Home · Guides

AI front desk for clinics: what it handles, what staff still do

Published September 24, 2026 · Last reviewed September 24, 2026
By the ClinicConvo team

An AI front desk can already answer a clinic's calls and messages, book or collect appointment requests, gather intake details, send reminders, run insurance eligibility checks and take routine payments. Staff still handle clinical questions, emergencies, exceptions, upset patients, in-person problems and any decision that needs judgment, so in practice these tools take over the repetitive part of the job and hand the rest to people.

This guide goes task by task through a medical or dental front desk and shows which tasks products on the market say they perform today, using only each vendor's own website, read on September 24, 2026. Features change quickly, so confirm them with the vendor before you sign. If you are comparing products for phone and message answering specifically, see our AI receptionist buyer's guide.

What does a clinic front desk actually do?

The US Bureau of Labor Statistics (BLS) says receptionists in hospitals and doctors' offices may collect patients' personal information and direct them to the waiting room, and that some handle billing and insurance payments (BLS, Receptionists). Medical secretaries and administrative assistants, a separate BLS occupation, take simple medical histories, arrange hospital admissions and process insurance payments, among other tasks (BLS, Secretaries and Administrative Assistants).

For this guide we split front desk work into ten jobs: answering the phone, answering messages, scheduling, intake and registration, reminders, insurance checks, check-in, payments, routing requests such as refills and records, and handling whatever does not fit a script. The sections below take them one at a time.

Which front desk tasks can AI do today?

Answering phone calls

Answering calls is the task most vendors in this guide offer. Zocdoc says its assistant Zo answers on the first ring, collects insurance, follows the practice's scheduling rules and books into the EHR, and that it can route urgent calls (Zocdoc). Freed positions its Front Desk product, which it labels as in beta, as a receptionist for the calls the team cannot pick up, from overflow to after-hours to the lunch rush, with a warm transfer to staff during business hours (Freed). Phreesia says its VoiceAI handles inbound calls and runs outbound ones for appointments, payments, refills, recall and collections (Phreesia VoiceAI).

Not every call is handled end to end. Freed publishes one customer's results, from Premonition Health, as 79% of calls handled autonomously; that is a vendor-reported figure from a single practice, but it shows that even a well-running setup passes a share of calls to people.

Answering texts, chat and other messages

Weave says its AI Receptionist responds to text requests and missed calls with full AI text conversations (Weave AI). Hyro lists call center, SMS, website and mobile apps among its channels (Hyro), and Luma Health lists omnichannel messaging, AI-powered web chat and two-way AI translation (Luma Health). Which channel matters depends on where your patients already write; our WhatsApp for clinics guide covers usage by country.

Speed is the main advantage of software here. On our own platform, the AI's median reply time was 15.4 seconds and the 90th percentile 24.8 seconds (1,058 replies, 7 days ending August 25, 2026), and 32.7% of customer messages arrived outside business hours (88,892 messages to 61 businesses, July 30 to September 21, 2026). Both samples cover several industries, mostly in Brazil, and are not clinic-only (our data and method).

Scheduling

There are two approaches. Some tools book directly into the practice-management system or EHR; Weave says it books appointments with qualified PMS. Others collect the request for a person to confirm; Weave also offers capturing inbound appointment requests for approval and scheduling (Weave AI). Direct booking is faster for patients but depends on the integration and your scheduling rules. Our guide on whether AI can book appointments covers the trade-offs.

Intake and registration

Assort Health says its AI agents automate patient registration over the phone or digitally, collecting demographics, insurance and clinical history, and that patients can photograph their insurance card mid-call for the AI to read (Assort Health, Patient Intake). Phreesia sends patients a secure link to fill out registration forms, update insurance and demographic information and complete intake questionnaires before or after their visit (Phreesia).

Reminders and confirmations

Reminders do not need AI: most are rule-based messages sent a set time before the visit. NexHealth lists reminders among its communication products (NexHealth), Weave lists appointment reminders and text confirmations in its plans (Weave pricing), and Arini offers automated confirmations for dental practices (Arini). The research on what makes reminders work is summarized in our no-shows guide.

Insurance eligibility checks

Many tools run these automatically, but "automatic" does not always mean conversational AI. NexHealth offers automated eligibility checks, before the visit or on demand (NexHealth Verification). Weave describes its dental Insurance Eligibility product as using robotic process automation, meaning software that logs into payer websites and extracts the data (Weave). Phreesia says it runs unlimited, automated eligibility and benefits checks in real time (Phreesia). Luma lists eligibility verification and prior authorization in its payments area (Luma Health), and Assort says eligibility checks run during its intake flow (Assort Health).

Check-in

Several vendors offer self check-in. Phreesia says it lets patients self-check-in (Phreesia), and Luma lists clinic flow and check-in management and in-clinic self-service (Luma Health). This is the task where a physical person matters most: someone still has to greet walk-ins, help patients who cannot use a phone or kiosk, and deal with whatever goes wrong in the waiting room.

Payments and balances

Weave lists taking payments among its AI Receptionist's tasks (Weave AI). Phreesia says VoiceAI lets patients check balances or pay bills anytime (Phreesia VoiceAI). Assort says its AI agents contact patients with open balances and guide them through payment (Assort Health, Payment Resolution).

Refills, records and other requests

Freed says Front Desk collects new patient information, visit reasons, refill requests and referral inquiries (Freed). Hyro lists prescription refills and status updates among its skills (Hyro). Phreesia goes further and says VoiceAI resolves routine requests, including prescription refills, billing questions and medical records, without staff involvement (Phreesia VoiceAI). Before switching any of this on, ask exactly what "resolves" means in your setup: who approves a refill, and what the patient is told while they wait.

Task by task: can AI do it today?

