5 Questions Health System Leaders Should Be Asking About Epic and AI

For many health systems, the Epic AI conversation is becoming less about what is possible and more about what is worth doing next. New capabilities are emerging across clinical, operational, financial, and patient-facing workflows. The challenge is determining which opportunities can improve performance without creating more complexity for teams already carrying heavy operational demands.

Epic, one of the most widely used electronic health record (EHR) platforms, supports many of the core processes that health systems rely on every day. As Epic continues to expand its AI roadmap through Art, Penny, Emmie, Agent Factory, Curiosity, and other capabilities, leaders need a clear way to separate near-term value from noise.

Health system leaders can start with five questions to evaluate where AI fits, what is ready to scale, and what it will take to achieve meaningful outcomes for the organization, its workforce, and the patients it serves.

1. Where is AI already embedded in our Epic environment?

Many health systems are already using AI-enabled functionality in some form, but adoption is uneven. Some capabilities may be live but underused. Others may sit within Epic-adjacent tools, existing workflows, or newer modules that have not yet been fully evaluated for value. Before pursuing the next AI use case, leaders should understand what is already available, where it is being used, and what remains latent across the enterprise.

BDO perspective: Start with an inventory of current Epic functionality, AI-enabled workflows, underused automation, adjacent tools, and where employees are already experimenting before adding new solutions.

2. Which use cases are mature enough to scale?

Eighteen months ago, many Epic AI conversations centered on familiar entry points such as ambient documentation, draft notes, and patient communications. Those use cases still matter, but the market has moved quickly. Health systems are now evaluating AI across clinician workflow support, revenue cycle performance, prior authorization, denials management, scheduling, billing, patient access, and patient engagement.

The opportunity is to distinguish between experimentation and readiness. Use cases that improve documentation speed, reduce administrative friction, support revenue cycle performance, or improve patient service may be strong candidates for broader deployment when they can be tied to clear measures of impact.

BDO perspective: Prioritize use cases tied to measurable operational, financial, clinical, employee, or patient outcomes. Test what works in practice, measure the results, and scale what delivers value.

3. Are our workflows ready for AI, or are we automating complexity?

AI performs best when the underlying workflows, build, and governance are sound. If core processes are fragmented, highly manual, or inconsistently adopted, AI may accelerate the wrong work rather than improve it. Before deploying new AI capabilities, organizations should understand where friction exists today and whether processes are working as intended.

Epic’s core capabilities continue to represent meaningful opportunities across workflow design, order and message routing, SmartTools, decision support, and other underused functionality already present in the platform. Recent progress in AI-supported charting, end-of-shift notes, coding, prior authorization, denial management, and patient service workflows reinforces the same point: The strongest outcomes tend to come from organizations that establish the right foundation first.

BDO perspective: Do not automate broken workflows. Simplify and strengthen the underlying process, then engage the people closest to the work to identify where AI can improve it before scaling.

4. Does our governance and change management approach support AI adoption?

As AI moves deeper into clinical, operational, and patient-facing workflows, governance and change management cannot be afterthoughts. Leaders need clear risk tiers, human-in-the-loop (HITL) expectations, monitoring processes, accountability models, and decision rights for when AI can recommend, draft, summarize, route, or act.

Even well-designed capabilities will struggle to deliver value without executive alignment and deliberate support for the people expected to use them. Health systems should engage clinical, operational, technology, and administrative leaders early; communicate how workflows and responsibilities will change; and equip teams with the training, support, and feedback mechanisms needed to build confidence and sustain adoption. Managers are often the link between strategy and execution, helping employees understand what is changing, address concerns, and incorporate new tools into daily work.

Management involvement becomes especially important as Epic advances workflow-level AI through capabilities such as Agent Factory and predictive intelligence through Curiosity. The more AI is used to orchestrate workflows or support decisions, the more health systems need governance and organizational change strategies that support safety, compliance, employee adoption, and patient trust.

BDO perspective: Set clear direction, guardrails, governance, and human-review expectations while equipping leaders, managers, and employees to safely experiment, adopt, and improve AI-enabled ways of working.

5. How will we measure value beyond implementation?

Staying current with Epic is increasingly important, but availability alone is not the same as value. Gold Stars and other measures of adoption can help signal progress, but executives also need to understand whether new capabilities are improving outcomes, reducing burdens, strengthening financial performance, supporting strategic priorities, and helping employees work more effectively.

Implementation is only the beginning. Health systems also need to determine whether new AI capabilities are producing measurable improvements in performance, reducing administrative burden, and supporting organizational goals. This often means starting with workflows that are already performing well, identifying latent AI, introducing automation where it can add value with minimal disruption, validating results, and then scaling more advanced capabilities over time.

Organizations should also evaluate how AI is affecting the workforce, including employee experience, adoption patterns, manager effectiveness, and the amount of time clinicians and staff can redirect toward higher-value work.

BDO perspective: Measure AI by adoption, sustained use, workflow impact, financial return, and the employee and patient experience, not by implementation alone.