One in five booked appointments never happens — and the message that would have booked the next one arrived at 10 p.m., when no one was at the front desk. The assistant answers at any hour, triages, books into the real schedule, confirms and sends reminders. It also drafts the visit note, which still goes out signed by the clinician who saw the patient.
This is not a matter of style: Resolution 2,454/2026 of the CFM (Brazil's Federal Council of Medicine) forbids delegating to artificial intelligence the communication of a diagnosis, a prognosis or a treatment decision. The product was built inside that line.
It isn't logged as an expense, it has no invoice, it never reaches the report. It just vanishes from revenue.
20–30%
in clinics with no active confirmation process. One municipal network in São Paulo recorded 21.4% in 2026.
15–30%
in Brazilian clinics and diagnostic centers. For a clinic billing R$ 60,000 a month, that is tens of thousands of reais every month.
87%
it is the channel where appointments are booked, confirmed and canceled — and where most messages arrive after hours.
Sources: Panorama das Clínicas e Hospitais (Doctoralia) · no-show surveys of Brazilian clinics and diagnostic centers. These are industry figures, not ours — on purpose: you can check them without having to take our word for it.
The math on what the empty chair takes every month, with your own numbers, and what to compare the price against. See the math
You can have your own server, domain and brand. It costs more, and it has a consequence we tell you about up front. See both options
Who signs, what the AI is not allowed to do, where the data lives, and what happens if it doesn't work. See all twelve
Good evening, do you take my insurance? I need an appointment this week.
10:14 p.m. · no reply until 8 a.m.
Which doctor should I book with? My knee has been hurting and it won't go away.
7:41 a.m. · the front desk can't triage
Was I able to reschedule? I sent a message yesterday and nobody answered.
9:03 a.m. · already booked somewhere else
With the assistant, all three would have been answered the minute they arrived — and two would have ended with a time booked.
And there is a third pain that shows up on nobody's phone: the medical record written from memory, at the end of the day, for six patients.
The split is neither subtle nor negotiable, and it organizes the entire product. On the left, nothing needs you. On the right, nothing happens without you.
The assistant answers the clinic's WhatsApp using the clinic's own material — insurance plans, address, hours, who sees what — and when the conversation reaches the point of booking, it checks the real schedule and offers slots that exist.
The front desk stops repeating the address and goes back to caring for the person standing in front of it.
This is not a recommendation written into its instructions. It is the design: it does not have the tools to do any of this, and the rule holds even when the patient asks point-blank.
Not as a hypothesis, not as “it could be”. It listens to the account, records it and routes it.
Over-the-counter drugs included. No dose, no dosing schedule, no “a painkiller will help”.
It can say the result will be reviewed at the visit. Reading a report is the clinician's act.
The six limits, and the requirement of the resolution behind each one · What the copilot points out during the visit
The visit is transcribed and the note is born a draft. It waits in a visible queue for the clinician who saw the patient. Until someone releases it, it is not a document — and the whole site uses this convention: dashed is draft, solid is signed.
Patient reports low back pain for about two weeks, worse at the end of the day. Denies trauma. Reports partial relief with rest.
Mentions ongoing use of an antihypertensive.
Written by the assistant from the visit. Not a document yet.
Patient reports low back pain for about two weeks, worse at the end of the day. Denies trauma. Reports partial relief with rest.
Mentions ongoing use of an antihypertensive.
Reviewed and released by the treating clinician. From here on, it is part of the medical record.
While it is a draft, the text is not part of the medical record. See the visit's audit trail, hour by hour — it is what answers “who decided what”.
Whatever you don't turn on simply doesn't exist in your account. Three possible setups, to show that the whole machine is not mandatory — and that the most valuable piece changes from clinic to clinic.
Three possible setups. They are illustrative, built from what the product delivers — they are not customers. Swipe the table sideways to see all three columns.
| Piece | Solo practice |
Clinic with four specialties |
Clinic with telehealth |
|---|---|---|---|
| WhatsApp front desk | on | on | on |
| Triage by specialty | off | on | on |
| Schedule per clinician | on | on | on |
| Pre-visit intake form | off | on | off |
| Written visit note | on | off | on |
| Telehealth with a video room | off | off | on |
Nothing to install, no need to replace the clinic's system, no contract commitment.
Most messages arrive when there is no one at the front desk. It isn't a problem of effort: it's a problem of hours.
