New · early demo

The AI receptionist for your clinic

The agent answers on Telegram and web chat, books, reschedules and cancels, sends reminders and upsells only with clinic-approved rules. Below, a live simulator that talks to the real booking engine.

Telegram + web chatRule-based upsellMeasured KPIsGDPR / EU

Early demo (phase M6): the simulator and AI agent run on the real engine (tenant “demo”), with Telegram channel and automations. Data is wiped on reset.

Create a demo for your clinic →Generate a demo tenant, then come back here with the code you receive.

Live simulator

Demo booking, step by step

Every step calls a real engine endpoint: lookup_patient, get_availability, book_appointment, reschedule, cancel.

engine: connecting…

1. Patient2. Service3. Day4. Time5. Confirm

Existing demo patientSearch the demo database (lookup_patient):

Or a new patient

First name

Last name

Phone (07…)

Email (optional)

All requests go to /v1/actions/… through the clinic-agent engine.

What the product includes

Not a chatbot — a receptionist

Built in small, verifiable phases with KPIs measured from the first pilot week.

OmnichannelWhatsApp, Telegram and web. The patient picks the channel; the agent keeps the same memory and rules across all of them.

Rule-based upsellOnly clinic-approved offers (e.g. air-flow after scaling), max 1 per conversation, never during urgency, permanent opt-out.

Measured KPIsBookings created, reminders sent, no-shows avoided, revenue added by upsell — monthly report with numbers.

GDPR & EUData processed in the EU, minimization (art. 9), logged consent, AI disclosure, append-only audit, export/erase.

Automations

Scheduled jobs — the invisible engine

n8n triggers the crons; the logic runs deterministically in clinic-agent (Postgres), deduplicated on the audit log. Every action shows up in the panel’s Events tab.

reminder_t24hReminder 24h before the visit, sent on the patient’s channel.

no_show_recoveryAppointments past by 30 min without attendance → no-show + recovery message.

waitlist_fillerCancellation in the last 24h → the freed slot is offered to the next waitlist patient (atomic claim, 1 offer per slot).

recall_6mPatient hasn’t visited in ~6 months → periodic check-up invitation.

Controlled upsell

The agent never invents offers

Rules live in the database (addon_rules), editable by the clinic: trigger service → offered service, with exact price and duration from the DB.

Deterministic evaluation, no LLMUrgency → zero upsell. Max 1 offer/conversation, 90-day cooldown, permanent opt-out per category.

Accept or decline, both logged“Yes” → the service is added to the same visit (staff overlap check). “No” → opt-out, never offered again.

Full auditupsell_offered / upsell_opt_out + tool calls visible in Events, with the conversation context.

KPI

Pilot numbers live in the audit log

Every action is logged append-only in clinic.events — the pilot KPIs are computed from it:

bookings

created, confirmed, cancelled

reschedules

rescheduled + freed slots

reminders

sent / delivered on channel

no-show recovery

marked + recovery message

waitlist filled

slots filled from the waitlist

upsell offered

offers sent, from DB rules

upsell accepted

services added to the visit

estimated revenue

DB prices, on confirmed bookings

The dedicated KPI dashboard is on the roadmap — the data already exists in the audit log.

Roadmap

Built in verifiable phases

Every phase has a concrete verification gate. You see the product growing, not promises.

M1Booking engine — multi-tenant Postgres, RLS, slot anti-duplicationdone

M2Showcase + live simulator (this page)done

M3AI agent: dialogue, tools, urgency triagedone

M4Telegram channel — first real conversationdone

M5Automations: T-24h reminders, waitlist, recalldone

M6Upsell engine with rules + KPIdone

M7WhatsApp Cloud API — real channelplanned

Competitive baseline (feature parity)

What a clinic expects from an AI receptionist — and where we stand:

inboxInbox + human takeoverin progress

The front desk sees the conversation live in the panel; the AI marks handed_off (urgency, “I want a person”). Manual replies on the same channel come next.

kpiKPI dashboardup next

Bookings, cancellations, reschedules, reminders, no-show recovery, waitlist, upsell, estimated revenue — from the append-only audit.

setupClinic setupplanned

Services, prices, staff, schedule, booking rules, tone, approved messages — clinic-configurable (data already in the DB; UI next).

notifyNotificationsplanned

Alert on handoff, new booking, urgency, job failure — events are logged, real-time notification comes next.

channelsChannelsin progress

Stable Telegram + web chat; then WhatsApp (M7), SMS, voice.

auditAudit / QAdone

Reviewable conversations, visible tool calls, handoff reasons, test scenarios on the gate.

Your clinic could be next

Let’s talk about a pilot: real channel (Telegram/WhatsApp), reminders and the first upsell rule, with KPIs measured from week one.

Product developed by Julien's IT Services · demo data: tenant “demo” · engine: clinic-agent (Postgres + RLS)