Safe conversational triage
The agent asks for the reason, perceived urgency and basic details. It does not diagnose. If it detects urgency signals (chest pain, bleeding, etc.), it escalates immediately to a human with priority flagged.
We design AI systems that qualify the patient via WhatsApp or web, route them to the right specialist, confirm the appointment and reduce no-shows. All with appropriate clinical language and a handoff to the team when needed.
In a clinic or medical practice, the first impression is the phone or WhatsApp. If the patient doesn't get a clear answer in under 5 minutes, they go to the competition — and it's usually the competitor with worse clinical care but a better digital front door.
Our AI agents for clinics do what an expert receptionist would: they listen to the reason for the visit, ask just enough for initial triage (without diagnosing), route to the right specialist based on schedule and insurance, book the appointment in your system and send reminders. All in neutral Spanish or English, with appropriate clinical terminology.
We build with the awareness that healthcare is a regulated sector: we don't give diagnoses, we don't recommend treatments, we comply with opt-in for communications and we maintain the traceability that HIPAA-adjacent and LFPDPPP require. When a patient shows a sign of urgency, the agent escalates to a human without asking.
The agent asks for the reason, perceived urgency and basic details. It does not diagnose. If it detects urgency signals (chest pain, bleeding, etc.), it escalates immediately to a human with priority flagged.
Read and write to your EHR, Jane, Kareo, DrChrono, Cliniko or Google Calendar. No double bookings. The patient receives the location, pre-visit instructions and a medical history form in the same chat.
48h ahead: reminder. 24h ahead: request for confirmation. If there's no response, it attempts an AI voice call. If they cancel, it frees the slot and offers it to the waitlist within minutes.
At 6 or 12 months, the agent reactivates patients who haven't returned: annual checkup, treatment follow-up, screening offer. A recurring pipeline that used to sit in a forgotten spreadsheet.
A high-volume clinic with a complex schedule and multiple specialists. We built automated triage that qualifies the patient, routes to the right specialist and manages confirmation.
A dental clinic with 4 chairs was losing bookings in Instagram DMs on weekends. We implemented an AI agent with confirmation + cleaning prepayment via Stripe.
A practice with 2 general physicians and heavy chronic-condition follow-up. We automated recall of inactive patients and reminders for pending lab work.
For clinics in the US we work with BAA-signed providers (WhatsApp Business API with WPP BAA, Twilio BAA, Anthropic enterprise). In LATAM we comply with each country's data law (LFPDPPP in Mexico, Law 1581 in Colombia). We don't store medical records in the agent — only what's needed to schedule.
The agent has escalation criteria defined with you in discovery (chest pain, active bleeding, suicidal ideation, severe allergic reaction). As soon as it detects one, it sends emergency instructions (911 or a local line) and immediately alerts staff with top priority.
Yes — we offer a supervisor mode where every agent response goes through human approval before being sent. It's useful for the first 2–4 weeks and in complex cases. After that, most clinics disable it for routine cases.
No. It's explicitly prompted not to diagnose, not to prescribe and not to interpret symptoms. Its role is administrative triage: it captures the reason, prioritizes and schedules. Any attempt by the patient to ask for a diagnosis triggers a handoff to the clinical team.
Yes. We integrate natively with Jane, Kareo, DrChrono, Cliniko and SimplePractice. For enterprise EHRs (Epic, Cerner) we use HL7/FHIR — this requires a custom project and IT team approval. For practices without an EHR, we set up Google Calendar + Airtable as an initial backbone.
Yes. In discovery we map: accepted insurance (PPO, HMO, Medicaid, cash), copays by visit type, prices for private services, packages. The agent responds accurately and updates whenever you edit the table in the dashboard.
Between 15% and 30% depending on the baseline. The biggest impact comes from the 24h reconfirmation with a fallback to an AI voice call if there's no response. For high-value appointments (surgeries, aesthetic treatments), automated deposits cut no-shows another 40%+.
Always available. The patient can ask to 'talk to a human' at any time and the agent hands off immediately to staff with full context. Typical escalation in clinics: 10–15% of conversations in the first month, dropping to 6–8% with continuous tuning.
Free 30-minute audit. We leave with a clear plan for clinics like yours. No commitment.
Book an audit