Skip to content
be-human.ai
← Back to the journal

AI by industry

AI Scheduling for Doulas: Client Intake and Due-Date Records

A practical guide for solo doulas on using an AI agent for intake forms and visit scheduling, keeping care decisions personal and records clean.

Be Human typographic cover with the title: AI Scheduling for Doulas: Client Intake and Due-Date Records
Conceptual illustration created with AI · Be Human

An AI agent can handle the repetitive logistics of a solo doula or small doula practice: sending intake forms, offering consultation times, booking prenatal and postpartum visits and sending reminders. It should write each step into a dated, structured record and pass anything about birth preferences, clinical concerns or care decisions to the doula personally. Set up this way, the practice gets a calmer calendar now and a reliable record base that future tools can use later.

Key takeaways

  • Let the agent own the logistics (forms, time slots, reminders and reschedules) with clear limits on what it may confirm.
  • Keep birth preferences, clinical questions and medical advice with the doula, and write the handoff as a rule.
  • Store due dates as dated facts with a stated source, and keep earlier values when they change.
  • Give the agent only the permissions each task needs, and treat intake text as untrusted input.
  • Test the agent against real past scheduling cases before widening its role.

What an agent can handle in a doula practice

Doula work is mostly coordination around a few moments that matter a great deal. Before a client's first visit there is a steady stream of messages: an inquiry, a consultation request, an intake form, a question about availability, a contract, and then a series of prenatal visits that move as pregnancies change. Most of that is administrative and repeatable, which makes it a good first role for an agent.

Tasks that suit an agent well include:

  • Answering first inquiries using the practice's published service area and availability rules
  • Sending the intake form and checking that required fields are complete
  • Offering consultation and visit slots from the doula's booking calendar, with buffers the doula sets
  • Sending reminders for upcoming visits and recording whether the client confirmed
  • Scheduling postpartum check-ins once a birth has been logged by the doula
  • Rescheduling when the client or doula asks, and recording the change with its reason

Adjacent care practices face the same intake and scheduling limits. The behavioral health intake guide explains where automated scheduling should stop, and the same boundary applies to a doula's inbox.

What stays with the doula

The line that matters is between logistics and care. An agent can tell a client that a visit is on Tuesday at ten. It should not interpret a symptom, recommend a provider, suggest a birth decision or reassure a client about a medical concern. Those belong to the doula and, where relevant, to the client's own clinicians.

Build the handoff as a fixed rule, not an intention. Define a short list of trigger topics and a standard reply that tells the client the message is being passed to the doula directly, with the usual response window. Typical triggers include requests for medical advice, pain or bleeding reports, emotional distress, complaints, questions about a change in provider or birth location, and any request to change a care plan. The agent stops scheduling on that thread and notifies the doula.

Any message that may describe an emergency needs a different response. The agent should immediately tell the client to contact their medical provider or emergency services, then notify the doula. It should never answer such a message with a booking reply.

Illustrative scenario: A client sends a message late in the evening asking whether a symptom they describe is normal and whether they should wait until morning. The agent does not answer the question. It sends the emergency instruction and the standard handoff reply, marks the thread as urgent for the doula, and leaves the calendar untouched.

A step-by-step plan for the first setup

Start narrow, then widen the role after reviewing real conversations. A workable order looks like this:

  • Step one: write the practice's scheduling rules in plain language, including consultation length, service area, cancellation terms and what the doula will never discuss by message.
  • Step two: list every field the intake form needs, and remove any question you would not actually use. Less collected data means less to protect.
  • Step three: define the trigger topics, the emergency instruction and the exact handoff reply.
  • Step four: connect the booking calendar with access limited to that calendar, not to personal or family calendars.
  • Step five: run the agent against past scheduling requests whose outcomes you already know, and check each proposed reply against your rules.
  • Step six: go live on inquiries and reminders only, and review a sample of conversations every week.
  • Step seven: add rescheduling and postpartum check-ins once the reviews show replies that follow your rules.

The most common early mistakes are letting the agent see the full intake form when it only needs scheduling fields, letting it guess at availability instead of reading the calendar, and failing to record which version of the form a client signed.

Designing due-date and visit records

The due date anchors most of a doula's scheduling, and it is also the field most likely to change. Record it as a dated fact with a source. Rather than overwriting the old value, add a new entry showing the date it was reported and who reported it. When a client's provider updates an estimate, the history then shows what was known at each point.

