AI + HumanContact center for healthcare

Every patient call answered, every judgment call made by your team

The AI answers at once and finishes booking, reminders and forms. Your nurses and front desk keep symptoms, results and upset callers. Together they cover more calls, with better care, than either alone.

  • Answers on the first ring
  • Can say it is an AI, every call
  • Nurse gets a summary, not a repeat
Healthcare: Monday 8:10 am · front desk on other lines, overflowAI agent on the line
AI agentAnswers on the first ring
OutcomeVisit moved to 9:20 am today · cough flagged to nurse with summary

Move a visit

A parent needs to shift Thursday's 3:40 visit to Wednesday morning.

Refill running low

Three days of blood pressure medication left, request needs a clinician.

Should I come in?

A child with a three-day fever. A nurse takes this call.

Upset about a bill

A patient disputes a balance and needs someone with authority.

01 · Why AI + Human

Why patient phones will run on AI and people together

One in three healthcare calls goes unanswered, and VA scheduling lines abandoned 11% of calls against a 5% standard. New-patient waits average 31 days. A people-only front desk cannot answer every call at 8:10 am Monday. An AI-only line cannot judge a worsening cough or tell a parent bad news. Patient phones need both.

AI agentsLow cost per call, instant scale, consistent
  • Answers the 8:10 am Monday surge on the first ring
  • Books, moves and confirms visits from open slots
  • Asks the same approved safety questions on every call
On its ownIt cannot judge a worsening cough, approve a refill or comfort a parent who just heard a hard result.
Your peopleJudgment, authority, care, new solutions
  • Hears a red flag in a patient's voice and acts on it
  • Approves refills and decides when to see a child today
  • Delivers bad news and settles a disputed bill with care
On their ownA front desk cannot answer every line at lunch, after 5 pm and during the Monday rush while also checking in patients.
What the combination delivers in a clinicOne desk, one record of every call
  • Calls that used to ring out are booked, so fewer patients wait weeks for a slot
  • The nurse gets a one-line summary and the transcript, so a worried parent tells the story once
  • Staff stop typing refill and referral details and spend that time on the patient at the desk
Side by side
People only (traditional)
AI only
AI + Human
Answer speed
Busy lines, voicemail at lunch and after 5 pm
First ring, every hour, every call
First ring, with a nurse or desk staff one step away
Cost of a routine call
Staff time on every reschedule and form request
Low, billed per second
Routine calls cost little; staff time goes to complex ones
Consistency
Varies by person, shift and how busy the day is
Same approved script and safety questions each call
Same script on every call, with staff fixing what it missed
Judgment calls
Strong: a nurse weighs symptoms and context
Weak: it can spot red-flag words, not decide care
AI flags the red flag, a nurse decides what to do
Patient outcome
Good when reached; many calls are never answered
Fast but limited: it cannot approve or advise
More patients reached and cared for by the right person
How to think about the split
Give AI facts, forms and slots

If the answer is a fact, a form or an open slot, let the AI finish it. Hours, parking, plans accepted and reschedules are the place to start.

Keep judgment and bad news with people

Symptoms, results, refill approval and billing disputes need a clinician or a staff member with authority. Route them there with the history attached.

Judge the hand-off, not the voice

Measure whether the task was finished and whether urgent calls reached a person fast. A pleasant voice that loses a worried parent is not a win.

The short answer

Which patient calls should AI handle in a medical or dental practice, and which should stay with staff?

Let AI answer every call and finish the administrative ones: booking, rescheduling, confirming, office information and forms. Let it start refills, referrals and new-patient intake, then hand them to staff with a summary. Send symptoms, urgency, results and billing disputes straight to a person. Start with one narrow job, such as reschedules on overflow and after hours.

Reviewed November 2026 · 18 sources cited below
  • AI owns: book, move and confirm visits, office information, forms
  • AI starts, staff finish: new patients, refills, referrals, insurance
  • Straight to staff: symptoms, urgency, results, bad news, disputes
  • Pilot reschedules on overflow and after hours, not the main line
02 · The stakes

The phone is where access to care is won or lost.

