AI in the Front Office

Website Chat Assistants for Practices: Do Patients Use Them?

Do patients use chat on a practice website? What survey data says, the use cases that work, what chat must not do, and how it fits with phones.

AssistCall research teamPublished Reviewed by Faruque Amin, VP of Sales and Marketing7 min read

On this page
  1. Do patients use chat at all?
  2. Who is typing, and on what
  3. What patients ask a practice chat
  4. What chat must not do
  5. Privacy and trust
  6. How chat fits with phones and booking
  7. Building the answer library
  8. Accessibility is not optional
  9. Example: a fictional dermatology practice
  10. Measuring whether chat is used and useful
  11. Frequently asked questions
  12. References
AI in the Front Office: Website Chat Assistants for Practices: Do Patients Use Them?Article coverAI IN THE FRONT OFFICEWebsite Chat Assistantsfor Practices: DoPatients Use Them?Ava Medical IntakeInsightsAI in the Front Office: Website Chat Assistants for Practices: Do Patients Use Them?Article coverAI IN THE FRONT OFFICEWebsite ChatAssistants forPractices: DoPatients Use Them?Ava Medical Intake

Key takeaways

  • Patients already turn to AI for health questions: about a third of US adults did so in early 2026[1].
  • Speed is the draw. Most users said they wanted quick, immediate advice[1], so a practice chat should answer office questions instantly and hand off everything else cleanly.
  • Chat works best for office facts, new-patient questions, booking requests and form help. It should never give clinical advice.
  • Privacy worries are widespread, so collect as little as possible in chat and tell patients where personal questions should go.

Do patients use chat at all?

The short answer is yes, though not always on your website. In a KFF poll from February and March 2026, about a third (32%) of US adults said they were turning to AI for health information and advice[1]. Among those users, 65% said a desire for quick and immediate advice was a major reason[1]. About one in five, 18%, said a major reason was not having a regular provider or not being able to get an appointment, rising to 38% among users aged 18 to 29[1].

Major reasons US adults use AI for health information, 2026Share of adults who use AI for health information or advice (KFF poll, Feb–Mar 2026)
Major reasons US adults use AI for health information, 2026Bar chart. Want quick, immediate advice: 65%; No regular provider or can't get an appointment (users aged 18–29): 38%; No regular provider or can't get an appointment (all users): 18%.Want quick, immediate adviceWant quick, immediate advice: 65%65%No regular provider or can't get an appointment (users aged 18–29)No regular provider or can't get an appointment (users aged 18–29): 38%38%No regular provider or can't get an appointment (all users)No regular provider or can't get an appointment (all users): 18%18%Major reasons US adults use AI for health information, 2026Bar chart. Want quick, immediate advice: 65%; No regular provider or can't get an appointment (users aged 18–29): 38%; No regular provider or can't get an appointment (all users): 18%.Want quick, immediate adviceWant quick, immediate advice: 65%65%No regular provider or can't get anappointment (users aged 18–29)No regular provider or can't get an appointment (users aged 18–29): 38%38%No regular provider or can't get anappointment (all users)No regular provider or can't get an appointment (all users): 18%18%

Source: KFF, 2026[1]

Show the data
ItemValue
Want quick, immediate advice65%
No regular provider or can't get an appointment (users aged 18–29)38%
No regular provider or can't get an appointment (all users)18%

For a practice, that data cuts two ways. It shows patients are comfortable typing questions into a box and expect a fast reply. It also shows that when they cannot reach a practice, some of them get health advice from general tools that know nothing about their history. A practice chat will not and should not replace a clinician, but it can make the practice easier to reach for the many questions that are not clinical at all.

Who is typing, and on what

Most visitors to a practice website arrive on a phone. About nine in 10 US adults own a smartphone[2], so the chat has to work with one thumb on a small screen: short answers, big tap targets, and links that open the right page rather than a PDF.

Many of your patients are also used to doing health admin online already. In 2024, 65% of US individuals were offered and accessed their online medical records or a patient portal, up from 25% a decade earlier[3]. That changes what a website chat is for. Established patients with portal access should be pointed to the portal for anything personal, because it is the secure, identity-checked channel. Website chat then serves three groups well: prospective patients deciding whether to book, established patients with quick office questions, and anyone who cannot get into the portal and needs help.

Design each answer with those groups in mind. A prospective patient asking about plans needs an answer and a booking route. An established patient asking about results needs a polite redirect to the portal and the nurse line. A locked-out patient needs portal help steps and a phone number. The same question can deserve different next steps depending on who is asking, so let the chat ask one simple question early, such as are you a current patient, before it suggests a route.

What patients ask a practice chat

Before choosing a tool, read a month of your contact form submissions, portal messages marked non-clinical and front desk call notes. The questions that keep repeating are your chat's job description. In most outpatient practices they fall into a handful of groups.

Typical website chat use cases for an outpatient practice.
Use caseFit for chatWhat a good answer includes
Hours, locations, parking, holiday closuresStrongAnswer plus link to the location page
Are you accepting new patients? Do you take my plan?Strong, from a current listAnswer, a reminder to confirm cover with the insurer, and a booking route
Booking or changing an appointmentGood, if connected to real availability or a request queueBooking link or a captured request with a callback time
Forms and portal helpStrongLinks, short instructions and who to call if stuck
Billing questionsPartialGeneral policy; personal balances routed to billing
Symptoms, results, medicationsNot a chat taskEmergency message where relevant, then the nurse line or portal

Scheduling deserves attention. Only 11% of medical group leaders said a majority of their patients self-schedule, and 73% said a quarter or fewer use digital scheduling tools[4]. A chat that helps a patient find the right visit type and either book it or leave a structured request can catch people who would never use a bare scheduling widget.

