Skip to content

Conversational AI for clinics and healthcare: appointments and LGPD

What conversational AI can do on WhatsApp for clinics and health services, such as confirming appointments and administrative triage, and what Brazil's LGPD requires.

Marlos Carmo

Founder of Tolky

Updated on November 16, 2026

10 min read
Conversational AI for clinics and healthcare: appointments and LGPD

Share

Clinics and health services live on WhatsApp. That is where patients confirm a time, ask whether their plan covers an exam, say they will be late and, sometimes, describe a symptom that nobody at the front desk should interpret.

That mix is the challenge. The front desk must be fast on administrative messages, which are the majority, and careful with the few that touch health. At the same time, any new tool touches data that the law treats more strictly.

This guide shows what conversational AI can do in this context, what it should not do and what Brazil's data protection law (LGPD) asks of those who handle this data. It is neither legal nor clinical advice: it is an informational outline for starting the conversation with your legal counsel and your clinical team.

Updated in November 2026.

What is conversational AI for clinics and healthcare?

Conversational AI for clinics and healthcare is an artificial intelligence agent that serves patients on WhatsApp in administrative tasks, such as confirming appointments, answering routine questions and organizing first contact, and hands off to the team when a person is needed. It works with the information the clinic registers, not with medical knowledge of its own.

Unlike a menu chatbot, it understands messages written the way the patient writes. But the limit is the same as any administrative service: it does not interpret symptoms, suggest treatment or replace a health professional. For the difference between the two technologies, see Conversational AI is not a chatbot.

What can AI do for the clinic?

AI can handle repetitive administrative tasks: confirm appointments, answer routine questions, organize administrative triage and log every contact. Four uses cover most of the volume.

Appointment confirmation

For confirmation, the AI sends the reminder on WhatsApp and talks with the patient about the reply: confirm, ask to reschedule or say they will not come. Outbound sending uses text campaigns, with a Meta-approved template and for people who gave opt-in. If the patient replies, the conversation continues with the AI, and the front desk receives the outcome.

Routine questions

The AI answers from what the clinic registered in the knowledge base: address, hours, documents to bring, administrative preparation, accepted insurance plans and payment methods. When the question falls outside the base, the AI transfers instead of making something up. To build the base, see how to train the agent with a knowledge base.

Administrative triage

Administrative triage is organizing first contact: who the patient is, what they are looking for, which unit or specialty and how urgently the front desk should respond, without assessing the health condition. The AI collects the name, the reason for contact in administrative terms (first visit, follow-up, exam, copy of a document) and routes to the right queue with a summary.

CRM record

Every conversation stays in the contact's history, with stage and activity. The front desk helps without asking for everything again. See the CRM solution.

What does the AI not do?

The AI does not diagnose, interpret symptoms or exams, recommend medication or give clinical guidance. That line must be explicit in the configuration and in the conversation.

In practice, three rules help:

  • Define what is clinical. Any message about a symptom, pain, medication, exam result or emergency leaves the AI flow and goes to a person.
  • Write the safety reply. For messages that suggest urgency, the AI replies with the clinic's standard guidance (for example, seek urgent care) and transfers. The clinic's responsible professional defines the text.
  • Introduce itself as AI. The patient must know they are talking to an AI assistant and can ask for a person at any time.

The AI is also not the clinic's calendar. Tolky does not currently have a confirmed native scheduling field, so we do not promise that the AI books appointments on its own. What it does is lead the conversation and route the booking or rescheduling request to the team, recording everything. More in the next section.

How do booking and rescheduling work?

The AI collects the patient's request and routes it to the front desk, which books or reschedules in the system the clinic already uses. That way the calendar stays with whoever controls it, and the patient does not repeat information.

Patient requestWhat the AI doesWhat the team does
Confirm appointmentSends reminder, records the replyUpdates the calendar if needed
RescheduleCollects preferred day and shift, recordsReschedules and confirms
First visitCollects administrative data, logs in CRMBooks and replies
Insurance questionAnswers from the registered baseCases outside the rules
Symptom or urgencySafety reply and immediate transferAttends the patient

If your clinic needs direct booking into the calendar, talk to the Tolky team about available integrations before counting on it. Details on the transition between AI and person are in handoff rules.

What does the LGPD require when health data is involved?

The General Data Protection Law (Law 13,709/2018) requires a clear purpose, the minimum necessary and security in handling personal data, and gives sensitive data its own rules, a category that includes health data. The text is at Planalto, consulted on November 16, 2026 (Portuguese only). For the term, see LGPD in customer service.

The points in the law's text that weigh most in a clinic conversation:

  • Principles (art. 6). Handling must have legitimate, specific purposes informed to the data subject (purpose), be compatible with what was informed (adequacy), limited to the minimum necessary (necessity), with clear information to the data subject (transparency) and with technical and administrative security measures.
  • Legal bases (art. 7). Among the grounds is the data subject's consent, and others are provided by law.
  • Sensitive data (art. 5, II, and art. 11). Sensitive data has narrower grounds. One is the data subject's consent in a specific and prominent way, for specific purposes. Paragraph 4 of art. 11 expressly addresses sensitive data related to health.
  • Automated decisions (art. 20). The data subject may request review of decisions made solely on automated processing.
  • Data protection officer (art. 41). The controller must appoint a person in charge of personal data processing.
  • Incidents (art. 48). The controller must notify the national authority and the data subject of a security incident.

