Books appointments and consultations
It offers the open slots, holds the chosen one, confirms with the patient and updates the schedule. Rescheduling and cancellations follow the same path.
Clinics, private practices and dental offices
A virtual receptionist on your clinic's WhatsApp
It books the appointment, confirms the time, reschedules whoever needs it and answers the patient at eleven at night. Then, at the end of the day, it reports back to you: who booked, who canceled, who was a no-show and what is still open.
Conversations run on Meta's official WhatsApp, through text, audio and images. Your front desk can step in whenever it wants.
The real problem
Your front desk can't keep up with WhatsApp — and a chatbot doesn't help
The message arrives while the receptionist is helping someone at the counter. The patient asks again, gets no answer, and disappears. The next day they miss the appointment and nobody noticed. A decision-tree chatbot doesn't fix that: it replies to the patient, but it does nothing for the operation.
Decision-tree chatbot
- Depends on a pre-drawn flow and stalls the moment the patient goes off script
- Doesn't confirm the appointment or remind the patient the day before
- Has no idea what happened yesterday with that same person
- Talks to the patient, but never talks to the practice owner
- Frustration only surfaces once it has become a public review
Hal-AI virtual receptionist
- Open conversation: it understands the request even with no script in sight
- Reminder the day before and confirmation on the day, to cut no-shows
- Memory and context: it picks the conversation up exactly where it stopped
- Recognizes the owner's phone number and switches roles — with you, it reports
- Flags complaints and open items the same day, so the clinic can act
This is not a chatbot with a flow. It is an operational agent trained to work.
On the patient's side
What it handles while the clinic is closed
Patients write whenever it works for them. The virtual receptionist answers within the minute, with your clinic's context — specialties, accepted insurance plans, payment options and everything published on your website.
Confirms appointments on its own
A reminder 24 hours ahead and a confirmation on the day. It is the work nobody at the front desk has time for, and the work that keeps your no-show rate down.
Answers the same questions, every time
Procedures offered, insurance plans accepted, payment options, address and opening hours — using the information the clinic registered.
Hands off to a person
It spots the sensitive case and transfers it to your team with the conversation history attached. The agent goes quiet in that chat until the person hands it back.
Understands audio and images
Older patients tend to send voice messages. It listens, understands and answers. It can also open the conversation with a short voice message introducing the clinic.
Stays on duty
24 hours a day on Meta's official WhatsApp. Saturdays, Sundays and the lunch hour — which is exactly when most people get a chance to write.
On the management side
It serves the patient and reports back to you
The same agent recognizes the phone number of whoever runs the clinic and switches roles. You ask in plain English, on your own WhatsApp, without opening a dashboard.
- End-of-day summary: bookings, cancellations, no-shows and anything worth flagging, sent when the day closes.
- Complaints and open items: it surfaces frustration while it is still a conversation, not a public review.
- Operational reminders: "tell me when supply X is running low," "remind me to call the lab."
- Factual answers: specialties, procedures and policies published by the clinic, along with the source it consulted.
- Honesty about what it doesn't know: until your clinic's system is connected, it states explicitly that a given number is a simulation — it never pretends to be integrated.
Clinic receptionist
online
Good evening. I'd like to book a consultation — are you open on Saturdays?10:41 PM
Good evening! We're open Saturday mornings. I have 9:00 and 10:30 this Saturday. Does either one work for you?10:41 PM
Voice message sent by the patient, nine seconds long.
Got your message: Saturday at 10:30, under the name Helena. The slot is held. I'll send the confirmation on Friday and a reminder Saturday morning.10:42 PM
Do you take my insurance plan?10:43 PM
We do — it's on our list. I'll ask the front desk to confirm coverage for this procedure first thing tomorrow and let you know right here.10:43 PM
Clinic receptionist
online
How did the schedule go today?7:10 PM
Busy afternoon. Two people asked to reschedule and I already fit both of them back in this week.7:10 PM
One patient complained about yesterday's wait time. I didn't answer on your behalf: I left the chat in the human queue.7:11 PM
What's still open?7:12 PM
Three of tomorrow's confirmations haven't come back. I'll send the reminder at 8:00 AM. And the supply you asked me to watch goes into Friday's alert.7:12 PM
The content depends on what each clinic configures in its agent.
The virtual receptionist's schedule
Today's confirmations, open reschedule requests and the queue of people still waiting to hear back.
New task
Tomorrow's reminders
Preparing
schedule_task → 21 slots tomorrow: 19 reminders queued, 2 already confirmed
Today's confirmations — 8 of the 24 slots
| Time | Patient | Appointment | Channel | State |
|---|---|---|---|---|
| 8:00 AM | Helena Prado | Consultation | Confirmed | |
| 8:40 AM | Otavio Ramos | Follow-up | Confirmed | |
| 9:20 AM | Bianca Sales | First visit | Awaiting reply | |
| 10:00 AM | Ivo Nogueira | Follow-up | Webchat | Reschedule |
| 10:40 AM | Celia Marques | Consultation | Confirmed | |
| 11:20 AM | Renato Bastos | Follow-up | No reply | |
| 1:30 PM | Amanda Vieira | First visit | Confirmed | |
| 2:10 PM | Sergio Lima | Follow-up | Phone | Canceled |
| 24 slots today = 17 confirmed + 6 pending + 1 canceled. The 6 pending are 2 awaiting a reply, 2 with no reply and 2 to reschedule. | ||||
Open reschedules — 2 open
Ivo Nogueira
Asked to move today's appointment. The 10:00 AM slot went back to the waiting list.
Thu · 9:20 AM
Fri · 2:10 PM
Sat · no openings
2 options offered 12 min ago
Douglas Prata
Wrote last night to say he can't make it that early. The 8:00 AM reminder was put on hold for this contact.
