Mansfield, TX
AI Appointment Request Chatbot for Mansfield TX Medical Practices
Someone lands on your Mansfield practice website at 9:15 on a Sunday evening. They want an appointment. Your contact page offers them a phone number that nobody will answer for fourteen hours and a form that promises a reply within one business day.
Book a Demo
Most of them leave. Not angrily, and not because your practice looked bad. They leave because they were ready to act and the site was not, and by Monday the impulse has passed or another practice has taken it.
Cleod9's website chat is tied to the AI Voice Concierge, so a visitor gets a conversation rather than a form. It answers the questions people actually ask before booking, collects appointment requests, and books directly when your rules allow it. And at any point the visitor can exit out to a real human. That exit is not a fallback for when the AI fails. It is a designed part of the experience and it should be visible from the first message.
Why the form underperforms
A contact form asks a visitor to do work with no feedback. They fill in six fields, press submit, and receive a page saying somebody will be in touch. There is no confirmation that a human will see it, no sense of when, and nothing they can ask.
Worse, the form asks for commitment before it gives anything. A person who is not yet sure your practice is right for them will not fill in their phone number to find out whether you take their insurance. They will go looking for that answer somewhere else, and the somewhere else is frequently a competitor's site that answers it plainly.
Chat inverts the order. The visitor asks first, gets something useful, and only then is asked for details. By the time the request for a name and number arrives, they have a reason to give it.
The questions that come before booking
Practices are consistently surprised by what visitors ask. It is rarely clinical. It is almost always logistics and eligibility, and the same handful of questions repeat.
- Do you take my insurance?
- Are you accepting new patients?
- How soon can I be seen?
- Where exactly are you and where do I park?
- What do I need to bring to a first visit?
- Do you see children, or patients my age?
Every one of those is answerable without a clinician and without any judgment call. Answering them well is most of the value, because a visitor who gets those six answers is a visitor who is ready to book.
Write the answers yourself rather than leaving them generic. The insurance answer in particular should name the plans you actually accept and say plainly what to do if theirs is not listed, because a vague answer on insurance sends people away faster than a negative one.
Collecting the appointment request
Once someone wants an appointment there are two possible endings, and both are legitimate.
The first is a booked appointment. The Concierge can book directly, and for established patients and routine visit types that is usually the right setting. The visitor leaves with a time, which is a materially different outcome from leaving with a submitted form.
The second is a captured request that a person confirms. Sensible for new patients, for anything where insurance has to be verified first, and for visit types where the schedule has constraints a rule cannot express. The visitor leaves knowing the practice has their request, what happens next, and when.
Whichever ending applies, keep the collection short. Name, date of birth, phone, the reason for the visit stated in a category rather than a paragraph, and their preference for timing. Anything else can be gathered at the visit or by a person on the confirmation call.
The exit to a person, and when to offer it
The visitor can always leave the automated conversation and reach a real human. Make that obvious rather than hidden, and offer it proactively in three situations.
When a visitor asks something clinical. The chat should not assess a symptom or advise on a medication, and the graceful move is to say so and offer to connect them to the office.
When a visitor repeats themselves or rephrases the same question twice. That is the signal that the automated path is not working for them, and continuing to try is how a helpful tool becomes an irritating one.
When a visitor says anything suggesting urgency. Urgency goes to a person, and anything suggesting an emergency gets told to call 911 immediately.
The Kite Contact System is the mechanism for the live side of this. A visitor clicks and is connected to an available person by chat, audio or video, from the browser, with nothing to download on either end. Where the AI handles the ordinary questions at scale, Kite handles the moment when a real conversation is what the person needs.
What it must not do
The same boundary that governs the phone governs the chat, and it should be written down before launch rather than assumed.
No assessment of symptoms. No advice about medications, including whether a dose can be skipped or doubled. No opinion on whether something can wait until Monday or warrants an emergency room tonight. These are clinical judgments and they belong to a clinician, and a chat window is a particularly bad place for them because the exchange is fast, informal, and permanently logged.
No confirmation of coverage as a guarantee. The chat can say which plans the practice accepts. It should not tell someone their specific visit will be covered, which depends on their plan details and the reason for the visit.
Have a clinician read the finished script and strike anything that drifts toward evaluation. Twenty minutes, before launch, and it is the most useful review in the project.
Where the widget goes
Placement matters more than practices expect, and the default of putting it everywhere is not the best setting.
It belongs on the pages where intent is highest: the contact page, the appointments or scheduling page, the new patient page, individual provider pages, and any service page describing a condition you treat. Those are the pages where somebody is close to acting.
It is less useful on a careers page or a patient portal login, where it interrupts a task rather than assisting one.
Set the opening behavior deliberately. A widget that expands automatically and covers the content on a phone screen produces closures rather than conversations. A quiet presence that the visitor opens when they want it converts better and annoys nobody.
After hours is when it earns its keep
Website traffic does not follow office hours. Evenings and weekends are heavily represented, and those are exactly the hours when the phone number on your contact page is a dead end.
Decide what happens outside business hours specifically, rather than letting the daytime configuration run unattended. The chat should still answer the eligibility and logistics questions, still collect requests, and be clear about when a person will follow up. The exit to a human should be honest about the hour: offering to connect someone to a person at eleven on a Saturday when nobody is there is worse than saying plainly that the office opens at eight and their request is already in the queue.
