates only on redacted, non-PHI prompts. PHI access, persistence in memory, tool exposure, external transmission, and agent-to-agent sharing are each treated as gated governance events.
The operating loop closes on a person. EDENA gates the move from proposal to action, a named human validates anything consequential, and every decision — allow, review, escalate, contain, or block — persists as evidence. Feedback returns to learning, so governance improves with use rather than drifting out of date.
You have the definition. Now see the doctrine.
EDENA is expressed as twelve operational principles and a published standards stack — each one anchored in regulation, clinical evidence, and nursing ethics. Start with the principles, then read how they become normative requirements.