Fri · 4:40 PM
Sat · 9:00 AM
waiting for his pick
Contact queue — 5 patients
Who is still waiting for an answer
Renato Bastos
11:20 AM
no reply since yesterday · second reminder pending
Bianca Sales
9:12 AM
today's confirmation still open
2
Douglas Prata
yesterday
wants to move tomorrow's early slot
Marisa Fontes
yesterday
complained about the wait at the front desk
1
Ivo Nogueira
8:55 AM
choosing between the two slots offered
AI paused automatically — Marisa Fontes's conversation was taken over by Marina (Front desk)
Release AI
Clinic system
Simulation
with no system connected, every number on this screen is stamped as a simulation
What you are looking at
It's the front desk's whole day on one screen: today's confirmation list, with every slot marked as confirmed, awaiting reply, no reply, to reschedule or canceled. Next to it sit the two open reschedules — with the slots the agent has already offered the patient — and the contact queue, which gathers everyone still waiting for an answer. Up top, the recurring task that fires tomorrow's reminders at 8:00 AM.
The advantage
Nobody had to call twenty-four people to find out who is coming: the confirmations arrived through the conversation and turned themselves into a state on the list. What's left on the screen is exactly what needs a human decision — six open items and two reschedules — instead of all twenty-four rows. And when someone at the front desk takes over a chat, the agent goes quiet in that conversation and the amber banner records who stepped in.
Only at Hal-AI
Notice what the screen does not show: no clinical data. The columns are time, name, type of appointment, channel and state — no symptom, no history, no test result, because the default configuration does not touch medical records. Any link to a clinical system depends on a specific project and a contract, with purpose, legal basis and scope of access defined up front. And until the clinic's system is connected, the screen itself stamps the numbers as a simulation instead of pretending to be integrated.
Names, times and counts are fictitious. The same list the front desk sees here is what becomes the end-of-day summary on your WhatsApp — you only open the screen when you want the detail.
Hal-AI creative
The same agent, the vocabulary of your operation
What changes from one clinic to the next isn't the technology: it's the agent's personality, the knowledge it consults and the tools it has at hand.
What it knows by heart
The agent is pre-trained on the clinic's website and rounded out by the Knowledge Vault: PDFs, text and images covering procedures, insurance plans and policies live in the base it consults before answering.
What it looks up in real time
Your appointment book and clinic systems connect through APIs: the endpoint is registered once and becomes a tool the agent can use. The integration happens at activation, with the scope you define.
In practice
The questions it already answers
The patient on one side, you on the other. Same virtual receptionist, same phone number.
The patient asks
- I'd like to book an appointment.
- Are you open on Saturdays?
- What times do you have available?
- I'd like to schedule a consultation.
- How many bookings did we get today?
- Did any patient complain?
- What's still open?
- Remind me when supply X runs low.
How to turn it on
From your clinic's website to a receptionist that's live
Pre-training starts with the clinic's own website. Whatever isn't there, you add.
Point it at your website
The agent reads what you've published — specialties, procedures, address, hours — and starts out already able to talk about your operation.
Fill in the gaps
Insurance plans accepted, payment options, rescheduling rules and the tone you want it to use with patients.
Connect official WhatsApp
The number is activated through Meta's own authorization flow. Hal-AI is a Meta Business Partner and runs on the Official WhatsApp Business Cloud API.
Decide who steps in
You define which situations send a conversation to a person and who receives the end-of-day summary. Human takeover is available the whole time.
Transparency
What the virtual receptionist does not do
Healthcare demands clear boundaries. Better to say this before you sign than after.
It gives no clinical advice
The agent does not assess symptoms, suggest a diagnosis, recommend treatment or weigh in on medication. It handles the front desk: booking, confirming, rescheduling and sharing what the clinic has published. Anything clinical goes to the human team with the context of the conversation.
It does not process health data by default
Medical records, clinical history and test results are sensitive personal data under Brazil's LGPD. The default configuration does not touch them. Any integration with a clinical system requires a dedicated project, with purpose, legal basis, access scope and responsibilities set out in a contract before any connection is made.
It does not invent numbers
Until your clinic's system is connected, the agent states explicitly when an operational figure is a simulation. It never claims to be integrated with a system it isn't connected to.
FAQ
Questions from the people who run the clinic
Does the virtual receptionist give patients medical advice?
No. It handles the front desk: booking, confirming, rescheduling and answering questions about hours, insurance plans, payment options and information the clinic has published. Anything clinical is passed to the human team.
Does the agent access patient records or health data?
Not by default. The initial setup works with what the clinic has published and with the appointment book. Access to a clinical system requires a dedicated project and a contract, with the purpose and LGPD legal basis agreed before any connection is made.
Is the number used an official WhatsApp number?
Yes. Hal-AI is a Meta Business Partner and runs on the Official WhatsApp Business Cloud API. The clinic's number is activated through Meta's own authorization flow.
Does the front desk still handle conversations?
It does. Human takeover lets a staff member step into any conversation; the agent goes quiet in that chat and the person picks up with the full history in view. The virtual receptionist takes the repetitive work off the front desk, not the decisions.
What if the patient sends a voice message?
It understands them. The channel handles text, audio and images, and the agent can open a conversation with a short voice message introducing the clinic's services.
Does it work for a solo practice?
It does. A solo practice, a dental office, a small or midsize clinic — the more WhatsApp is your main channel, the more work it takes off your hands.
Your clinic answering after hours, too
Turn on the virtual receptionist on your clinic's WhatsApp and watch the end-of-day summary arrive on your own phone.