On the clinic's WhatsApp, at whatever time suits the person. The assistant answers right away, using the clinic's own material — what the clinic treats, where it is, which insurance plans it accepts.
Before any conversation about times, the message is read for signs of an emergency. If there are any, the guidance is to seek immediate care — and the assistant does not keep negotiating the schedule. It does not assess severity and never says it isn't serious.
Based on what the person described, the conversation goes to the right specialty among those the clinic offers. That avoids the question the front desk can't answer — “which doctor should I book with?” — and the visit wasted with the wrong clinician.
The assistant checks that clinician's real schedule and offers what is open. There is no “let me check and get back to you”: the slot offered is a slot that can be booked, and on confirmation the appointment is created.
Before the visit, the person receives a short form to explain the reason calmly. It is a pre-visit intake form, not a medical history: taking the history is a medical act and happens after the visit, with the clinician.
The clinician opens the schedule and finds the reason for contact and the patient's answers on the form. The visit starts on the topic, not on data collection.
Patients don't know — and shouldn't have to know — which specialty handles their case. Triage figures that out from what they said. And before all of that comes a fork that can't wait.
Swipe the drawing sideways to see all of it.
The emergency check happens before any conversation about times. It does not assess severity or rule out risk: when there is a sign, the assistant tells the person to seek immediate care and stops negotiating the schedule.
Most of what reaches a clinic's WhatsApp is the same question in a different order: do you take this insurance, where are you, do you have a slot on Thursday, do I need a referral. None of it requires judgment — it requires the right information, available on the spot.
The front desk is still the front desk: it takes care of the person in front of it, not the phone buzzing while it's helping someone.
If the visit is a telehealth appointment, the assistant creates the room when it confirms and sends the link for that specific visit. It does not adapt an old link or reuse one from another appointment — a plausible but invalid link is only discovered at the appointment time, with the patient waiting alone.
And a telehealth confirmation carries no address and no “please arrive early”: there is nowhere to arrive.
Three documents, three different requirements, and none of them becomes part of the medical record without passing through a person.
A transcription of what was said, with each line attributed to its speaker. It goes through no artificial intelligence at all — it is the raw text, and it exists so the other two have a basis.
Written from the visit, and only from what the person actually said. It is created after the visit because taking the history is a medical act — what is collected beforehand is the pre-visit intake form.
Drafted from what the clinician indicated during the visit. The assistant organizes the text; the treatment decision belongs to the clinician who saw the patient, and so does the release.
Putting everything in one text looks practical until the day someone needs to tell what the patient said from what the clinician decided. Kept apart from the start, that never becomes an argument.
In a single text, “the patient reports pain” and “rest was recommended” become the same prose. Kept apart from the start, the report stays a report and the decision stays a decision — and nobody has to reconstruct it from memory months later.
It also changes who may do what: a transcription error in the report is fixed by editing; a treatment decision can only be changed by the clinician. They are documents with different owners and different requirements.
Dashed is what the machine produced and does not count yet. Solid is what has passed through a person. The document changes its stroke at a single point — and it is always the same point.
Swipe the drawing sideways to see all of it.
The release queue stays in view of whoever has to release, showing how many documents are waiting and for how long. That is not dashboard decoration: a queue no one sees becomes a graveyard, and a late medical record is the worst place to let things pile up.
Each clinician sees their own queue. What they haven't released doesn't count as a record, and the system won't let anyone else release it for them.
The CFM resolution turned the medical record into a risk-management tool: it must make it possible to reconstruct later which tool was used, for what, and that there was human review. The audit trail is what does that — and it is what protects the clinician.
Swipe the drawing sideways to see all of it.
This reconstruction — which tool was involved, for what, and that there was human review — is what the resolution requires of the medical record. And it is what supports the “diligent use” on which the clinician's protection depends.
A medical record that is stored but can't be consulted is paper at the back of a drawer. You open a patient and see their history at this clinic — visit by visit, with who saw them and what was signed.
The date of the visit, the name and medical board registration of the clinician who saw the patient, and the documents released at that visit. Drafts are left out: until someone signs, it isn't a record — it's a queue.
A visit that happened with no author on record still shows up, and says so. Hiding it would leave a silent gap in the history, and a silent gap is worse than a declared one.
On each document: when it was released, and how many passages you changed from what the machine had written. “Released without changes” is also an answer — and it is different from not knowing.
CFM Resolution 2,454/2026 turned the medical record into a risk-management tool: it must make it possible to reconstruct later which tool was involved, for what, and that there was human review. The one who answers for that is the one who signed.
