American Family Care (AFC), a US urgent care chain with more than 400 locations, has put a conversational voice AI from a vendor called Flip on its phone lines. CX Dive reported it on 13 November 2026. AFC started with a pilot at a handful of clinics. As of 22 September it had rolled the system out to 80 corporate clinics and more than 10 franchise groups.
The AI handles routine calls: booking, clinic hours, directions, billing, self-pay pricing, walk-in availability and lab-result status. It also rebooks cancelled and missed visits. Callers can ask for a person. The stated aim is that front desk staff can focus on patients already in the clinic instead of juggling the phone.
The excitement is about voice
Stories like this tend to get read as “voice AI is here, and conversation is moving to voice.” I’d test that before repeating it.
Is there a real need? Yes, and the article shows where it sits. Flip’s CEO says the typical urgent care operator has around a 30% missed call rate. AFC reports that call abandonment fell from 40% to 14%. The share of inbound calls that ended in a scheduled visit rose from 11% to 34%. These are AFC’s reported figures. The article gives no detail on how they were measured.
What changed? Phones used to force a choice between the person at the desk and the person on the line. Someone always lost. Now the routine call has somewhere else to go.
The channel is the phone, and that is all
AFC’s CEO said the phone is still where many patients go first when they need care quickly, so that is where they started. That sentence does not say voice is the best way to talk to a machine. It says patients were already calling, and AFC met them there.
I read this as a channel decision. Voice helps on input, which is why I dictate instructions to Claude with Wispr Flow. Voice helps much less on output and on two-way conversation, and I expect that gap to widen. More of these conversations will be agents talking to agents, and agents exchange data, which is text. A clinic’s phone line is a legacy channel worth serving well. It is not evidence about where conversation is heading.
What I expect customer service to look like
I think customer service teams in their current form will cease to exist. The transactional and operational work, such as booking, hours, pricing and status checks, no longer needs people. AFC’s call list reads like a list of exactly that work.
What stays is the human-to-human layer, and it becomes the premium one. AFC’s front desk is the clearest example in the article. Staff who are no longer answering the phone can look after the patient at the counter. In a clinic, that is where a human experience counts.
The opposite position, “we only use humans,” doesn’t hold up. Human-only service doesn’t scale economically. A competitor that answers every call instantly will take the bookings you let go to voicemail.
How an operator should measure it
Take a small clinic, or any service business with a busy front desk. Two measurements matter.
- Resolved calls. Count calls where the caller’s problem got solved. A booking made, a price given, a result status delivered. Don’t count calls that were merely deflected or avoided. AFC’s headline numbers lean this way, since scheduled visits and abandonment both track outcomes. The article does not report a resolved-call rate, so that is the number I would ask any vendor for.
- What staff do with the freed time. The article says clinic teams can spend more time on medical care and patient outcomes. It doesn’t say what AFC measures there. If you free up an hour of desk time and nobody can say where it went, you have saved a little money and changed nothing about the patient experience. Decide in advance. Greeting patients, following up after a visit, handling the cases that need judgement.
Also look at the handover. AFC lets callers ask for a person, though they may wait on hold depending on what is happening in the clinic. That wait is the part to design carefully. A caller who reaches a human should not have to repeat everything they just told the machine.
Where this touches my own work
AeroChat, which I build, works on the text side of this. Customers write in on channels like WhatsApp and chat, and the conversation turns into actions in the business’s own systems. I mention it only because it shows the other half of the picture. Meeting customers in the channel they already use is the principle, and the channel differs by business.
What a business should do
List every kind of call or message your team handles in a week. Mark which ones are transactional. Give those to automation, in whichever channel your customers already use. If that is the phone, use voice. If it is WhatsApp, use chat.
Then decide, before you switch anything on, what your people will do with the time. Write down the one or two moments where a person in front of the customer sets you apart, and put your staff there. Track resolved calls and that list, and ignore the rest.
Source: CX Dive: How a healthcare clinic chain uses conversational AI to improve patient experience