AI outreach
Agents that call, text, and actually book the visit.
Natural voice calls and two-way SMS for every patient with an open care gap — in English or Spanish, inside your calling windows, with every outcome written back to the record.
Anatomy of a conversation
Not a robocall. A conversation that handles the follow-up question.
Patients don’t follow scripts. They ask about insurance, push back on timing, switch languages. The agent answers from approved conversation flows, books the visit, and saves the full transcript to the record.
- Voicemail detection. No awkward messages read to a beep — the agent knows when a human answered.
- Approved answers only. Every flow is reviewed by your team before a single call goes out.
- Escalation with context. Anything outside the flow routes to your staff with the transcript attached.
Hi, may I speak with Denise? I’m calling from her health center — she’s due for her mammogram screening.
That’s me. Will my insurance cover it, though?
Based on the coverage we have on file, screening mammograms are covered as preventive care — typically at no cost to you. Would you like me to schedule it?
Oh good. Yes — sometime next week?
Tuesday 11:20 AM at the women’s health center is open — you’re booked. I’ll text you the confirmation now.
How campaigns work
From “who has a gap” to “who has an appointment” in four steps
A campaign is something your quality team launches in an afternoon — not a project with a kickoff meeting.
Define the cohort
Start from any measure, report, or saved list on the unified record. “Patients overdue for an A1c at the east site, Spanish-preferred” is a filter, not a data request.
Approve the conversation flows
Your team reviews what the agent will say, ask, and answer — in every language — before anything sends. Template variables personalize each message from the record.
Agents call and text within your windows
Voice and SMS run inside business-hours calling windows, with voicemail detection, quiet hours, and capped attempts. The agent reschedules, answers, and books.
Outcomes write back, humans get escalations
Bookings, declines, and opt-outs land on the patient record automatically. Anything the agent can’t handle routes to your staff with the full transcript and context.
Guardrails
Built to be trusted with your patients’ phones
Outreach at scale only works if every conversation is respectful, logged, and easy to stop. These rules are enforced by the platform, not by policy documents.
Instant opt-out, quiet hours, attempt caps
Reply STOP or say it on a call — opt-out is honored immediately and recorded on the patient record. Calls stay inside configured hours, and attempts are capped per campaign.
Voicemail detection and respectful cadence
The agent knows when it reached a machine instead of a person. Follow-ups wait two days between touches — persistence without harassment.
Full transcripts and outcome logging
Every call and text thread is saved to the patient record, word for word. Your team can audit exactly what was said, to whom, and what came of it.
Multilingual with live switching
English and Spanish today — and when a patient switches languages mid-conversation, the agent switches with them instead of transferring or hanging up.
Semantic understanding, not keyword matching
“I work Tuesdays” means reschedule. “My daughter handles my appointments” means a different conversation. Replies are understood by meaning, not by hunting for YES or NO.
Human escalation with context
Clinical questions, billing disputes, anything outside the approved flow — the agent doesn’t guess. It hands off to your staff with the transcript and the patient’s record attached.
Your staff, supercharged
Not ready for full automation? Start with your own callers.
Most clinics don’t flip a switch to AI — and they shouldn’t have to. Your outreach staff get the same engine through a calling workspace built for speed: a deadline-ordered queue, the patient’s gaps, history, and language on one screen, and outcomes logged per measure in your team’s vocabulary.
- One queue, two workforces. Humans and AI agents pull from the same prioritized lists and log to the same record — switch any cohort between them without changing your data.
- Built for volume. Keyboard shortcuts, session goals, and pre-loaded patient context keep your callers in flow instead of in chart-hunting.
- Graduate on the evidence. Run staff and AI side by side, compare outcomes per measure, and expand automation only where it proves itself.
And when a cohort does move to AI, capacity stops being the constraint — one agent comfortably carries a mid-size clinic’s entire year of routine outreach, tens of thousands of calls and texts, without a hiring plan.
Outreach analytics
Know which gaps closed because you called
Every call and text — human or AI — logs an outcome against a specific measure. So outreach stops being a vibes-based activity and becomes an instrumented one: you see what each campaign closed, which channel works for which population, and where the next hour of effort pays most.
- Gap-level attribution. Outcomes tie to the measure they touched, so closure trends trace back to the outreach that drove them.
- Channel and cohort performance. Voice vs. text, language, site, time of day — see response patterns, not anecdotes.
- Staff and AI on one scoreboard. Same outcome vocabulary, same dashboards — compare honestly and staff accordingly.
Beyond reminders
The same agents do more than “you have an appointment”
Forms completed by text
Send a form link in the SMS thread — screenings, intake questions, pre-visit paperwork. Patients complete it from their phone and the responses land on the record.
Report-driven personalization
Campaigns pull from any measure or report, and template variables fill in the patient’s name, due service, site, and provider. Mass outreach that reads like it was written for one person.
Appointment scheduling in the conversation
The agent offers a concrete time and books it on the spot — not “call us back during business hours.” The visit is the outcome, not the referral to a phone queue.
No-show follow-up
A missed appointment triggers a follow-up conversation to find out what happened and rebook — the moment when a reminder is worth the most and staff have the least time.
Measured outcomes at our first deployment — a Los Angeles community health center. Read the case study. The need is national: across all 1,510 federally funded health centers, 5.1 million patients are overdue for colorectal cancer screening alone (HRSA UDS 2024) — see our research.
Questions
What teams ask before turning the agents loose
No. The agents use modern natural voices with realistic pacing — not the touch-tone robot you’re picturing. Outreach identifies itself as automated contact from your health center, and patients respond the way our mockups show: they ask follow-up questions, push back on times, and book visits. And you don’t have to take that on faith — every flow is approved by your team before it dials, and you can hear it live in your demo.
English and Spanish today, including live switching — if a patient answers in Spanish or changes language mid-conversation, the agent follows without transferring the call. We expand language coverage based on the populations our partners serve.
It doesn’t guess. The agent answers only from conversation flows your team approved. When a patient raises something outside that — a clinical question, a billing dispute, a complaint — the agent says a team member will follow up, and the case escalates to your staff with the full transcript and the patient’s context attached.
Reply STOP to any text, or say so during a call. Opt-out is honored instantly, recorded on the patient record, and respected across every campaign — it isn’t a per-campaign setting someone has to remember to check.
Your staff can review the full transcript of every call and text thread — each conversation is written back to the patient record, word for word, alongside its outcome. That makes auditing what the agent said as easy as reading a chart note, whether you’re reviewing one patient or a whole campaign.
See it on your own data
Watch our AI call a patient.
Then imagine it calling
thousands of yours.
A 30-minute demo: live AI outreach, your quality measures on a unified record, and an honest conversation about what we'd build for your workflows.