There is a difference between rubber-stamping what the artificial intelligence wrote and reviewing it — and that difference used to be recorded nowhere. Now it is: the text the machine produced is kept, and what you changed is counted.
Another clinician at the same clinic cannot release the document from your visit — the system refuses. And the front desk, which organizes the queue and can see that the document exists, whose it is and when it was created, cannot read the clinical content.
You see patients in more than one place with a single login, and the record does not cross the door: custody belongs to the facility where the visit took place. What you recorded there does not show up here — and that is how it should be.
None of this is dashboard decoration. It is the difference between saying there was human review and being able to show what it was.
Another clinician at the same clinic cannot release the document from a visit that wasn't theirs. It isn't a usage policy: the system refuses.
The pre-visit intake form and what was recorded at previous visits at this clinic come into the room with you. And while the conversation goes on, the copilot points out what was left half-done — without ever suggesting a diagnosis, a medication or a dose.
Dr. Helena · you
Antônio R.
RECORDING
Reason: low back pain for 2 weeks, worse at the end of the day
Intake form: denies trauma; has already tried rest
Current meds: ongoing antihypertensive
At this clinic: 2 previous visits
The intake form mentions ongoing use of an antihypertensive, and that hasn't come up in the conversation yet.
He mentioned tingling in his leg early on and the conversation moved on without coming back to it.
You talked about an imaging test, but nothing was settled about when or where.
Staged screen with fictional data. What it shows is real; the patient is made up.
Each one is a switch, and all three start off: the clinic turns on what it uses, and the cost stays proportional.
Swipe the drawing sideways to see all of it.
The raw text, filed in the archive. It is the basis for the other two.
no artificial intelligence at all
transcriptfetches the call's transcript, with the role of each speakerarchivesaves it to the archive with time, duration, who saw the patient and the recording referencepipelinemoves the patient to Seen — which is what answers “how many showed up this month?”What the patient reported, in the order of a clinical text.
affiliationfinds out which clinic the visit belongs to — the record does not cross the doortranscriptthe same basis, with no reprocessingscribewrites only what was said; anything not said becomes “not reported”, not an assumptiondraftsaved awaiting release, and added to the treating clinician's queueWhat the clinician indicated. Same discipline, different document — which is why there are two workflows.
affiliationsame clinic resolutiontranscriptsame basisscribenever corrects a dose: it writes what was heard and flags (verify) if it was ambiguousdraftalso waits for a signatureThis one doesn't fire when the call ends: it fires when a person signs.
archivestores it with the audit trail: who wrote it, who reviewed it, how many passages changeddeliversends the patient only what can be delivered — the care instructions go out; the medical history is an internal record, and the signed prescription is issued by the clinicThe costliest mistake in this work is attributing to the clinician something the patient said. That is why the transcript carries the role of each speaker, and the scribe separates the two before writing a single line.
Three, and none of them is a clinical decision. The difference is not one of tone, it is one of category: it has no way to issue a treatment suggestion.
What the intake form brought up and the conversation didn't touch, or the detail that is usually missing in that specialty: how long it has been going on, what has been tried, allergies, current medications.
“The intake form mentions ongoing use of an antihypertensive, and that hasn't come up in the conversation yet.”
Something the person said that the conversation let slip. It does not assess severity or say what it means — it says it was left behind.
“He mentioned tingling in his leg early on and the conversation moved on without coming back to it.”
A test, follow-up or referral that was mentioned and left hanging. It is what most often goes missing between one visit and the next.
“You talked about an imaging test, but nothing was settled about when or where.”
What changes is not the visit — it is what is left of it. On the left, the path almost every small clinic follows today.
Saving keystrokes is the argument everyone uses, and it is the smallest of all. What changes a clinic's week is somewhere else.
What gets written from memory, for six patients, is not what was said — it is what was left. And the record is what protects the clinician later.
A document stalled without anyone knowing is the worst kind of backlog. Here it shows up, along with whose it is.
Booked, intake completed, seen, no-show — without anyone writing anything down.
Nobody builds an audit trail after the problem. It counts because it was recorded when nothing was happening — and because you can open it the day you need it, patient by patient.
The medical history is the clinic's internal record: it exists for care and for the clinician's defense, not to circulate. What goes out on its own, once someone releases it, is the visit's care instructions — what was agreed, in the hands of the person who was seen. The signed prescription is still issued by the clinic: a prescription is only valid with the clinician's signature, and nothing here signs on their behalf.
