Source: GlobeNewswire — WellSpan Health / Hippocratic AI — 2026-07-30
Summary
WellSpan Health expanded its partnership with Hippocratic AI from a set of individual use cases into a platform-wide, multi-year co-development agreement, with a dedicated Hippocratic AI team now embedded at WellSpan's York, PA campus. "Ana," the generative-AI voice agent the two built together, already handles more than 160,000 patient calls and 7,000+ hours of conversation per month — previously limited to inbound calls and primary-care scheduling. The expanded deal moves Ana into post-discharge follow-up and chronic-disease check-ins, with the first new workflow targeting outreach to patients who missed imaging appointments, extending the agent across inbound, outbound, ambulatory, and inpatient settings.
Key Takeaways
- 160,000+ calls and 7,000+ hours of conversation per month is a real production usage number, not a pilot metric — this is one of the largest documented deployments of a voice AI agent inside a U.S. health system.
- The structural change matters more than the volume: going from "individual use cases" to a "platform-wide, multi-year co-development" deal signals WellSpan is treating Ana as core infrastructure, not an experiment, complete with a vendor team embedded on-site.
- The first new workflow — proactively calling patients who missed imaging appointments — is a concrete, high-value target: missed imaging follow-ups are a known driver of delayed diagnoses, so this is closing a specific, costly care gap rather than adding a generic feature.
- Expanding from inbound-only to inbound-and-outbound, and from scheduling-only to post-discharge and chronic-disease check-ins, is the difference between a call-center replacement and a proactive care-coordination tool — a materially bigger scope of responsibility for the agent.
Reel Script
Hook (~16s, 36 words): A health system's AI voice agent already handles over 160,000 patient calls a month. Instead of stopping there, it just got a multi-year mandate to start calling patients who missed appointments before anyone notices.
Core Concept (~55s, 125 words): "Ana" started as an inbound-only tool: patients called in, and the agent answered questions or booked a primary-care appointment. The expansion flips that model to include outbound calls — Ana proactively reaching out — and extends its job from scheduling into post-discharge follow-up and chronic-disease check-ins. That's a meaningful scope change: inbound agents just need to handle whatever the caller brings up, but an outbound, proactive agent needs to know who to call and why, which means it has to be wired into the health system's actual clinical data — who missed an appointment, who was just discharged, who's due for a chronic-condition check-in — not just answer a phone line.
Hands-On (~50s, 115 words): The concrete numbers: 160,000+ patient calls and 7,000+ hours of conversation every month, currently. The first new outbound workflow specifically targets patients who missed an imaging appointment — a well-known gap where a delayed follow-up can mean a delayed diagnosis. Worth sketching as a simple expansion diagram: a box labeled "Ana v1 — inbound calls, scheduling only" growing into a wider box labeled "Ana v2 — inbound + outbound, post-discharge, chronic-disease check-ins, missed-imaging outreach," with the 160,000-calls-per-month number sitting underneath as the base this expansion is building on.
Takeaway (~22s, 48 words): The real signal here isn't the call volume, it's the shift from reactive to proactive — an agent that reaches out before a problem compounds is worth more than one that just answers the phone faster. Health systems evaluating AI voice agents should be asking about outbound capability now, not later.