Which legal basis applies to each flow in your clinic is a question for your legal team or your data protection officer. For official guidance, also see the ANPD page.

How do you apply this in the conversation? A checklist

Applying the LGPD in the conversation means asking only what is necessary, telling the patient they are talking to AI, limiting who has access and having a human path for anything sensitive. The checklist below turns the law's principles into configuration decisions.

ItemControl questionPractical decision
PurposeWhy is this data requested?Ask only what the administrative flow requires
NecessityCan we solve it without this data?Do not ask for symptoms, diagnosis or exams in the chat
TransparencyDoes the patient know it is AI?Intro message as an AI assistant
AccessWho sees the conversations?Access profiles by team and activity log
Sensitive dataWhat if the patient describes a symptom?Safety reply and transfer to a person
Data subject rightsHow does the patient request access or deletion?Channel indicated by the clinic and internal flow with the officer
IncidentWho notifies and how?Clinic's plan, with the officer

Which mistakes should you avoid at the start?

The most common mistakes are asking for too much information in the chat and leaving the AI without a ready reply for the sensitive case. Both are solved before you turn the AI on.

  • Asking for symptoms "to speed things up". It seems useful but widens the handling of sensitive data without need. Ask for the type of visit, not the clinical picture.
  • Forgetting the urgent case. Define with the responsible professional the text the AI sends and the immediate transfer path.
  • Outbound messages without opt-in. Reminders only for people who authorized them, with an approved template and an option to opt out.
  • Open access for everyone. Not everyone on the team needs to see every conversation. Use access profiles.
  • Not reading the conversations. Reserve a weekly slot to review a sample and adjust the base and rules.

Where does Tolky fit, and what does it not promise?

Tolky is a conversational CRM with AI agents on WhatsApp (official API), webchat and voice, with handoff to the team along with history. In healthcare, the fit is the administrative layer: confirmations by text campaign, routine answers, administrative triage and CRM records.

On the security side, the platform has access profiles by team and an activity log with export. It does not claim security certifications or data residency in any specific country, does not offer anonymization or configurable retention today, and does not diagnose. To learn how to evaluate an AI vendor, see security and privacy in AI platforms and AI governance in customer service. The clinic operates the data, and compliance is a shared responsibility, defined in contract.

Frequently asked questions

Does the AI give a diagnosis or clinical guidance?

No. The AI handles the administrative side. Any message about a symptom, medication, exam or urgency should leave the flow and go to a person, with the safety reply defined by the clinic.

Does the AI book appointments in the clinic's calendar?

Tolky does not have a confirmed native scheduling field. The AI leads the conversation, collects the patient's preference and routes the request to the front desk, recording everything in the history.

Can I confirm appointments on WhatsApp?

Yes, with text campaigns: a Meta-approved template, sent to people who gave opt-in, with a conversation with the AI when the patient replies.

Can health data be handled in a conversation with AI?

The LGPD classifies health data as sensitive, with narrower grounds for handling. The decision on legal basis and flow belongs to the clinic, with its legal team and officer. The safest design is to ask for the minimum and not collect clinical information in the chat.

Does Tolky have a security certification or store data in Brazil?

We do not claim certifications or data residency. For security and data handling details in your case, talk to the team.

Can the patient talk to a person?

Yes. The AI introduces itself as an AI assistant and transfers the conversation with context when the patient asks or the case requires.


Conclusion

In clinics, conversational AI pays off for what is administrative and repetitive: confirming, answering the basics, organizing first contact and recording. What is clinical stays with professionals, and sensitive data requires a clear purpose, controlled access and human review, as the LGPD asks.

Let's talk? See the WhatsApp solution, see the plans or talk to the team →

Share

Tags

  • conversational AI for clinics
  • chatbot for clinics and healthcare
  • WhatsApp for clinics
  • appointment confirmation on WhatsApp
  • LGPD in healthcare
  • administrative triage
  • #conversational CRM
  • #AI agent

About the author

Marlos Carmo

Founder of Tolky

Marlos Carmo is an AI entrepreneur and founder of Tolky, the conversational-era infrastructure and AI CRM that unifies intelligent service, multi-channel support (such as WhatsApp and voice), live CRM, and operational intelligence in a single ecosystem. He is a finalist for the SXSW Innovation Awards and a member of Francesco's Economy, a global network of young entrepreneurs focused on innovation and social impact. He works connecting Artificial Intelligence and digital transformation in projects for large organizations.

Next step

Want to bring this to your operation?

Tati, Tolky's AI assistant, understands your scenario in a few questions, suggests the best path and connects you with someone from the team, with full context.

Read also