An unnecessary alert in the middle of a visit costs two things: the clinician's attention and their trust in the tool. The second time it gets in the way, nobody reads it anymore.
Not even if the clinician asks directly. Clinical decisions belong to the doctor, and the CFM resolution does not allow them to be delegated.
Suggestions appear only to the person leading the visit. The patient sees none of it.
If the point was covered, it stays quiet. That is the difference between a copilot and an alarm.
When the visit is by video, the room is created at confirmation time and that visit's link goes in the message. No adapted old link — a plausible but invalid link is only discovered at the appointment time, with the patient waiting alone.
It is the same brain behind all four — same triage, same schedule, same clinic material. But if the clinic only looks at conversations, it will see less than half of what is running.
Swipe the drawing sideways to see all of it.
The only thing that changes between them is the place. Triage by specialty, the real schedule and the limits on what the assistant may say are exactly the same — because it is literally the same agent behind all four.
Swipe the drawing sideways to see all of it.
Where the conversation already happens. It is the busiest door, and the one that gets the most messages after hours.
same brainThe same conversation inside a web page. It works for people who'd rather not call, and for people already waiting at the counter.
same brainTelehealth, with the room created on the spot and the patient's context coming in with you.
with copilotWhere clinicians see the schedule, the patients, the queue of documents awaiting release — and each patient's record at this location.
team onlyA patient can start on WhatsApp, be seen in the video room and come back through the website: it is the same patient file, and they don't repeat their story.
This is the part nobody sees and nobody buys in a demo — and it is where most of the time the clinic gets back comes from. None of these lines was triggered by someone clicking on something.
A few hours before the visit, timed from the appointment. Whoever fills it in arrives with the reason already explained.
triggered by the booking, not by a personNo artificial intelligence at all in the raw record: it is the text of what was said, with the role of each speaker.
triggered by the end of the callWritten from what was said — the patient's report on one side, the clinician's decision on the other, never mixed.
two separate workflows, each its own switchShowing how many documents are waiting and for how long. A queue no one sees becomes a graveyard, and a late medical record is the worst place to let things pile up.
each clinician sees their ownWho wrote it, who reviewed it, how many passages changed and who released it. And only what can be delivered reaches the patient.
triggered by the signatureContact, visit booked, intake completed, seen, no-show. Nobody needs to drag a single card for the clinic to know how many booked and how many showed up.
and that is what answers “how many no-shows this month?”A clinic that only turned on the WhatsApp front desk would already save the receptionists' hours. But a clinic that stops there is still writing records from memory at the end of the day, still finding out next month how many patients didn't show, and still has documents stalled without anyone knowing.
The question that separates the two is not “how many messages does it answer?”. It is what happens after the conversation ends.
See the document's cycle, from the end of the visit to the archive
Each piece is a switch, and whatever you don't turn on doesn't exist in your account.
Neither “a clinic has doctors” nor “a doctor has clinics” describes this without duplicating someone. They are two separate things and an affiliation between them — and it is the affiliation that carries the schedule.
Custody of the medical record belongs to the facility, not the clinician: the clinic is answerable for it, and it is the clinic's number the patient knows. An independent practitioner is no exception — they are a clinic of one, with the same structure and without the steps that only make sense with a team.
What changes from place to place is the affiliation: the same person has different hours at each address where they practice, and what answers “when does she see patients here?” is the person-clinic pair, never either one alone.
Either design forces you to duplicate someone: the same person becomes three profiles, or the same clinic shows up in three places. What solves it is recognizing that the schedule belongs to neither — it belongs to where the two meet.
Swipe the drawing sideways to see all of it.
She is the one who uses the system all day, and she is the one who abandons the tool if it treats her like a visitor. That is why she has what she needs to do a good job at the counter: the schedule, the patients, what is booked and what changed.
What isn't hers is the clinical content and the release of documents — not because of hierarchy, but because the one who answers for the act is the one who saw the patient.
Since February 27, 2026, a resolution of the CFM (Brazil's Federal Council of Medicine) spells out what artificial intelligence may not do in medicine. It is not a footnote detail: it is the design of this product, item by item — and that is why this is the most concrete page on the site.
Delegating to artificial intelligence the communication of a diagnosis, prognosis or treatment decision is prohibited.