Patients still call first, schedules are tight, and every unanswered call or empty slot has a cost. These numbers come from public audits and named surveys, not from vendors. Use them to compare against your own phone log.

32%

of healthcare calls missed, versus 14% in home services and 9% in real estate. CallRail benchmark of 1.1 million leads, 7 industries, January 2025.1

11%

of scheduling calls abandoned at VA clinical contact centers in fiscal 2024, against a 5% standard. 14 of 17 centers missed both service standards.2

31 days

average wait for a new-patient appointment across six specialties in 15 US metros, up from 26 days in 2022. AMN Healthcare survey of 1,391 offices, 2025.3

18.8%

average no-show rate across 10 clinics at one large VA medical center, 1997 to 2008, from 12.6% in audiology to 25.7% in gastroenterology. About $196 each in 2008.4

03 · The whole journey

One patient relationship, from first call to coming back

A patient touches your practice many times before and after a visit. AI answers and reaches out at each step. Your team steps in where judgment, clinical care or a difficult conversation is needed.

New patient inquirySTAGE 1
Intake and first visitSTAGE 2
Between visitsSTAGE 3
Recall and added servicesSTAGE 4
Billing and referralsSTAGE 5
Staying with the practiceSTAGE 6
Lapsed patientsSTAGE 7
The moment
Calls at 7 pm to ask about insurance and openings. Voicemail loses them.
Forms are unfinished and the patient is unsure what to bring or where to park.
Refill requests, test result questions and 'should I come in?' calls.
Overdue checkups and screenings go unbooked because no one has time to call.
A surprise bill or a stalled specialist referral leaves the patient frustrated.
Patients drift after a long wait or an unreturned call, and never say why.
Patients who haven't visited in over a year, some moved or switched.
AI agent
Answers at once, explains services and plans from your knowledge base, books the first visit.
Calls and texts with what to bring, directions and the forms link; confirms the slot.
Logs refill requests, answers routine questions, flags worrying words and hands off at once.
Calls overdue patients, offers open slots and mentions services the care team has approved.
Explains statements in plain words, checks referral status and sets a call-back.
Reaches out after visits for feedback, and records every request and complaint in one place.
Calls inside quiet hours and consent rules; offers a visit or records a request to stop.
Your team
Takes complex coverage questions and referral-only cases, with the transcript in front of them.
Handles patients who need help with forms, accessibility or a language not covered.
Nurses and clinicians decide on symptoms, results and prescriptions, using AI Assist notes.
Talks through hesitant patients' concerns and decides who needs a clinical conversation.
Settles disputes, arranges payment plans and calms upset patients.
Calls back unhappy patients, fixes the cause and tells the team what to change.
Speaks with patients who left unhappy, or who ask for a person, and invites them back.
What to track
Inquiries answered; first visits booked
First-visit no-show rate; forms done
Right-person hand-off; calls resolved first time
Recall booked; added services scheduled
Billing calls resolved; referrals completed
Return visit rate; open complaints
Lapsed patients reactivated

Read down a column for one stage, or across a row to see what the AI and your team each do from first contact to win-back.

04 · AI or person

Which patient calls should AI own, and which stay with your team?

Sort calls by what the answer needs. If it is a fact, a form or a slot on the schedule, the AI can finish it. If it needs clinical judgment, authority to decide or care for a worried person, a human takes it. Tag two weeks of your own calls before you decide.