What chat must not do

  • Assess symptoms or suggest whether a patient should be seen. Any symptom language triggers a fixed emergency message and a route to the nurse line or portal.
  • Discuss a specific patient's records, results, balance or appointments without verified identity through a secure channel. A public website chat is not that channel.
  • Answer from the open internet. Answers should come only from facts your practice has approved.
  • Collect more than it needs. Name, contact details and the type of request are usually enough to route a message.
  • Pretend to be a person. Label the chat clearly as an automated assistant, with an obvious way to reach staff.

Privacy and trust

Patients worry about what happens to health details typed into AI tools. In the same KFF poll, 77% of the public said they were concerned about the privacy of personal medical information provided to AI tools[1]. Meet that concern head on in the chat itself.

Opening message: β€œHi, I'm the practice's automated assistant. I can help with hours, locations, new-patient questions, appointment requests and forms. Please don't share medical details here. For questions about your health or results, use the patient portal or call us. If this is an emergency, call 911.”

Behind the scenes, ask the vendor where chat transcripts are stored, for how long and who can access them, and whether it will sign a business associate agreement if patient information could end up in transcripts. Assume some patients will share health details despite the warning, and plan retention and access accordingly.

How chat fits with phones and booking

Chat should be one more door into the same front office, not a separate system with its own answers. Patients who start in chat and then phone should hear the same facts, and requests from either channel should land in the same queues.

  1. One answer library

    Phone, chat and website pages draw on one approved set of facts, owned by one person and updated when anything changes.

  2. One request queue

    Booking requests, callback requests and form problems from chat land in the same tracked queues as phone messages.

  3. Clear handoffs

    During hours, offer a callback or a phone number with the expected wait. After hours, give a realistic response time.

  4. Shared review

    Review a sample of chat and phone conversations together each week, so a wrong answer is fixed everywhere at once.

Building the answer library

  1. List the top questions from your call notes, contact forms and portal messages.
  2. Write one short approved answer for each, with a link where it helps.
  3. Mark every answer with an owner and a review date. Hours and accepted plans need checking at least monthly.
  4. Add a fallback for anything not in the library: the chat says it is not sure and offers a route to staff.
  5. Test with fictional questions, including awkward phrasing, typos and questions in the languages your patients use.

The fallback matters more than the library size. A chat that says I don't have that answer, here's how to reach the team builds more trust than one that guesses.

Accessibility is not optional

Chat widgets are web content. Under the amended Section 504 rule, recipients of HHS federal financial assistance with 15 or more employees must make web content and mobile apps meet WCAG 2.1 Level AA beginning May 11, 2027, and smaller recipients by May 10, 2028[5]. Ask vendors for an accessibility conformance report, and test the widget with a keyboard only and with a screen reader. Check that it does not cover your page's own navigation on a phone.

Example: a fictional dermatology practice

Example: a fictional two-location dermatology practice adds chat to its website with an answer library of about forty questions. In the first month, the most common topics are whether the practice sees new patients for skin checks, which location has parking, cosmetic consultation availability, and portal login problems.

Example, continued: the weekly review shows two gaps. Patients asking about a mole that changed were correctly sent to the nurse line, but the after-hours message did not say when the nurse line opens. And cosmetic questions often ended with no next step. The practice adds nurse line hours to the clinical handoff and a consultation request form to cosmetic answers. Chat requests now land in the same queue as phone requests, so the front desk works one list.

Measuring whether chat is used and useful

  • Conversations per week and the share that end with an answer, a booking or a captured request.
  • Fallback rate: questions the library could not answer. Each one is a candidate for a new approved answer.
  • Handoffs that reached a person within the promised time.
  • Booking requests from chat that became appointments.
  • Phone calls about topics now answered in chat, which should fall over time.

Patients' feelings about automated voices are covered in will patients talk to an AI voice, and texting rules in AI-assisted patient texting.

Frequently asked questions

Do patients use chatbots on doctors' websites?

Many patients are comfortable typing health-related questions into AI tools; about a third of US adults used AI for health information in early 2026. On a practice site, chat gets used most for hours, new-patient questions, booking requests and forms.

Can a practice website chat answer medical questions?

It should not. A practice chat should answer approved office questions, route symptoms to an emergency message or the nurse line, and send personal questions to the secure portal or phone.

Does website chat need to be HIPAA compliant?

If patient information could end up in chat transcripts, treat the vendor as a business associate: get a signed agreement, limit what chat collects, and set retention and access controls.

References

  1. KFF (2026). KFF Tracking Poll on Health Information and Trust: Use of AI For Health Information and Advice (fielded Feb 24 - Mar 2, 2026) (opens in a new tab) ↑

  2. Pew Research Center (2026). Americans' Use of Mobile Devices in 2026 (Mobile Fact Sheet) (opens in a new tab) ↑

  3. Office of the National Coordinator for Health IT (ASTP/ONC), HHS (2024). Individuals' Access and Use of Patient Portals and Smartphone Health Apps, 2024 (Data Brief No. 77) (opens in a new tab) ↑

  4. Medical Group Management Association (MGMA) (2024). Putting the power of scheduling into patients' hands (opens in a new tab) ↑

  5. Legal Information Institute, Cornell Law School (Electronic Code of Federal Regulations text) (2026). 45 CFR Β§ 84.84 - Requirements for web and mobile accessibility (opens in a new tab) ↑

Book a demo

See Ava handle your calls.

Book a short demo. We’ll walk through your calls and workflow, and agree what Ava should take care of.

Book a demo