Based on the vendor pages above, read September 24, 2026. "Example vendor" means a company that says it offers this, not a recommendation.

What still needs a human, and why?

• Clinical judgment. Deciding whether a symptom can wait is not a front desk task for software. In a 2024 study in The Lancet Digital Health, six radiation oncologists reviewed GPT-4 drafts replying to 100 simulated patient messages; they judged that the unedited drafts posed a risk of severe harm in 11 of 156 survey responses (7.1%) and of death in one (0.6%), mostly because the drafts misjudged how urgent the situation was (Chen et al.). That study tested clinical replies, not front desk software, but it is the reason to keep clinical questions away from the AI.
• Confident wrong answers. ECRI, an independent patient safety organization, put misuse of AI chatbots in healthcare at the top of its 2026 list of health technology hazards, noting that these systems are built to sound confident and always give an answer (ECRI, January 2026). ECRI was writing about general-purpose chatbots, but the same failure can happen in a front desk tool that is asked something outside its instructions.
• Patients who want a person. Surveys point both ways, and both of the recent ones come from companies with a stake. WellReceived, which sells human medical reception services, reported that 90% of patients prefer talking to a real person instead of AI (OnePoll survey of 6,000 consumers in the US, UK and Canada, June 2026) (WellReceived). Salesforce, which sells AI agents, surveyed 3,200 adults in eight countries (March 24 to April 10, 2026): 67% said they would rather have 24/7 AI help than wait for someone to pick up during office hours, and 89% said a clear "escalate to human" option is essential for trusting AI administrative support (Salesforce). The point both agree on: the route to a person has to be easy.
• Clinical AI is a harder sell than admin AI. In an independent Pew Research Center survey of 11,004 US adults (December 2022), six in ten said they would feel uncomfortable if their own provider relied on AI to do things like diagnose disease and recommend treatments (Pew). Keeping the AI to logistics is both safer and easier for patients to accept.
• Exceptions and the physical office. Billing disputes, a patient who arrives on the wrong day, a family managing an elderly parent's appointments, the person at the counter who is upset: these are judgment calls, and they are a large part of the job that remains.

Will an AI front desk replace receptionists?

Official projections point to a shift, not a disappearance. The BLS projects employment of receptionists to decline 2 percent from 2025 to 2035, from 947,500 jobs in 2025, and says employment is expected to be constrained as organizations continue to automate or consolidate administrative functions. It still projects about 105,100 openings for receptionists each year, on average, and healthcare and social assistance employed 42% of receptionists (BLS, Receptionists). Medical secretaries and administrative assistants move the other way: 979,000 jobs in 2025, projected to reach 1,026,100 in 2035 (5%), which the BLS attributes primarily to the growth of the healthcare industry (BLS, Secretaries and Administrative Assistants).

For cost planning, the BLS reports a median wage for receptionists of US$ 18.27 an hour (US$ 38,010 a year) in May 2025. We did not find an official source that measures how much front desk time AI tools save in clinics, so we don't estimate one here; ask vendors for data from practices like yours and check it against your own call and message logs.

What privacy rules apply?

• United States. Under HIPAA, a vendor that creates, receives, maintains or transmits protected health information for you is a business associate, and HHS lists a health app developer providing patient messaging as an example. You need a business associate agreement before patient data flows (HHS).
• EU and UK. Data concerning health is special category data under the GDPR (ICO), and a vendor processing it for you needs a contract that meets Article 28 (ICO).
• Elsewhere. Most countries have their own health data and AI disclosure rules. Our AI receptionist guide covers California's AB 3030, the EU AI Act's transparency duty, Brazil and Canada. This is not legal advice; check with your regulator or lawyer.

When is an AI front desk not the right choice?

• You only need reminders. Your practice-management or patient communication system may already send them; Weave, for example, lists appointment reminders in its plans (Weave pricing).
• Your after-hours calls often need a clinician. A nurse line or on-call service handles what software must not. Our after-hours guide compares the options.
• Your patients strongly prefer a person. Human answering services for clinics exist; WellReceived, for example, describes itself as providing 24/7 medical reception services (WellReceived).
• The vendor won't sign the privacy contract you need. Then don't send it patient data.
• Your volume is low and your team keeps up. Measure missed calls and unanswered messages first. If there are few, the problem an AI solves may not exist in your office.

Implementation checklist

Sources

Vendor pages read September 24, 2026.

1. US Bureau of Labor Statistics. Occupational Outlook Handbook: Receptionists.
2. US Bureau of Labor Statistics. Occupational Outlook Handbook: Secretaries and Administrative Assistants.
3. Zocdoc. Zo, AI phone assistant.
4. Freed. Front Desk.
5. Phreesia. Home, VoiceAI, End front-desk chaos and Increase revenue.
6. Weave. AI, pricing and Introducing Weave Insurance Eligibility.
7. Hyro. Home.
8. Luma Health. Home.
9. Assort Health. Patient Intake and Payment Resolution.
10. NexHealth. Home and Verification.
11. Arini. Home.
12. Chen S et al. The effect of using a large language model to respond to patient messages. Lancet Digit Health. 2024. doi:10.1016/S2589-7500(24)00060-8 (PMC11829255).
13. ECRI. Misuse of AI chatbots tops annual list of health technology hazards. January 21, 2026.
14. WellReceived. New Research Reveals Patient Frustrations With AI. July 15, 2026.
15. Salesforce. Patients Trust Their Doctor's AI Agents 3x More Than Public AI. June 24, 2026.
16. Pew Research Center. 60% of Americans Would Be Uncomfortable With Provider Relying on AI in Their Own Health Care. February 22, 2023.
17. US Department of Health and Human Services. Business Associates.
18. Information Commissioner's Office. Special category data and What needs to be included in the contract?
19. ClinicConvo. Our data.