AI handles it end to endAI starts, a person finishes with the contextStraight to a person
Call type
Volume
Who takes it
Why
Book, move or cancel a visit
A parent needs to move Thursday's 3:40 visit to Wednesday morning, or an established patient wants a physical within three weeks.
high
AI handles it
The rules are known: visit type, length, provider, location. The AI reads open slots, books, and texts a confirmation.
Reminders, confirmations and recall
Call or text the day before a 9:00 visit, or an overdue hygiene recall: confirm, move or cancel in the same call.
high
AI handles it
A Cochrane review found text reminders raised attendance about 14% (relative). One that can also rebook saves a slot a plain reminder cannot.
Hours, location, forms, what to bring
Where to park, which plans you accept, whether Dr. Rao sees children, a link to the intake form before the first visit.
high
AI handles it
Answers live in your knowledge base and rarely change. A wrong answer is easy to catch in a transcript review.
New-patient intake and first booking
A caller asks if you take their plan and when a first visit or cleaning is open. The AI takes name, birth date, reason and plan.
high
AI, then a person
The first call decides if the patient stays. Speed matters, but eligibility and clinical fit need a person's check before the day.
Prescription refill requests
A patient on blood pressure medication has three days left. Drug, dose, pharmacy and last fill are captured for the clinician.
medium
AI, then a person
The AI collects and routes. Only a clinician approves. A queue with a deadline beats a voicemail pile.
Referrals and specialist scheduling
A referral order has sat unscheduled for two weeks. The AI calls the patient, offers slots and flags missing records to staff.
medium
AI, then a person
In one large health system only about 35% of referral attempts ended in a documented visit. Chasing is repeatable work.
'Should I come in?' symptom calls
A parent says a child has had a fever for three days. A patient reports chest tightness or a sudden severe headache.
medium
Straight to a person
Advice rests on clinical judgment and protocol. The AI listens for red flags, says to call 911 if urgent, and brings a nurse in.
Test results, diagnoses, bad news
A patient calls about a biopsy result, or asks what a new diagnosis means for their work and their family.
low
Straight to a person
Personal and sensitive. A clinician decides what is said and when. An AI voice should never be the first to say it.
Bill disputes and upset callers
A patient disputes a balance after a visit, or has waited three weeks for a call back and is angry.
low
Straight to a person
It needs authority to adjust and real empathy. The person gets the history and the caller's mood, so the story is not told twice.
TakeawayThe line is not easy versus hard. It is fact versus judgment. Facts, forms and slots go to the AI. Judgment, authority and bad news go to people, with the AI having done the typing first.
05 · One call, start to finish

A refill request at 7:05 pm, from first ring to the chart.

One after-hours call, step by step. The AI does the gathering and filing. A clinician makes the decision. Nothing is approved by a machine.

0:00

Answers and says what it is

First ring: practice name, 'I'm an AI assistant, this call is recorded', then 'how can I help?' A person is offered at any point in the call.

0:20

Checks who is calling

It asks for full name and date of birth, finds the chart through your system, and reads back the phone number on file. No visit or drug detail is spoken until the check passes.

0:50

Asks the safety questions you approved

Your clinicians write them: 'Are you out of this medicine today?' 'Any chest pain, trouble breathing or fainting?' A yes goes to the on-call line, or to 911 advice.

1:30

Takes the request, reads it back

Drug, dose, pharmacy and last fill date are captured and read back. The AI does not approve, change or advise. It tells the patient when the clinic will respond.

2:15

Files the task and texts a receipt

A task lands in the refill queue with a one-line summary, the transcript and a deadline. The text says the request was received and leaves out the drug name.

8:15 am

Staff finish it

The medical assistant opens the task and checks the protocol. The clinician approves in one click. AI Assist drafts the note and disposition for the chart.

06 · Start small

Start with reschedules on overflow, not with the whole front desk.

Large health systems that moved fast still started narrow, listened to real calls first, and set grading and urgency rules before launch. Do the same at practice scale: one call type, one entry point, twelve weeks, and written conditions for what comes next.

Start with this

Reschedules and cancellations for existing patients, on overflow and after hours

Low clinical risk, one checkable outcome (the right slot, confirmed), and the calls that go unanswered today. Cancelled slots can then be offered to others. Refills and new patients come second.

How calls reach the AI

The main number rings your front desk as usual. If no one answers in four rings, at lunch, or after 5 pm, your phone system forwards to a VoiceGenie number. Nothing else changes.

