Talking to Patients About AI Scribes: What to Say in the First 30 Seconds

The first 30 seconds set the tone

Most patients have never seen a clinician use an AI scribe. The first time they encounter one, the script you use during the opening of the visit determines whether the rest of the encounter feels normal or feels weird. Get those first 30 seconds right and the AI sits invisibly in the background for the rest of the visit. Get them wrong and the patient stays distracted by the device the whole time.

This is a practical guide to the wording that works, the wording that doesn't, and what to say when patients ask the predictable follow-up questions.

What to actually say

Here is a script that consistently works for primary care, mental health, and specialty visits:

"Before we start, I want to let you know I use a tool that listens to our conversation and helps me write up the notes after, so I can pay full attention to you instead of typing. Nothing is shared outside our practice and it's deleted once the note is done. Is that okay with you?"

That is about 25 seconds delivered at a normal pace. It does four things:

1. Frames the tool as something that benefits the patient (you paying attention to them, not typing)
2. Addresses the privacy question before they ask it
3. Mentions deletion, which patients consistently ask about
4. Closes with a yes-or-no question that requires consent

The vast majority of patients say yes. A small percentage want more information. A very small percentage decline, and that is fine.

Words to avoid

A few phrases reliably make patients more anxious instead of less:

• "recording" carries connotations of surveillance. Use "listens" instead.
• "AI" is fine for some patients but spooks others. "A tool" is neutral and accurate.
• "transcript" sounds like a legal document. "Notes" sounds like what a doctor writes anyway.
• "sent to the cloud" raises a question you don't want to answer in 10 seconds. Just say "kept secure."
• "FDA approved" or any regulatory claim. Don't get into it unless asked.

The goal of the introduction is not to fully educate the patient about ambient AI clinical documentation. The goal is to get consent so you can focus on the visit. Education can happen if they ask.

The four questions you'll get most often

After a few hundred patient encounters using an AI scribe, the same four questions account for almost all of the follow-up:

"Where does this go?"

"It stays within our practice's account. The tool processes the audio to make the note, and we keep the note in your chart like any other visit. The audio itself is deleted once I sign the note."

The honest answer about your specific platform's data flow matters more than memorizing a script. Know how your scribe handles retention before you start using it with patients.

"Can I see the note before it goes in my chart?"

"Absolutely. I review and edit every note before it gets saved, and I'm happy to read through it with you at the end of the visit if you'd like."

Most patients who ask this don't actually want to read the note. They want reassurance that they could if they wanted. Offering openly gets you to a yes faster than any other answer.

"What if I say something I don't want recorded?"

"Just tell me and I'll pause it. Or I can leave it out of the note. The tool only writes what I confirm, so anything sensitive that we discuss but don't want documented stays between us."

This is especially common in mental health visits. Have a clear "pause" or "stop" workflow ready in your scribe tool. Practice it before you need it.

"Is this going to be used to train AI?"

"No. Our practice's contract with the tool specifically prevents that. Your conversation only gets used to make your note."

Confirm this is true for your platform before you say it. Many ambient scribes do use de-identified data for model training, and patients are right to be cautious. Choose a vendor that contractually excludes training use of patient data, and then you can answer this honestly.

When patients decline

Some patients will say no. Common reasons:

• They have privacy concerns specific to their care (mental health, substance use, immigration status, custody situations)
• They are uncomfortable with technology generally
• They had a bad experience with surveillance or recording in another context
• They want to think about it before agreeing

Your response should be unhurried and unbothered:

"Of course. Let's do this visit the regular way. We can revisit it next time if you change your mind."

Then close the scribe tool, take notes the old way, and move on. Patients who feel pressured will not return.

For Canadian providers, PIPEDA and provincial privacy laws require this kind of clear opt-out path. For US providers, state-by-state consent law varies, and the patient-consent state guide covers the legal floor in your jurisdiction.

Language matters in specialty visits

The generic opening script works for primary care. For specialties where the visit content is more sensitive, slight adjustments help:

Mental health and psychiatry:

"I want to make sure you're comfortable. I have a tool that helps me write notes by listening to our conversation, so I can give you my full attention. It's private, it's deleted after, and you can ask me to stop or skip something at any time. Is that okay, or would you rather we do this without it?"

The explicit "or would you rather we do this without it" matters. Patients in psychiatric care need the path of least resistance to be opting out, not opting in.

Pediatrics with a parent present:

"I want to let your parent know that I use a tool that listens during the visit so I can write up the notes without typing. It stays in our practice and gets deleted. Is that okay with you both?"

Address the parent, but also make eye contact with the child if they are old enough to understand. Including them in the consent conversation reduces parent anxiety.

OBGYN, sexual health, sensitive history-taking:

"I use a tool that helps me write the notes from our conversation. Some things we might talk about today are sensitive. The tool only writes what I keep in the note, and I'll skip anything you ask me to. Is that okay with you, or would you prefer we not use it today?"

The acknowledgment that you understand the sensitivity matters more than the tool itself. Patients who feel seen are far more likely to say yes.

Visible signals matter

Some practices put a small sign in the exam room: "We use AI tools to help with documentation. Ask your provider for details." A pre-visit notice in the waiting room or appointment confirmation email reduces the need for the in-room script. Patients who already know about it before sitting down don't need the full explanation; a quick "as the front desk mentioned, I'll be using the documentation tool" works.

If you have a website, put a one-paragraph explanation on a page like /privacy or /how-we-use-ai-tools. Link to it from the appointment confirmation. Patients who want to read more before the visit will do so on their own time.

What the data says about patient acceptance

Across published studies and clinic-reported numbers in 2026, patient acceptance of AI scribes when introduced clearly runs in the 88 to 95 percent range. The handful of patients who decline tend to be consistent: they have a clear privacy reason or a general discomfort with new technology. Acceptance rates rise the second time a patient sees the scribe in use, because by then they have the lived experience of the doctor making more eye contact instead of typing.

The clinics with the highest acceptance share three things:

1. A consistent opening script that every provider uses
2. A visible sign or pre-visit notice so it isn't a surprise
3. A genuine and unhurried willingness to skip it for patients who decline

If your acceptance rate is below 85 percent, the script or the framing is probably the issue more than the technology.

The compounding benefit

The first week of using an AI scribe involves having this conversation 20 to 50 times. By the second month, you have it less because returning patients already know. By month three, the conversation has become a one-line check-in: "tool is on as before, still okay?" That compounding makes the early effort worth it.

The conversation pattern is also generalizable. Once you have a way to introduce one piece of healthcare technology, you have a template for introducing the next one.

---

If you're trying ambient AI in your practice, Transcribe Health includes patient-facing consent screens and customizable opening scripts in the provider dashboard. The free trial lets you try them on real visits without committing to a contract.