This insight was featured in the November 19th, 2025 edition of the AI Catalyst Pulse.
Epic’s patient-facing AI assistant, Emmie, is shaping up to be the company’s boldest attempt yet to redefine how patients navigate care. But behind the polished demos and ambitious roadmap, health system executives are already asking a sharper question:
Is this the beginning of a more streamlined digital front door? Or the start of a deeper dependency on Epic’s ecosystem?
During AI Catalyst’s November 13 session, Epic’s VP Sean Bina previewed where Emmie is heading: scheduling automation, AI-assisted triage, proactive care recommendations, and eventually voice. After the vendor left the room, health system leaders were split. While optimistic about the integration upside, most were also cautious about the operational and financial commitments that come next.
This is the kind of tension that signals a true inflection point.
Epic’s real advantage: Integration
Epic isn’t trying to win by launching the flashiest agent. Its strategy is simpler and more powerful: Put AI inside the channels where patients already live, particularly MyChart, which now reaches nearly 200 million users.
Emmie isn’t a standalone AI bot, rather it’s an intelligence layer woven into existing workflows:
Invisible routing and automation behind the scenes
Patient-facing conversational workflows in MyChart and SMS
Interoperability with Epic’s merged MyChart and Cheers teams, call-center tools, and infrastructure like HelloWorld and Fabric
This is a fundamentally different play from third-party AI vendors. Instead of adding a new digital front door, Epic wants to be the front door (and the hallway, and the back office).
For systems trying to simplify fragmented patient communication, that proposition is compelling.
What’s live now – and what’s coming
Emmie is already in limited use among “pioneering customers,” especially through SMS functionalities like text-based scheduling and billing assistant workflows. Early data from Rush Health shows a 48% reduction in billing-related customer service messages and reduced wait times as patients navigated billing concerns through the new self-service features.
Epic’s longer-term roadmap includes:
November 2025:
Conversational Self-Triage (clinically reviewed reboot of symptom checking)
Anticipate My Needs (next-step care recommendations)
2026 and beyond:
Adaptive engagement profiles (consumer-archetype communication preferences)
Visit Agenda (AI-generated pre-visit plans)
Simplify My Results (contextualized test explanations for lab results)
Health Coach functions (general wellness recommendations and coaching)
However, attendees felt that the timeline for voice-based interactions remains unclear. Epic was adamant that outbound call answer rates (hovering about 15-30%) and infrastructure costs make voice the hardest channel to get right - something that many executives agreed with. While Epic does support outbound phone calls today for appointment canceling, confirming, and rescheduling, building a full-scale voice platform that handles volume, nuance, and advances the case in the Epic system will take time.
What health system leaders really think
Once Epic left the session, the temperature in the room shifted. Executives were candid about what excited them – and what worries them – on the Emmie roadmap.
“Epic may be behind…but it doesn’t matter.”
Most leaders acknowledged that Emmie’s functionality trails existing AI agents, but they also noted that integration often matters more than innovation. As one executive put it, “Even if it doesn’t have every functionality or the most advanced model, the value from tight integration far exceeds that value gap.”
Based on our in-session poll, 60% of participants anticipate implementing Emmie at some point in 2026 – whether as an enthusiastic early adopter (19%) or an optimistic fast-follower later in the calendar year (41%). The next highest frequency answer (26%) implied additional hesitancy as members anticipated adopting a wait-and-see approach across 2026. Many hedged on pricing clarity, wondering whether Emmie will be included through existing Epic licenses or as a separate add-on. Combined with a desire to see how early implementation shapes up for other organizations, there remains significant considerations to adoption.
Integration is the sell. It’s also the risk.
Epic’s ecosystem works beautifully when organizations fully commit. When they don’t? Friction multiplies.
Leaders raised concerns about:
Unclear boundaries between foundational vs. paid add-on capabilities
Future reliance on a single vendor’s AI roadmap
Operational divergence across systems adopting at different speeds
A health system’s ability to stay up to date with Epic’s most recent versions and updates (e.g., HelloWorld, Cheers) will be critical to finding success with Emmie – and will only become more so as new features, like voice integration, are realized. One member put it bluntly when considering market-wide operational discrepancies, “This could get out of control really quickly.”
Voice, governance, and the human factor
Voice was the biggest dividing line. Some saw Epic’s cautious stance as smart; others felt they may be conceding ground to vendors already strong in conversational AI (and felt unwilling to wait on Epic to begin their own implementations).
More universal concerns about:
Governance overload as demand for AI workflows accelerates
The need to define ROI guardrails early
Ensuring equitable care for patients who prefer (or require) human interaction
“What about patient who won’t opt in?” one attendee asked. That question hung in the air.
What this means for health systems now
Emmie is not just another Epic feature, but a strategic fork in the road. For most systems, the question isn’t whether Emmie will be part of the future. Instead, they’re considering how much control they’re willing to cede to Epic to get there.
The early moves that matter most:
Start with measurable, low risk workflows. Canceling and rescheduling appointments, billing, and message deflection will give you the clearest operational and financial proof points.
Pair ROI expectations with the unglamorous backend work required to realize them. Tools like online scheduling only pay off after hard work aligning templates, negotiating with clinicians on access, and cleaning up workflows. Clear ROI on Emmie will depend on this kind of operational plumbing as much as on the technology itself.
Stand up governance before you begin. Without intake discipline, AI workflows can proliferate faster than teams can manage.
Don’t fall behind on Epic’s release cycles. Rapid iteration means the gap between “current” and “obsolete” will widen quickly.
Plan for hybrid models, not automation fantasies. Even at full maturity, Emmie will require oversight and patient choice will remain essential.
Define your organization’s 80:20 threshold for Epic, one use case at a time. Many leaders default to Epic once its functionality is about 80% of what competitors offer, given the value of native integration. However, maturity will vary across Emmie’s workflows, so set that bar deliberately for each use case.
Emmie’s long-term success won’t hinge on the model. It will hinge on workflow fidelity, executive discipline, and patient trust.
Questions to consider
Where would Emmie create real, measurable impact in the first 90 days at your organization? And where might it simply create noise?
What governance structure will determine which Emmie features to adopt, defer, or decline – and who owns those decisions?
What happens to patients who will not opt in? How will you maintain equitable access across digital and human channels?
Which parts of Emmie’s roadmap align with your strategy and which reflect Epic’s strategy more than yours?