Weeks 1-2

Baseline and scope

  • Tag two weeks of calls by reason; log unanswered, voicemail and abandoned calls by hour of day
  • Sign the business associate agreement and finish privacy and security review before any patient call
  • Write the do-not-answer list: symptoms, results, medication changes, billing disputes, anything urgent
  • Clean up appointment types, lengths and provider rules; the AI books only what the template allows
Weeks 3-4

Staff test

  • Staff place 40 scripted calls, 10 of them traps: 'my chest hurts', an angry caller, a wrong birth date
  • Read every transcript; fix wrong slots, long pauses and any urgent call that did not reach a person
  • Test the warm transfer to the nurse by day and the on-call line after hours, with the summary showing
Weeks 5-8

Live on overflow

  • Forward only the four-ring, lunch and after-hours calls; the main line stays as it is
  • Review every call in week one, then 20 percent plus every transfer; a clinician reads the urgent ones
  • Hold a 15-minute weekly review with the practice manager and update the knowledge base
  • Post five numbers weekly: answered, finished, transferred, corrected by staff, urgent routed
Weeks 9-12

Decide and add

  • Compare results with your baseline and the exit conditions below
  • If they hold, add one thing: new-patient booking, or outbound reminders for one provider's schedule
  • Keep refills and referrals for the next round; symptom calls stay with the nurse
  • If they do not hold, narrow the scope or stop. Nothing else in your phone system was changed

Expand when all of these are true

  • Every urgent or red-flag call in the audit reached a person within your set time, with no exceptions
  • At least 90% of audited reschedules were correct with no staff fix (planning range; set from your baseline)
  • Unanswered and abandoned calls in the forwarded windows fell at least by half against your baseline
  • No rise in patient complaints, and transferred calls needed no repeated questions in the audit sample
07 · What to measure

Measure access and outcomes, not how human the voice sounds.

Write definitions down and capture the baseline before you switch anything on. Published phone benchmarks for private practices are thin, so several baselines below come from large public systems or surveys. Yours may differ. Targets are planning ranges, not promises.

Metric
Typical today
Planning range with AI and people
Answer rate5
Inbound calls answered by a person or the AI, divided by all inbound calls, by hour of day. Voicemail counts as not answered.
54% answered by a person (Invoca, 2026)
95% or more answered
Abandonment rate2
Callers who hang up before anyone answers, divided by calls offered. Report scheduling lines on their own.
VA scheduling about 11%; standard 5% (2024)
Under 5%
Service level (80/30)2
Share of calls answered within 30 seconds, the VA's 80% in 30 seconds standard for scheduling lines. Measure by hour, not by day.
VA: about 57% in 30 s at the 14 centers that missed
80% in 30 seconds for staff; AI answers at once
AI task completion
AI calls where the request was finished and confirmed with no staff fix, found by audit. Not containment, which counts hang-ups as wins.
Not published; audit 30 calls to set yours
60 to 80% of routine scheduling calls
Urgent-call routing
Calls with red-flag words or urgency that reached a person within your set time. Review all of them, not a sample.
No benchmark; measure from the first week
100% reach a person; any miss triggers a review
No-show and late-cancel rate4
Visits where the patient did not arrive or cancelled too late to refill, divided by booked visits, by provider and visit type.
About 19% at a large VA center; clinics 13 to 26%
1 to 4 points lower
New-patient call conversion5
New-patient calls that end in a booked first visit, divided by new-patient calls. Review 100 calls a month to count it.
Invoca: 45% of leads convert on the call
55 to 65% of new-patient calls booked
Referral loop closure6
Referrals ordered that reach a documented completed visit, divided by referrals ordered, within 60 days.
35% of referral attempts completed (one system)
10 to 20 points higher

Planning ranges are targets to test in a pilot, not promises. Measure your own baseline for two weeks before you change anything.

08 · The return

What a group practice recovers when fewer calls go unanswered.

