Superintelligence for Physiotherapy Clinics: Scheduling and Follow-Ups

Superintelligence for Physiotherapy Clinics: Scheduling and Follow-Ups

How a superintelligence agent handles scheduling and follow-up reminders in physiotherapy clinics, and what always stays with the physiotherapist.

•

Want to apply this in your company?

Book a call with Juan →

An AI agent in a physiotherapy clinic handles the operational work: booking, rescheduling, confirming appointments, keeping a waiting list and sending follow-up reminders according to a rule the team has defined. It does not assess pain, suggest exercises, decide whether a patient should return or change the session frequency prescribed by the physiotherapist. The line between administrative and clinical work is the most important part of the project, and it should be written down before any testing begins.

Key takeaways

  • The agent manages the schedule: booking, rescheduling, confirmation and waiting lists, always within the time slots the clinic makes available.
  • Follow-up reminders follow an administrative rule written by the team, never a clinical recommendation created by the agent.
  • Assessment, treatment, exercises, symptoms and progress stay with the physiotherapist, and the agent hands them off without trying to answer.
  • Every conversation needs a record, a handoff to a person, and a list of situations that trigger an immediate handoff.
  • Before expanding, the clinic tests a real workflow against an agreed measure and decides whether to expand, adjust or stop.

What the agent handles in the schedule

In a physiotherapy clinic, a large share of front desk time goes to booking calls, time changes and confirmations. These tasks follow known rules: which time slots exist, which professionals work at each location, how much notice is needed to cancel, and what happens when a patient misses an appointment. Because these rules can be written down, they can be executed in a predictable way.

In practice, the administrative scope of the agent usually includes:

  • Booking the first appointment and later sessions from time slots the clinic has released.
  • Rescheduling and canceling within the notice policy the coordination has set.
  • Confirming attendance and keeping a waiting list for openings that appear.
  • Collecting only the registration data needed: name, phone number, location and health insurance plan, when applicable.
  • Providing the address, documents to bring and administrative rules, such as the no-show policy.

All of this depends on organized data. If the schedule lives in a spreadsheet separate from the clinic's system, the agent will reproduce the mess faster. Before connecting any tool, check that time slots and professionals are registered in a single, consistent way.

Follow-up reminders: an administrative rule, not a diagnosis

The follow-up reminder is the point where the agent comes closest to clinical work, so it needs a simple rule. The reassessment date should come from a field completed by the physiotherapist in the medical record or patient file. The agent reads that date and sends a message within the agreed timeframe. It does not calculate the date on its own, does not infer it from something the patient said, and does not change the interval because a patient reports feeling better.

The message text should also be fixed and approved by the coordination and the technical lead. Messages the clinic starts outside WhatsApp's 24 hour customer service window require an approved template, and the patient's permission to receive this kind of contact must be recorded. These requirements are covered in more detail in outbound WhatsApp messages.

Illustrative scenario: a patient has finished a block of sessions, and the physiotherapist has recorded that a reassessment should happen by the date noted in the file. The agent sends the message on that date, asks whether she wants to book the reassessment, and offers available times. If she replies that she has some discomfort or wants to talk about her treatment, the conversation goes to the front desk or to the responsible physiotherapist, and the agent does not comment on the symptom.

What stays outside any clinical guidance

The list of what the agent does not do needs to be as concrete as the list of what it does. General statements like "no medical advice" do not guide anyone in practice. The clinic should explicitly exclude:

  • Interpreting pain, symptoms, test results, reports or images.
  • Suggesting exercises, loads, posture changes, stretches or home remedies.
  • Saying whether a patient can train, return to sport, drive or work after a session or procedure.
  • Changing the frequency, duration or type of session that was prescribed.
  • Giving advice about medication or suggesting combinations of treatments.
  • Classifying a clinical emergency on its own.

When a question falls into one of these areas, the agent responds with a fixed text, for example saying that the question needs to be answered by the physiotherapist and that the team will be notified. If a message contains signs that require immediate attention, such as sudden loss of strength or shortness of breath, the text should direct the patient to emergency care right away and alert the team. This response must be validated by a health professional at the clinic before going live, because the agent should not improvise at this point.

The term superintelligence appears often in market content, but what matters to a clinic is what the system does today, with visible limits. The difference is covered in superintelligence in practice: what an agent does today.

How to build the workflow in six steps

A scheduling and follow-up workflow can be designed in an afternoon with the front desk team and a responsible physiotherapist. The order below avoids the most common mistake, which is starting with the tool and only then thinking about the rules.

  • Step 1: list the administrative tasks and explicitly mark what stays out, using the exclusion list above.
  • Step 2: define the source of the follow-up date, which must be a field completed by the professional and never an inference by the agent.
  • Step 3: write the fixed messages and have each one approved by the coordination and the technical lead.
  • Step 4: define the words, topics and situations that trigger an immediate handoff to a person.
  • Step 5: record every conversation, every handoff, the reason and the time, so the team can review them later.
  • Step 6: test with scripts based on real cases and edge cases, including ambiguous messages, before releasing the flow to patients. A low-risk testing approach is described in mock tools for testing AI agents safely.

Health data, LGPD and the communication channel

Brazil's LGPD treats health data as sensitive personal data, which calls for more care than an ordinary customer record. For a scheduling agent, the practical principle is data minimization: it needs the name, phone number, location, appointment time and follow-up date, and it does not need the full medical record. The less clinical content reaches the system, the smaller the exposure.

Conversation logs are also personal data, so the retention period, who can access them and for what purpose must be defined. If the tool processes data outside Brazil, the international transfer question enters the analysis, as explained in international data transfer under LGPD when using AI providers. Brazil's National Data Protection Authority (ANPD) publishes guidance and is the reference for specific questions. It is also worth consulting the clinic's lawyer before publishing the workflow.

How to start with a pilot

The method we use for projects like this has four stages: map, limit, pilot and hand over. First, the clinic records how scheduling works today, with a baseline of no-shows, reschedules and follow-ups that go unbooked, using data that already exists. Then it limits the scope to one location or one professional, and to a single workflow, usually the follow-up reminder, because its rule is clear.

During the pilot, every conversation is logged and reviewed at short intervals, and the team agrees on a success measure beforehand, such as the share of follow-ups booked within the agreed timeframe, without promising a specific number. At the end, the decision is objective: expand, adjust the rules or stop. Clinics that already use scheduling systems and want a comparable case in healthcare can look at superintelligence for podiatry clinics, with schedule, records and follow-ups.

Frequently asked questions

Can the agent answer questions about symptoms before an appointment? No. Questions about pain, injury or progress are clinical and should be passed to the physiotherapist or the front desk, following the clinic's rules. The agent can share the next available time and let the patient know the team will get back to them, but it should not try to interpret what the patient describes.

Who sets the follow-up date? The responsible physiotherapist sets the date in the patient record, and the agent only reads that field. If the field is empty, the system should not invent a timeframe. The case should go to a review list for the team.

Do patients need to give permission to receive reminders? Yes. The clinic should have that permission recorded for this kind of contact, and messages started outside the 24 hour window require a template approved by the platform. The topic of consent is covered in more detail in WhatsApp opt-in and how to collect permission for outbound messages. Confirm the legal position with the person responsible for the clinic.

How do you know if the agent is going off track? The team reviews a regular sample of conversations, tracks handoffs to people, and keeps records with the reason and time for each one. If clinical answers appear that should never be given, the workflow is paused, the rule is adjusted and the test is repeated before it goes back into operation.

Sources