The assistant doesn't give a diagnosis, not even as a hypothesis, doesn't recommend medication — over-the-counter included — and doesn't interpret test results. The restriction lies in what it is able to do, not in a request written in its prompt.
The final word on diagnostic, therapeutic and prognostic decisions always belongs to the physician, and using the tool does not relieve them of responsibility.
Every document is born a draft and waits for release. And only the clinician who led the visit can release its documents: another clinician at the same clinic is refused by the system.
The medical record must record the tool used, the purpose of its use and the human review of its suggestion — so that who decided what can be reconstructed later.
The audit trail records each step with a timestamp: what the assistant wrote, that it was left waiting, that it was reviewed, how many passages changed and who released it. That trail is what supports the “diligent use” on which the clinician's protection depends.
Patients have the right to be informed, clearly and accessibly, whenever artificial intelligence is used.
The conversation never pretends to be a person, and the record that the interaction went through the assistant stays in the visit's audit trail.
Under the LGPD — Brazil's General Data Protection Law — the assistant asks only for what it needs to guide and book. Never a CPF (Brazilian taxpayer ID), ID card, credit card, password, full insurance member number or a photo of a document. It isn't conservatism: it is what the law classifies as sensitive, and whatever is left stored is what can leak.
The data lives on our own infrastructure, in Brazil, with each clinic's content isolated from the others — and the material the clinic uploads is consulted at answer time, without feeding any model training.
A refusal is never an “I can't help you”. The assistant records the account, routes it to someone who can answer and steers toward the visit — which is where the question will really be answered.
A slot it didn't check, an insurance plan that isn't on the list, the name of a clinician who doesn't practice there. If it doesn't have the answer, it says it will confirm.
That is up to the insurer. It says what the clinic accepts and offers the alternative.
No error codes, no system names, no failure notices. If something doesn't work, it takes another route and logs it for the team.
Guessing which system the clinic uses signals an assumption, and guessing wrong signals carelessness.
If the line between what the machine does and what the person signs is in the right place, the rest follows.
It isn't booked as an expense, it doesn't appear in the report and it isn't discussed in meetings. It is the gap between the schedule you booked and the schedule that actually happened — and it disappears from revenue without a trace.
The percentages are industry figures and are on the home page, with sources. The rest is arithmetic.
In the example below, 300 — a clinic with four or five clinicians.
In the example, R$ 200. This is the only number that doesn't come from research — it is illustrative, and it is the first one you replace.
A clinic with no active confirmation process sits between 20% and 30%. In the example, 20% of 300 appointments at R$ 200 comes to R$ 12,000 a month that the schedule promised and the till never saw.
And that is only the half you can see. The other half is the message that arrived at 10 p.m. and got answered at 8 the next morning — when the person had already booked somewhere else. That one doesn't become a no-show: it never becomes an appointment, so it never shows up in any statistic.
The most common price comparison is with practice management software, and it leads to the wrong conclusion — because management software waits for someone to type. This assistant is what happens when there is no one.
Stores the schedule, records and finances. It does all of that very well, and answers no one: someone on your team has to sit in front of it. You still need it — and you don't have to replace it.
Answers, triages, books, confirms and reminds. That is exactly what the assistant does — only also at 10 p.m., on Sundays and on holidays, which is when a good share of the messages arrive.
Transcribing the visit and drafting the note is the expensive part, because it costs something on every visit. That is why it is a separate switch: if all you want is a full schedule, you don't pay for it.
Not per registered clinician. A doctor who saw no patients that month doesn't count, and a clinic that saw more pays proportionally more — not least because it gave the system more work.
The unit is the recorded visit, chosen for a practical reason: it is the one thing here you already count every day, and it is the same unit you already use to think about your own revenue. No messages, audio minutes or credits — three units nobody knows how to translate into money.
The higher the volume, the lower the price per visit. The figure comes out of a conversation, once we know what you'll turn on — because a number in isolation, without knowing which pieces you want, misleads in both directions.
The first step is just the front door: answer, triage, book, confirm and remind. No visit notes, no transcription, no medical record. It is the step that goes after the empty chair, which is the cost behind the numbers above — and it is the cheapest precisely because it leaves out the expensive piece.
The note comes in when you want it, and not before. Turning something off uninstalls nothing: whatever you don't turn on doesn't exist in your account, and whatever has already been written remains yours, in a format you can take with you.
If the clinic prefers its own server and brand, that exists and costs more — and it has a consequence we'd rather tell you before selling. It's explained here.