This example is a group of six to eight clinicians. Change the inputs to your own phone log. The calculator counts the AI minutes, the front-desk time freed, and the visits recovered from calls that used to go unanswered.

A worked example. Move the sliders to match yours.

3.5 minaverage call length
$5,500loaded monthly cost per person
60 a daycalls one person handles
$220 · 20%value of one kept visit from a recovered call · share of recovered calls that become one
Estimated monthly gain
$33,701a month
Today, per call
$3.67
With VoiceGenie
$1.72
720more calls answered a month
144extra kept visit from a recovered calls a month
$31,680value of those outcomes
1,440calls your team still takes, with context

AI at $0.14 a minute with volume credit applied, plus Insights seats for the people who take handed-over calls. Phone line charges are your carrier's and are not included. Illustration, not a quote. Open the full calculator

Where these numbers come from

  • Staff cost: BLS median pay for medical secretaries and administrative assistants is $45,930 7. At an estimated 1.43 loaded ratio that is about $5,500 a month.
  • Missed calls: CallRail's benchmark put healthcare at 32% 1. Invoca finds 54% answered by a person 5. Definitions differ, so the example uses 30%.
  • Conversion: Invoca reports 43% of answered healthcare calls are leads and 45% of leads convert on the call, about 19% of answered calls (derived) 5.
  • Visit value: $220 is a placeholder. Use your own revenue per visit. For scale, a VA study put the cost of one no-show at $196 in 2008 4.
  • Calls per person: 60 a day is a planning figure, not a benchmark. Front desk staff also check patients in and out.

What the dollar figure leaves out

  • Front desk staff spend less time on hold and in refill queues and more time with the patient in front of them
  • Every call is recorded, transcribed and QA-scored, so a complaint becomes a reviewable fact
  • Cancelled slots can be offered to other patients the same day instead of sitting empty
09 · Rules to get right

Rules you must get right before the first patient call.

This is a practical guide, not legal advice. HIPAA, federal calling rules and state AI laws all apply to patient calls. Ask counsel to review your specific states and services before launch.

US

Business associate agreement before the pilot8

A vendor that records or transcribes patient calls handles protected health information for you. HHS says a covered entity needs a compliant business associate contract with a cloud service that stores or processes ePHI. Sign it first.

US

Say less on voicemail9

HHS allows appointment reminders on answering machines but tells providers to limit what is said, such as name, number and what is needed to confirm. No diagnosis, result, drug or balance. Honor reasonable requests for confidential contact.

US

Reminder calls to cell phones have conditions10

Counsel summaries of the FCC health care exemption list: the number the patient gave, provider named up front, about one minute, one a day and three a week, easy opt-out honored at once, no marketing or billing. The FCC treats AI voices as artificial voices (2024).

US states

California: AI in clinical messages11

AB 3030 requires an AI notice and human-contact instructions on generative AI patient messages about clinical information. It excludes scheduling and billing, and messages a clinician has read and reviewed. Scope gets tested at the clinical edge.

US states

Texas: tell patients when AI is used12

TRAIGA (HB 149), effective January 1, 2026, requires providers to tell patients when AI is used in their treatment, in clear plain language. SB 1188 adds duties for AI in diagnosis. A first-sentence AI notice on every call helps, but ask counsel if it is enough.

US states

Recording consent varies by state13

Federal law needs one party's consent to record, but 13 states need every party's. Announce recording and AI at the start of every call, before any patient detail, and apply the strictest state you serve.

Four questions to ask any voice AI vendor, including us

  1. Will you sign a business associate agreement before any patient call, and what exactly does it cover?
  2. Is call audio or text used to train any model, and can we delete recordings and transcripts on request?
  3. What does the AI say and do when a caller mentions chest pain, bleeding or a child in trouble?
  4. How is identity checked before any appointment detail is read out, and where is it logged?
10 · Where it plugs in

The systems the AI has to read from and write to.