A useful visit record separates what happened from what was said about it:

  • Scheduling facts: date, time, format, attendees, and whether the visit was moved
  • Client-reported information: what the client said, recorded with the date
  • Doula notes: observations and decisions, written and kept by the doula
  • Messages and reminders: what was sent, what was delivered if the channel reports it, and what was answered

Illustrative scenario: A doula takes on a client whose estimated due date comes from the provider. The intake form stores that date with its source. Several weeks later the client mentions an updated estimate, and the doula confirms it. The agent moves the postpartum check-in slot tied to the new date, keeps the earlier date in the history with its original source, and posts a note asking the doula to review the prenatal visit schedule. Nothing is erased, and the doula decides what the change means for care.

The visit recall approach in chiropractic intake follows a similar pattern of treating each visit as its own dated event, which is worth borrowing.

Permissions, privacy and consent

Pregnancy and family information is sensitive, and clients often share more than a form asks for. The legal duties depend on where the practice operates and how it stores data, so get local advice instead of relying on a general rule. In Brazil, the LGPD sets the framework for processing personal data, including health data, and the [ANPD](https://www.gov.br/anpd/pt-br) publishes guidance on how it is applied. In the United States, state rules differ. Whatever the jurisdiction, the practical controls are similar.

  • Give the agent access only to the fields each task requires
  • Keep booking calendars separate from personal and family calendars
  • Use individual logins, and revoke access when a helper or contractor stops working with you
  • Set a retention period for intake forms and state it in the consent text
  • Tell clients which channel is used and that an automated assistant handles logistics

Intake text is also untrusted input. A client may paste instructions into a message or form field, and an agent that treats that text as commands can be steered into revealing records or changing bookings. Keep permissions narrow so that one bad message cannot do much, and check authorization on every action rather than only at login. The [OWASP guidance on authorization](https://cheatsheetseries.owasp.org/cheatsheets/Authorization_Cheat_Sheet.html) and its [prompt injection prevention notes](https://cheatsheetseries.owasp.org/cheatsheets/LLM_Prompt_Injection_Prevention_Cheat_Sheet.html) are good starting references for the technical team.

How Be Human's managed approach handles this

Be Human starts by defining the agent as one role: the scheduling tasks it performs, the records it may read and write, the tools it may use, the approval boundaries for changes and handoffs, and one observable success criterion, such as the share of booking requests closed without a manual correction. Managed operations then cover monitoring of live conversations, recurring evaluations against real cases, permission reviews whenever staff or connected tools change, maintenance when a calendar or form provider updates its interface, and controlled changes to the rules, each reviewed before it goes live.

Building a record base for future tools

A clean record is useful even if the practice never adopts another tool. Consistent fields, dated entries and visible sources make the history easier to search, export and review. They also make the agent measurable: you can check whether a reply followed policy, whether a reminder was sent, and whether a handoff happened when it should. The [NIST AI Risk Management Framework](https://www.nist.gov/itl/ai-risk-management-framework) is a useful reference for thinking about monitoring and accountability in this kind of system.

Keep the record in plain fields rather than long free-text notes that a tool must reinterpret later. Name each field once, use one date format, and store the version of the intake form each client signed. Memory for the agent should hold scheduling facts and stated preferences, not clinical conclusions. Future tools will only be as reliable as the record they read, so the work done now on fields and sources carries forward. The acupuncture clinic guide on packages and follow-ups shows how a similar record structure supports repeat visits over time. If you want to find the first scheduling task in your practice that is worth handing to an agent, take the free Be Human AI diagnostic.

Frequently asked questions

Can an AI scheduling agent answer clients' questions about birth options? No, and it should not try. It can describe the practice's general services and availability, then route questions about birth preferences, medical concerns or provider choices to the doula. A clear handoff message tells the client who will reply and when.

Do I need to tell clients that an agent is involved? Yes. Say so in the consent text and in the first message, and state which channel they are using and how to reach you directly. Disclosure also makes it easier to correct a mistake quickly, because the client knows the reply came from an automated assistant.

What is the best first task to hand to an agent? Usually new inquiries and appointment reminders, because the rules are clear and mistakes are easy to spot. Avoid starting with rescheduling close to a due date or with any message that mentions symptoms. Expand the role only after a review of real conversations shows replies that follow your rules.

Does this replace my calendar or my intake form? No. The agent works with the calendar and form you already use, and the record stays under your control. The goal is to reduce repetitive back and forth so you have more time for the client work that depends on your judgment.

Your next step

See where an AI agent fits in your company first.

Take the free Be Human AI diagnostic. A few questions about your work show which task to hand to an agent first, what it needs to know and where a person approves.

Be Human editorial guidance. Examples are illustrative unless explicitly attributed. Sources provide background; implementation choices depend on your workflow.

Less busywork. More room for you.

Bring us the work you wish you had time for.

Book a call to try a personalized AI agent free for 7 days ↗Or take the free AI diagnostic →