By usage, not by seat — you don't pay for a registered doctor who saw no patients. The figure comes out of a conversation, once we know what you'll turn on.
The account is created in minutes. What determines the rest is how much of the clinic's material is already organized: insurance plans, hours, who sees what.
Turn on one piece, see the effect, and only then turn on the next — the minimum-dose logic applies here too. Start with the front desk alone, with no notes at all. And turning something off uninstalls nothing: whatever you don't turn on doesn't exist in your account, and whatever has already been written remains yours, in a format you can take with you.
It doesn't need a system change, staff training or convincing anyone to type more. It needs someone answering when nobody is there.
The product is the same. What changes is whose address it is — and the choice has a consequence we'd rather tell you before selling, not after.
The product is the same. What changes is whose address it is — and the choice has a consequence we'd rather tell you up front.
There is no expensive wrong choice: starting on the shared platform and migrating later is a planned path, and the visit records come along.
It is the fastest way to see the assistant answering your number, and migrating later is planned for — the visit records come along.
The twelve that come up most, answered straight — including the ones that don't work in our favor.
No, and that isn't a setting. It doesn't diagnose, not even as a hypothesis, doesn't recommend medication — over-the-counter drugs included — and doesn't interpret test results. CFM Resolution 2,454/2026 (from Brazil's Federal Council of Medicine) prohibits delegating those communications to artificial intelligence, and the product was built inside that line.
The clinician who saw the patient. The note is born a draft and sits in a queue until a person reviews and releases it. On top of that, only the clinician who led that visit can release its documents — another clinician at the same clinic can't.
Yes. You open the patient and see their history at this clinic: visit by visit, with who saw them and the documents released — and, on each one, when it went out and how many passages you changed from what the machine had written. That reconstruction is what the CFM resolution requires of the medical record, and it only counts if someone can open it.
Through the audit trail. The text the assistant wrote is saved when you make the first correction, and the number of passages changed is recorded at release. “Released without changes” is also an answer — what doesn't exist is releasing without a record of what was released.
The emergency is recognized before anything else and jumps the booking queue: the assistant tells the person to seek immediate care and does not keep negotiating times. It does not assess severity and never says it isn't serious.
Yes. The CFM resolution gives patients the right to be informed, clearly, whenever artificial intelligence is involved in their care — and the conversation never pretends to be a person.
On the shared platform, no: the person has a single login and sees, at each clinic, only what belongs to that clinic. The medical record belongs to the facility, so it doesn't cross the door — seeing patients at clinic A gives no access to records made at clinic B. A dedicated install is different, by definition: a separate install is an island, so anyone who also practices at a clinic on the shared platform will have two logins. That is the price of isolation, and it is exactly what you are buying there.
Yes. In a dedicated install the domain and the identity belong to the clinic, the server and database have no neighbors, and the clinic decides when to update. It costs more, because it is per-customer infrastructure and operations. The consequence we'd rather mention up front: a separate install is an island, so a clinician who also practices at a clinic on the shared platform will have two logins. Starting on the shared platform and migrating later is a planned path, and the visit records come along.
On our own infrastructure, in Brazil, with each clinic's content isolated from the others. Health data is sensitive data under the LGPD (Brazil's General Data Protection Law), and the assistant asks for the minimum needed to guide and book — never documents, cards or passwords. To be precise about what leaves: the text of each conversation is processed by an artificial intelligence provider to generate the reply, and that provider states it does not use the content to train models or retain it afterwards. The archive — medical records, schedule, history — never leaves here.
Not to get started. The front desk, triage and schedule work on their own, and the visit note is available for you to take wherever you want. Integration with third-party systems is a separate conversation, and depends on which one.
Yes. An independent practitioner is treated as a clinic of one — same structure, without the steps that only make sense with a team. Triage by specialty, for example, you simply don't turn on.
The receptionist is a role, not a lesser level of access: she handles the schedule, the patients and whatever she needs to do a good job at the counter. Clinical content and the release of documents stay with the clinician who saw the patient.
Create the account and try it with your own number — it's faster than us answering in writing.
You create the account, connect the schedule and invite the team. The clinic code comes out at the end — it is what links WhatsApp to your account.
By creating the account you agree that the conversation with the assistant is recorded in the clinic's account.
Your clinic code
—
Keep this code: it is what identifies your location on the shared number.
It is what lets the assistant offer slots that exist, instead of promising to get back to people.
Each person signs in with their own login and their own role, at however many clinics they practice.
