A worried or unwell person calls a healthcare line. They connect with an agent, and may need to read out their member ID or a complex medication name. If the agent sounds off, talks over them, or mishears them, the caller may hang up. For an agent to work in healthcare settings, callers have to trust it, and the team that deployed it has to be able to control what it says and does.
ElevenAgents for Healthcare helps healthcare companies earn that trust. Agents can now use Scribe v3 Live Medical, our first real-time medical transcription for agents, to accurately transcribe clinical language as people speak. With ElevenAgents, healthcare teams decide how every agent behaves, what they can say and do, and when they hand a call to a person.
What teams are building with ElevenAgents for Healthcare
Patient calls for health systems
At Banner Health, one of the largest nonprofit health systems in the US, with 33 hospitals and more than 400 clinics, agents answer inbound patient calls around the clock. Patients can book, reschedule, and cancel primary care appointments directly in Banner’s EHR without waiting on hold.
Cadence, which monitors patients with chronic conditions, uses agents to call patients when Cadence’s monitoring systems detect that a vital sign reading has come back as abnormal. The median time to reach a patient fell from nearly two hours to 3.5 minutes, and patients answer the agent’s calls 47% more often than calls from nurses.
Provider calls for payer and pharma teams
Angle Health, which administers health plans for small and midsize employers, uses agents on its provider and member lines. Agents now resolve more than 90% of provider calls they handled without a transfer to a person.
We work with teams at Pfizer, AstraZeneca, and several of the world’s largest pharmaceutical and health insurance companies.
Agents that understand the context that matters
ElevenLabs already offers medical transcription for recorded audio in clinical settings. Scribe v3 Live Medical brings that same capability to live calls in ElevenAgents. Built on Scribe v3 Live, it’s optimized for drug names and clinical terms. On doctor-patient conversations in the PriMock57 benchmark, it had the lowest word error rate of the medical models we tested, with 17% fewer errors than the next closest one.
On Eka’s dataset of more than 3,600 English medical recordings, Scribe v3 Live Medical makes 4.7% fewer errors than Scribe v3 Live, our general real-time model (11% vs. 11.54%).
Note: These results come from ElevenLabs’ internal testing, on a selection of real-time models, each in its best available configuration. They don’t include every model available for medical transcription.
If an agent doesn’t catch a word the first time, the caller has to repeat it, and the call slows down. Teams can now give the agent up to 10,000 terms of their own, like the names of their plans, therapies, devices, and locations. And if a caller gives their member ID in English and then explains their question in Spanish, Scribe v3 Live Medical follows the switch.
Effective communication, from tone to timing
WithEleven v4 Turbo, an agent can stay steady with someone frustrated about a bill, and slow down for someone anxious about an upcoming procedure. The agent also knows when to wait. Ourturn-taking model can tell when a caller has finished talking and when they’ve only paused to find their insurance card, so the agent doesn’t cut in. Teams can also adjust how long it waits before responding. A clinic primarily serving older callers, for example, could build in more time for its patients to finish a thought.
As we also build the speech recognition, turn-taking, and voice models ourselves, we can tune them as one system and keep response times low. Faster agents also hold more successful conversations: In a study of ElevenAgents deployments, we were able to reduce end-to-end latency by an average of 16%, which correlated to a 26% lift in conversation success rate. For healthcare teams, faster response times can mean more appointments booked or more medical information requests resolved on the first call.
Appointments booked and claims checked on the first call
Picture a patient who calls to book a 3D mammogram at the clinic closest to them, ideally on a Saturday. Scribe v3 Live Medical helps the agent hear the request correctly, but booking the appointment requires access to your systems. ElevenAgents connects to the scheduling, electronic health records (EHR), CRM, and phone systems you already use. The agent can then find an open Saturday slot at the nearest clinic and confirm it before the patient hangs up. The same connections let an agent check a claim for a provider’s office or look up the status of a prior authorization.
Some healthcare organizations run on legacy systems or custom-built software. Ourforward-deployed engineers can work with your team to connect those systems too. With routine calls handled from start to finish, your staff can spend their time on the more complex calls that need a person.
Safe to deploy in healthcare settings
Say the patient booking that mammogram asks whether a symptom is worth worrying about. That question belongs with a clinician. With ElevenAgents for Healthcare, the agent can transfer the call to your clinical team, along with what was said, so neither the patient nor your staff has to start over.
Your team decides where each agent’s job ends. Built-in guardrails keep an agent on topic, and custom guardrails let you write your own rules in plain language, like sending any question about symptoms to a person. Workflows can also require a step before a sensitive action, such as confirming a caller’s identity before discussing a record or a claim. Before launch, your team can run simulated calls to check that the agent follows your rules.
ElevenAgents supports HIPAA-eligible deployments for Enterprise customers who sign a BAA and turn on Zero Retention Mode, so audio and transcripts aren’t stored on ElevenLabs infrastructure. Teams that need a record of each call can send it straight to their own systems.
Faster paths to working agents
At our largest healthcare customers, forward-deployed engineers embed with internal teams to build and launch production agents, then help expand them across the organization. Other teams would rather build their own. Templates, like our Healthcare Receptionist, give those teams a head start. The system prompt, call flow, and integration setup are already in place, so they can have a working agent in minutes. Then they connect their own systems and customize the template for their organization.
ElevenAgents Architect, our built-in AI assistant, makes building from a template, or from scratch, even easier. Teams describe what they want in plain language — like, “offer a telehealth visit when the next in-person slot is more than a week out” — and Architect edits the agent and tests the change before it goes live.
Agents that are yours to own and configure
Say a health system has a scheduling agent running and wants to add one for billing questions. A solution built for scheduling may not cover billing, which can mean onboarding another vendor, with another security review, BAA, and integration project. ElevenAgents is flexible enough to handle billing and additional use cases across channels. Each new agent reuses the integrations and approvals already in place.
Every agent can be managed by you. Your team can change and test any step of an agent without waiting on a vendor. When you want help, our engineers can build for you. You can also monitor every agent in one place making it easier for you to align agent configurations to your organization’s governance and protocols.
Get started
Visit ElevenAgents for Healthcare, read our report, “The State of Conversational Agents in Healthcare,” or talk to our team about your deployment.












