Deploying a voice AI agent in a medical practice is not a plug-and-play installation. Each specialty has its own rhythm: appointment durations, urgency reasons, call filtering rules, integration with practice management software. Generic configuration produces average results — specialty-tuned configuration moves the tool from useful to indispensable. Here's how Vocalis AI adapts to the main specialties, while strictly staying in the administrative domain.
Principles common to all specialties
- AI announcement: call start = "Hello, I am the AI voice assistant for Dr X" (AI Act Article 50 obligation)
- No diagnosis, no advice: explicit refusal, human transfer or emergency routing
- Urgency detection: systematic keywords (chest pain, loss of consciousness, hemorrhage, suicidal ideation) → emergency services
- Minimized data: name, contact, slot, administrative context. No detailed clinical reason
- Human transfer at any time upon patient request
General practitioner
Average duration: 15-20 min | Daily call volume: 40-80 | Typical no-show: 12-18%
Common administrative reasons: first consultation, follow-up, check-up, prescription renewal, medical certificate, vaccination, home visit.
Configuration rules:
- Default 15-min block, 30-min for 1st consultation or annual check
- Daily urgency slot (8-9 AM, 5-6 PM) reserved for acute reasons
- Home visit: collect address, administrative reason, transfer to doctor for validation
- Automatic urgent detection → emergency services immediately
- Prescription renewal: no direct renewal by agent. Follow-up appointment mandatory.
Dentist
Duration: 30-60 min | Volume: 30-60 calls/day | No-show: 15-25%
Common reasons: consultation, cleaning, cavity treatment, prosthesis, orthodontics, dental emergency, check-up.
Rules:
- Variable durations: cleaning 30 min, treatment 45 min, prosthesis 60 min, emergency 20 min
- Agent asks administrative reason (first visit/follow-up/emergency) to block appropriate duration
- D-1 reminder mandatory by SMS + brief call (>20% no-show without reminder)
- Dental emergency (tooth pain, trauma): same-day slot if available, otherwise routing to emergency dentist
- Prosthesis quote: agent never gives verbal quote. Schedule quote appointment or email request.
Physiotherapist
Duration: 30 min | Volume: 20-40 calls/day | No-show: 10-15%
Reasons: 1st consultation with prescription, session series, post-op follow-up, sports rehab.
Rules:
- Specific: session series. Agent manages full series planning (10-15 sessions) in single conversation
- Request medical prescription at start (mandatory administrative reason for reimbursement)
- D-2 vs D-1 cancellation policy configurable per practice
- Weekly reminder for ongoing series
- End-of-prescription assessment automatically scheduled
Ophthalmologist
Duration: 30-45 min | Volume: 50-100 calls/day | No-show: 22-28%
Reasons: routine, annual check, urgency (red eye, vision loss), post-op follow-up, AMD/glaucoma screening.
Rules:
- Structural long waits (3-6 months for routine) → automatic waiting list critical
- Distinguish reasons: routine, check, post-op follow-up, visual emergency
- Visual emergency: same-day urgent slot mandatory or emergency ophthalmology routing
- Waiting list patients: SMS when slot frees <72h
- Post-op follow-up systematically scheduled D+1, D+7, D+30 for cataract/refractive surgery
Dermatologist
Duration: 15-30 min | Volume: 40-80 calls/day | No-show: 20-25%
Reasons: consultation, mole check, acne, eczema, dermato-aesthetic follow-up, urgency (suspicious lesion).
Rules:
- Administrative distinction routine consultation vs evolving mole check (urgent slot <2 weeks)
- Agent never decides on mole urgency — proposes both options
- Dermato-aesthetic (laser, peelings, botox): consultation MANDATORY before any procedure
Aesthetic medicine
Duration: 30-90 min | Volume: 20-50 calls/day | No-show: 10-15%
Reasons: quote consultation, procedure, post-procedure follow-up, touch-up.
Rules:
- Consultation MANDATORY before any 1st procedure
- Agent gives no advice on procedures
- Variable durations: consultation 30 min, botox 30 min, hyaluronic acid 45-60 min, peeling 60-90 min
- Post-procedure follow-up automatically scheduled D+15 and D+30
- Reflection period reminder: for 1st procedure, legal 15-day delay between quote and procedure
Psychiatry & psychology
Duration: 45-60 min | Volume: 15-40 calls/day | No-show: 20-30%
Rules:
- Calibrated empathy: soothing tone, slower pace, reassuring phrasing
- Distress/crisis detection: suicidal ideation → immediate routing to local suicide prevention hotline
- No clinical message-taking
- In-office payment
- Late cancellation: configurable policy
Pediatrics
Duration: 20-30 min | Volume: 30-60 calls/day | No-show: 10-15%
Rules:
- Agent dialogues with parent (never child)
- Daily urgency slot for acute pediatrics
- Detection of vital pediatric emergency: fever >40°C, seizure, loss of consciousness → emergency services immediately
- Vaccination follow-up: automatic reminders per pediatric vaccination calendar
Cross-specialty rules: what the agent NEVER does
Whatever the specialty, the Vocalis AI voice agent has non-negotiable prohibitions:
- No diagnosis, even by allusion
- No prescription, no direct renewal
- No medical triage
- No therapeutic advice
- No verbal quote for aesthetic or prosthesis procedures
- No exam result communication
- No sensitive health data storage without validated certified hosting
Vocalis AI configuration process
- D1 — Briefing: practice type, specialties, hours, administrative reasons, cancellation rules, agenda integration
- D2-3 — Configuration: durations, urgency keywords, transfer rules, announcement scripts
- D4 — Tests: 20 standard scenarios + 10 edge cases with practice
- D5 — DPIA validation: review with DPO, sub-processor contract
- D7 — Go Live: redirect secondary line first, then main line after 1 incident-free week
"Generic configuration delivers 60% of the value. Specialty configuration delivers 95%. The difference is the initial briefing — 3 hours with practitioner and reception team are worth all the post-launch tweaks." — Vocalis AI deployment team
Conclusion: configuration IS the product
A voice AI agent in a medical practice is not a "box" product but a tool tuned to measure. The underlying technology is the same for everyone. Differentiation happens in the rules: durations, urgencies, administrative reasons, vertical integrations. Bad configuration produces a frustrating agent. Good configuration produces a tool no one would want to remove.
For more, see our medical voice AI pillar, the GDPR healthcare guide, and no-show impact analysis. General framework: home, inbound calls pillar.