Most of the FAQs on this topic are written by marketers who have never heard a bad AI phone call. This one is written by people who have. If you're a MedSpa owner, a cosmetic practice operator, an orthodontist, or a therapist thinking about an AI receptionist, these are the questions we get on the first call — and the answers we actually give.

Will patients know they're talking to an AI?

Probably, within the first few sentences. Modern voice AI is good — good enough that older patients often don't clock it immediately — but anyone under 40 who's used ChatGPT will recognize the cadence within a turn or two. The better question is: does it matter? In our experience, patients care less about whether it's AI and more about whether it solves their problem quickly. A bad human phone experience is worse than a competent AI one.

That said, we recommend being transparent — a short disclosure at the start of the call builds trust and is required in some states. "Hi, this is the virtual receptionist for Dr. Patel's office — I can book appointments, answer questions about our services, or connect you to someone if you'd prefer."

What happens if the AI can't handle a call?

This is where the difference between a good and bad implementation shows up. A well-built agent has clear handoff rules: certain keywords, certain topics, and explicit requests for a human all trigger a transfer. The transfer itself should be a warm handoff when staff are available, and a "we'll call you back within X hours" when they're not — with that callback actually scheduled.

Don't accept a vendor whose answer is "oh, the agent handles almost everything." Push them: show me a recording of a handoff.

Does it work after hours?

Yes, and this is usually where the biggest ROI comes from. Most aesthetic practices, ortho offices, and therapy groups lose significant lead volume between 5 PM Friday and 9 AM Monday. A 24/7 agent turns that dead zone into booked consults.

What about HIPAA?

If you're a covered entity — most therapy practices, most medical-adjacent aesthetic practices with a physician on record, and most cosmetic dental and ortho offices are — you need a vendor who signs a Business Associate Agreement and can show you their privacy architecture. We wrote a separate post on this. If a vendor gets squirrelly when you ask about a BAA, walk.

How long does it take to set up?

Honestly? Between two and four weeks for a practice that takes the process seriously. The AI itself deploys fast — the script, the knowledge base, the handoff rules, and the calendar integration take work. Practices that try to go live in three days usually end up with an agent that sounds right in demos and fails in production. Build it once, build it right.

Can it handle emotional calls? Like a distressed therapy client?

No, and it shouldn't try. A good setup detects distress markers and routes to a crisis line or an on-call clinician within seconds — it does not attempt to comfort, counsel, or de-escalate. The appropriate response to a client in crisis is a human, immediately. Anyone who tells you their AI handles that is selling something dangerous.

For routine therapy intake — "Do you take my insurance?" "How long is the waitlist?" "Do you see couples?" — the AI is fine and arguably better than voicemail, because it actually responds.

Will it understand accents? Bilingual callers?

Modern models handle accents reasonably well — better than IVR systems, not as well as an attentive human. For practices with significant Spanish-speaking patient populations, a bilingual agent is often worth the extra cost. Test this during a demo with callers from your actual community; generic demos don't tell you what production looks like.

What does it cost?

For a single-location practice, expect $300–$900/month all-in, depending on volume and features. That typically includes the platform, the voice minutes, the integrations, and some level of ongoing tuning. Expect setup fees of $500–$2,500 for the initial build.

Enterprise or multi-location setups are quoted differently and are not well represented by the per-location price.

Can it book directly into my scheduling software?

It should. If the answer is no — if the agent only "sends a booking request" — it's barely doing the job. Ask specifically about your software: Nexhealth, Dentrix, Curve, Modento, SimplePractice, TheraNest, whatever you use. "We're working on an integration" is not the same as "it works."

Can I change the script on my own?

In most serious platforms, yes. Small changes (updated hours, new service, price update) should be a self-serve edit that takes five minutes. Bigger changes (new handoff logic, new use case) typically involve the vendor. If everything requires a support ticket, that's a bad sign.

What happens if the AI gets something wrong?

It will. The question is how fast you find out. Good platforms surface failures — flagged calls, handoff rates, drop-off points — in a dashboard you can actually read. Plan to spend 15 minutes a week in that dashboard for the first month, then monthly after. Practices that "set it and forget it" end up with agents that slowly drift from the actual business.

Is this going to replace our front-desk team?

At a healthy practice, no. It changes what they do. The routine stuff — reschedules, price questions, insurance lookups, after-hours intake — goes to the AI. Your front desk spends more of their day on patients in the chair, harder cases, and the calls that actually benefit from a warm human. Most practices we work with don't reduce headcount; they reduce burnout.

If your main goal is headcount reduction, be honest about that — it changes the implementation.

Is there a trial?

Most reputable vendors offer a 30-day pilot or a month-to-month arrangement. Be wary of annual contracts with no out. The pilot is for both sides — you're checking whether it works for your patients, and the vendor is checking whether you'll actually maintain the system.

How do I know it's working?

Three numbers to watch:

1. Answer rate — should go from 60–75% to 95%+.

2. Call-to-booking conversion — typically holds or improves compared to human handling of the same call.

3. After-hours booking volume — a new category, previously zero for most practices.

If all three move in the right direction after month one, it's working. If any of them regress or stay flat, the implementation needs tuning.

If you have a question we didn't cover, reach out. Real questions get real answers.