An AI that cannot see your schedule can answer questions but not finish a booking. VoiceGenie typically connects to these through the open API, webhooks or Zapier, and to HubSpot, Salesforce, Zoho and Google directly. Many practice systems limit write access, so the pilot can file tasks first.

Medical EHR and practice management

  • EPEpic
  • ATathenahealth
  • ECeClinicalWorks
  • NENextGen
  • MOModMed
  • TETebra

Dental practice management

  • DEDentrix
  • EAEaglesoft
  • ODOpen Dental
  • CDCurve Dental
  • DEDenticon

Scheduling, intake and patient messaging

  • ZOZocdoc
  • PHPhreesia
  • LHLuma Health
  • KLKlara
  • SOSolutionreach

CRM, calendars and phones

  • HUHubSpot
  • SASalesforce
  • ZOZoho
  • GCGoogle Calendar
  • TWTwilio
  • Any SIP trunk or PBX
Keep your numbers and carrierTwilio, Plivo, Telnyx, Vonage or any SIP trunk or PBX. Put the AI behind one menu option, on overflow or after hours.
CRM and calendarsHubSpot, Salesforce, Zoho and Google connect directly.
Everything elseThe systems above connect through the open API, post-call webhooks or Zapier.
11 · One platform

One platform for the AI, the desk and the tools behind your staff

You do not need one vendor for the dialer, another for the AI agent, a third for the agent desk and a fourth for tools that help staff. VoiceGenie is one platform. You set it up once, and every input improves both the AI and your people.

The usual way

Four tools, four set-ups

Dialerown contract · own integration
own training · own reports
Voice AI agentown contract · own integration
own training · own reports
Agent deskown contract · own integration
own training · own reports
Agent assist and QAown contract · own integration
own training · own reports
With VoiceGenie

One platform, set up once

AI agents
AI-native dialer
Human Agent Desk
AI Assist
Shared by the AI and your people
  • Your phone numbers and carrier
  • CRM and calendar connections
  • Knowledge and call rules
  • Recordings, QA and analytics
One connection to your schedule

Connect your scheduling system and calendars once. The AI books from it, and staff on the desk see the same visits.

One knowledge base

Write hours, plans accepted and prep rules once. The AI answers from it, and AI Assist shows the same answers to staff.

One recording and QA view

Every AI and staff call is recorded and scored in one view. A corrected answer or a QA finding improves the next call on both sides.

12 · What goes wrong

Five ways AI phone projects fail in clinics.

Switching the AI on for the whole main line first
Begin with overflow, lunch and after hours, so the front desk keeps every call it answers today and you learn from the rest.
Letting the AI answer 'should I come in?' questions
Write the red-flag list with your clinicians. Any hit goes to a nurse or the on-call line, or to 911 advice. Test it with trap calls before launch.
Judging success by how human it sounds or by 'containment'
Measure whether the request was finished and confirmed, from audited calls. A caller who gave up is not a win.
Automating a messy schedule template
Clean up visit types, lengths and provider rules first. The AI books what the template allows, so confusion spreads faster.
Reviewing privacy and the BAA after the pilot starts
Sign the agreement and complete the review in week one, with your privacy officer and counsel. Patient calls are not a place to find out later.
13 · Questions

Questions people ask before they start

Which patient calls can an AI phone agent handle?

An AI agent can finish booking, rescheduling, cancelling, confirming, office information and form links. It can start new-patient intake, refill requests and referral scheduling, then hand them to staff with a summary. Symptoms, urgent issues, results, bad news and billing disputes should go straight to a nurse, clinician or front desk person.

Is an AI voice agent HIPAA compliant?

Compliance depends on how it is set up, not on the word AI. Before any patient call you need a signed business associate agreement, limits on what is said, a way to delete recordings, and a clear way to opt out of model training on your audio. Ask every vendor for these in writing and have your privacy officer review them.

How much does an AI receptionist cost for a medical practice?

