If you work in a hospital running Epic, you are already using AI — whether you know it or not. Epic has been building machine learning capabilities into its platform since the mid-2010s and has accelerated that investment significantly since 2022, following its partnership with Microsoft's Azure OpenAI platform. The question isn't whether AI is in your EHR; it's which AI features your institution has activated and what you as a nurse need to understand about them.
This article covers the Epic AI features most directly relevant to nursing practice in 2026, with a clear distinction between what is standard in most Epic installations versus what requires separate licensing or is still being rolled out.
Epic's most visible AI feature for documentation is its ambient listening integration with Nuance DAX Copilot (now owned by Microsoft). In hospitals that have activated this feature, clinicians — including nurses in some implementations — can speak naturally during patient interactions and have an AI-generated note draft appear in their Epic flowsheet or nursing note section.
For nursing specifically, ambient listening implementations vary widely. Some hospitals have configured Epic ambient listening for nursing admission assessments, shift handoffs, and discharge education documentation. Others have deployed it for physician notes only. The configuration is determined at the health system level, not by individual nurses.
For a detailed review of Nuance DAX Copilot specifically, see our full article: Nuance Dragon Medical and DAX Copilot for nurses.
The Epic Deterioration Index is one of the most widely deployed AI clinical decision support tools in U.S. hospitals. It is a machine learning model — trained on Epic's enormous multi-institutional patient dataset — that predicts the probability of patient deterioration (including sepsis, respiratory failure, and cardiac arrest) at regular intervals throughout a patient's stay.
The EDI score appears in the nursing view of the patient list and flowsheets in Epic installations where it has been activated. It generates a 0–100 score, with higher scores indicating higher predicted risk. Alert thresholds are set at the institutional level — most hospitals configure alerts at scores above 60-65, though this varies.
What nurses should understand about EDI: The EDI score is a prediction of risk based on current vital signs, lab values, nursing documentation, and other data already in the Epic record. It does not replace clinical assessment — it augments it. The score updates in real time as new data is entered. A score that rises from 40 to 70 while you are charting is a signal worth investigating, even if the individual data points look individually manageable.
The EDI is most useful when nurses understand the model's inputs. High scoring is driven most by: tachycardia, declining oxygen saturation, increased respiratory rate, rising lactate, reduced urine output, and documentation of altered mental status. If a patient's score is rising and you can identify which of these inputs is driving it, you have an early target for intervention.
See also: AI-powered sepsis early detection: what nurses need to know
Epic's InBasket AI features use large language models to assist with message triage and response drafting. In most current implementations, this is most visible to providers (physicians and APPs) rather than bedside nurses, but nursing managers, case managers, and nurses who handle InBasket communications will increasingly encounter it.
InBasket message drafting: When a nurse manager or case coordinator opens an InBasket message requiring a patient response or inter-departmental communication, an AI-generated draft response may appear. The nurse reviews, edits, and sends — rather than writing from scratch. This is a productivity tool, not a clinical decision tool.
InBasket prioritization: In health systems with this feature activated, AI ranks InBasket messages by urgency based on keywords, sender type, patient acuity, and message content. Urgent clinical messages from labs, pharmacies, or high-acuity patients surface first.
Epic's SmartForms — the templated documentation structures used for nursing assessments — have incorporated AI-powered predictive text and auto-population features that have been live in many implementations since 2023-2024.
What this looks like in practice:
The handoff auto-generation feature is the one bedside nurses encounter most directly. In Epic implementations with this feature configured, you can generate a structured handoff report with a single click rather than building it from scratch — then edit what the AI generated rather than writing from nothing.
Epic has built AI-powered compliance tracking into its sepsis bundle tools. For nurses managing sepsis patients, this appears as smart reminders and workflow nudges — not just passive alerts.
When a patient has a sepsis alert (from EDI or from a nurse-triggered sepsis protocol), Epic tracks bundle elements and uses AI to prompt the nurse about incomplete elements in real time. If lactate has not been drawn 2 hours after the initial draw, a reminder surfaces. If repeat blood cultures are overdue, the flowsheet highlights the gap. If antibiotics are due and haven't been administered, the MAR generates a time-sensitive reminder.
This is AI as a checklist assistant — ensuring that the time-sensitive bundle is completed in sequence even in the chaos of a busy ICU. Studies of Epic sepsis bundle compliance AI have shown statistically significant improvements in bundle completion rates.
Epic's medication reconciliation workflows at admission and discharge now include AI-assisted identification of discrepancies between the patient-reported medication list and the EHR medication record. The AI compares medication names, doses, frequencies, and routes across sources and flags potential discrepancies for pharmacist and nursing review.
For nursing specifically, this reduces the burden of manual reconciliation comparison — particularly at admission, when collecting a medication history from a patient who may be acutely ill, cognitively impaired, or without a clear medication list. The AI surfaces the discrepancies; the nurse and pharmacist make the clinical judgment.
Epic's roadmap for 2026-2027 includes several features in limited rollout or development that nursing staff should be aware of:
Epic AI features are activated and configured at the health system level. Not every hospital running Epic has the same features enabled, and the same feature may look different at different institutions depending on how it was configured.
To find out what AI features are active where you work:
Understanding which AI tools are embedded in your EHR is increasingly part of nursing competency. Nurses who understand what the EDI score represents, how InBasket drafting works, and what the sepsis bundle AI is tracking are better positioned to use these tools critically rather than passively.
See also: AI tools ICU nurses are actually using | Clinical decision support in the ICU
Get The ICU Notebook Newsletter
Clinical tools and career insights for ICU nurses. One email per week, no fluff.
Yes, send it freeNo spam. Unsubscribe any time.