Following up by text
A visitor who gives their mobile number in chat has handed you the most reliable way to reach them. Cleod9 supports two-way SMS on the practice's business number, so the confirmation goes out from the same number the patient will see on caller ID later.
Two things to get right. Business texting from a ten digit number has to be registered through The Campaign Registry, which Cleod9 handles, but registration depends on the website carrying specific consent language and a compliant privacy policy. If the chat is collecting phone numbers, the consent disclosure needs to be part of that exchange rather than buried in a footer.
And a patient can revoke consent by any reasonable means, not only by replying STOP, with the request honored within ten business days. Since April 2026, an opt-out given in one context extends to your other messages too, so opt-outs need one shared record rather than living wherever they were received.
What to measure
- Conversations started per hundred visitors on the pages carrying the widget, which tells you whether the placement is right.
- Share of conversations ending in a booked appointment or a captured request, rather than the visitor closing the window.
- How often visitors take the exit to a human, and at what point. A spike at one question means that question needs rewriting.
- After-hours share of total conversations, which is usually the number that justifies the whole thing.
Read actual transcripts in the first fortnight. Every practice finds at least one question its site answers badly and one question it never thought to answer at all.
Privacy, kept simple
Collect the minimum the booking requires. Category and contact details, not narrative. A shorter exchange is both a better experience and less sensitive information sitting in a chat log.
Cleod9 states end-to-end encryption and its tools run in the browser with no plug-in downloads. Before patient information moves through any platform, put these to the vendor in writing:
- Will you sign a business associate agreement covering this service?
- Where are chat transcripts stored, and for how long by default?
- Can we set our own retention, and can a transcript be deleted on request?
- Who on your side can access transcripts, and through what process?
- What happens to our transcripts and our number if we leave?
This page is operational guidance rather than legal or compliance advice, and your privacy officer should set the policy.
Common questions
Can visitors reach a real person?
Yes, at any point. The exit to a human is part of the design rather than a failure path, and the Kite Contact System connects them by chat, audio or video without a download.
Will it answer medical questions?
No. It handles eligibility, logistics and scheduling, and clinical questions route to the office.
Can it book without staff involvement?
Yes, for the visit types and patient categories you allow. Many practices begin by capturing requests only and enable direct booking once they have read a few weeks of transcripts.
Does it work on phones?
Yes, and most of the traffic will be phones, which is the argument for a widget that stays quiet until opened rather than one that expands over the page.
Is this the same system that answers our calls?
Yes. The website chat is tied to the AI Voice Concierge, so the questions, the boundaries and the routing are configured once rather than maintained separately in two places.
The visitor who is not ready yet
Not everyone who opens the chat wants an appointment today, and treating every conversation as a booking attempt loses the ones that were never going to convert on the first visit.
Some visitors are comparing two or three practices. Some are checking whether a referral their physician mentioned is somewhere they can actually get to. Some are researching on behalf of a parent. For all of them the useful outcome is not a booking, it is a good answer and a reason to come back.
Give the chat somewhere to send those people. A link to the page that answers their question in more depth, an offer to text them the address and hours so it is on their phone when they need it, or simply a clear statement of how to book when they are ready. A conversation that ends with the visitor better informed and not pressured is a conversation that did its job.
Name the visit types in the patient's words
A large share of wrong bookings come from a menu written in the practice's language rather than the patient's.
Internal names carry meaning only staff have. A patient reading a list picks whichever option sounds closest, and if two sound similar they will choose either at random. The cost lands on the practice: a slot of the wrong length, a provider who was not expecting this, and a call to move it.
Name each type the way a patient would describe it and add one plain line saying who it is for. If two still sound alike afterward, the practice probably should not offer both for self-service.
Start with two or three, watch what people actually choose for a month, and rename anything selected for the wrong reasons.
Hold back part of the day
A practice that exposes every open slot loses the ability to fit in the patient who calls and genuinely needs to be seen this week.
Reserve a portion of each day for staff to allocate. It keeps the phone useful and it keeps the schedule flexible, and the decision should be deliberate rather than the accidental result of showing everything.
Set a floor as well as a ceiling. Nothing should be bookable inside the next day or two without a person seeing it, because a request made at eleven at night for nine the next morning arrives before anyone can prepare for it.
Review both after a month against what actually happened. Most practices adjust once and then leave them alone for a year.
The questions to ask any provider, and Cleod9's answers
A practice comparing providers should ask the same short set of questions of each and keep the answers in writing. Cleod9 has answered them as follows.
Will you sign a business associate agreement? Yes, through Wildix, the platform the service runs on. Which services does it cover? Voice, voicemail, video, call recording and transcription. What is excluded? SMS text messaging. Where do recordings live and for how long? In the platform's AWS environment, for a period the practice selects, from one week up to ten years. What independent audits do you hold? SOC 2 Type 1 and Type 2. How is call audio protected in transit? DTLS-SRTP for media, with TLS for signaling and web traffic.
Those answers are the vendor's half of the arrangement. The other half is the practice's own configuration, access decisions, training and documentation, and whether all of that meets the practice's obligations is a matter for its own privacy officer or counsel rather than for any vendor to assert.
Talking to Cleod9
Cleod9 is based in Dallas-Fort Worth and supports its customers directly, so a Mansfield practice deals with someone local. The platform is described on the Cleod9 services page, and the website requirements for texting registration are set out on the texting registration page.
Bring the six questions your front desk answers most often. Those become the chat, and the conversation stops being about software.