VoiceGenie charges $0.14 per minute for AI, billed per second, with the phone line not included. Human seats on annual billing are $29 for Desk, $59 for Insights and $99 for Copilot per person per month. There is free trial credit and no minimum commitment. Enterprise pricing and in-country deployment are quoted privately.

How long does it take to set up AI on patient phone lines?

Plan a twelve week pilot. Most of the time goes to baseline, writing the do-not-answer list, staff testing and call review, not to technology. Weeks one and two set scope and the privacy review, three and four test with staff, five to eight run live on overflow, and nine to twelve decide what to add.

Will patients hate talking to an AI?

Patients dislike being stuck, not a voice. Say it is an AI in the first sentence, finish the task, and offer a person at any point. Hand over with the history so the patient never repeats it. Systems that only talk and cannot book or route are the ones that frustrate callers.

What happens if a caller describes an emergency?

The AI should stop the routine flow, tell the caller to hang up and dial 911 if it is an emergency, and bring in a nurse or the on-call line with a summary. Your clinicians write the red-flag list and you test it with trap calls before launch. Review every urgent call after go-live.

Does it work with our EHR or dental software?

VoiceGenie typically connects to practice systems through its open API, webhooks and Zapier, and to HubSpot, Salesforce, Zoho and Google calendars. Whether it can read open slots and write bookings depends on what your system exposes. Many do not allow full write access, so the pilot may file a task for staff instead.

Can it speak with patients in Spanish and other languages?

VoiceGenie agents speak 90 or more languages, so scheduling and reminders can be offered in the patient's language. Test medical terms, names and place names in each language you serve before launch. For clinical conversations, keep qualified interpreters and ask counsel what language-access rules apply to your practice.

Try it on one call type

Hear it handle your calls before you change anything.

Talk to our AI agent in your browser, or start a trial and point one line at it. Your team keeps every call that needs them.

No credit card · No minimum commitment · Keep your carrier

Sources

  1. CallRail via Business Wire, CallRail benchmark report for small businesses (1.1 million leads, 7 industries) (2025).
  2. VA Office of Inspector General, Review of Clinical Contact Centers to Assess Leadership and Oversight, report 25-00228-214 (2025).
  3. AMN Healthcare, 2025 Survey of Physician Appointment Wait Times (1,391 offices, 15 metros) (2025).
  4. BMC Health Services Research, Prevalence, predictors and economic consequences of no-shows (Kheirkhah et al.) (2016).
  5. Invoca, The Invoca Healthcare Lead Conversion Benchmarks Report 2026 (2026).
  6. Journal of General Internal Medicine, Closing the Referral Loop: Primary Care Referrals to Specialists in a Large Health System (2018).
  7. US Bureau of Labor Statistics, Secretaries and Administrative Assistants, Occupational Outlook Handbook (2025).
  8. US Department of Health and Human Services, May a HIPAA covered entity use a cloud service to store or process ePHI? (FAQ 2075) (2023).
  9. US Department of Health and Human Services, May health care providers leave messages for patients at their homes? (HIPAA FAQ 198) (2022).
  10. Bass, Berry & Sims, TCPA Exemptions for Healthcare Companies (2018).
  11. Morgan Lewis, California Law Requiring Disclaimers by Healthcare Providers Using GenAI (AB 3030) (2024).
  12. Texas Medical Liability Trust, New AI disclosure requirements for physicians passed into Texas law (2025).
  13. KTS Law, Wiretap laws in the United States (2024).
  14. MGMA, Phones are still a backlog costing medical practices time (MGMA Stat poll, 294 responses) (2026).
  15. MGMA, Patient access priorities for 2026 (MGMA Stat poll, 236 responses) (2025).
  16. Cochrane Library, Text message reminders for improving attendance at health appointments (Cochrane CD007458) (2013).
  17. Cooley LLP, FCC: AI-generated robocalls are illegal under the TCPA (2024).
  18. Notable (vendor blog), Catholic Health: a practical guide for your next AI deployment (n.d.).