Do You Actually Need an AI Scribe? A Decision Framework for Mental Health Clinicians

The AI scribe question has become the default question. At every practice-management talk, every clinician listserv, every conference hallway: should I get one, and which one?

It is a reasonable question. It is also, for a substantial number of practices, the wrong first question — the loudest complaint rather than the largest leak.

This is a familiar clinical pattern. The presenting complaint is rarely the formulation.

Before evaluating vendors, get an honest number. For one week, track the minutes you spend on clinical documentation outside of session time. Not your impression — the actual minutes.

Then compare it against the other candidates for where your practice is losing time and money:

• Unfilled or no-showed appointment slots
• Inquiries that never convert to booked clients
• Claims that go out late, get denied, or never get resubmitted
• Time spent on scheduling logistics that could be automated
• Unbilled or under-coded sessions

If documentation genuinely tops that list, an AI scribe is likely the right investment. If it doesn't, buying one will make you feel more modern without making you more solvent — and it will consume the attention and budget that the actual problem needed.

Then compare your number against what the field actually reports, because most clinicians badly underestimate how far outside the norm they are — in either direction.

Behavioral health documentation burden runs roughly one to two hours of documentation per hour of direct clinical care, which is dramatically heavier than general medicine, where a routine visit can be charted in a few minutes. Estimates put documentation and administrative work at 25–50% of total work hours; for a clinician carrying 25–30 clients in a 40-hour week, that lands somewhere between 10 and 20 hours weekly. Surveys of behavioral health clinicians consistently find 60–70% regularly completing notes outside scheduled hours.

Two things follow from that.

First, if you are at 10+ hours weekly, you are not failing at documentation — you are experiencing the field's baseline, and a scribe is likely to be among your highest-leverage purchases.

Second, if you have somehow gotten your documentation down to three or four hours a week, you have already solved this problem better than most of your colleagues. Buying a scribe will yield real but modest returns, and your money is almost certainly better spent on whatever else is on your list above.

When documentation burden is real, the effect is not merely economic. Note-writing that spills into evenings and weekends is one of the more reliable contributors to clinician burnout, and burnout degrades clinical work in ways that are hard to see from inside it.

The honest framing is that an AI scribe buys back time and cognitive load. Whether that time converts into more sessions, better sessions, or simply a life outside of practice is a decision worth making deliberately rather than by default.

1. HIPAA posture and BAA. Non-negotiable. A vendor that will not sign a Business Associate Agreement is disqualified, regardless of how good the output is. Ask specifically where audio is processed, whether it is retained, and for how long.

2. Whether it was built for mental health at all. Most scribes were trained on and designed for general medical encounters. A psychotherapy session is structurally different from a fifteen-minute medical visit: longer, less procedural, and the clinically meaningful content is often affective and relational rather than factual. A scribe that produces an excellent HPI may produce a useless psychotherapy note.

3. Note format fit. Does it produce the format you actually use and your payers actually accept — SOAP, DAP, BIRP, or your own? Can you customize the template, or are you adapting your documentation to the software?

4. What the note omits. This matters more in our field than in most. A verbatim-leaning scribe may capture material that has no business being in a permanent record, particularly around third parties, or content a client would experience as exposing. Evaluate what the tool leaves out as carefully as what it captures.

5. Consent and disclosure. You will need a consent process for recording, and it should be an actual clinical conversation, not a form buried in intake paperwork. Some clients will decline. Some clinical presentations — paranoid, trauma-related, adolescent — make recording a live therapeutic issue rather than an administrative one. Plan for how you will handle that before you deploy, not after.

6. Cost against recovered hours. Monthly subscription versus hours actually saved, valued at your session rate. Run the arithmetic; the answer is often clearer than the marketing.

Does the note need to be as long as you are making it?

A meaningful share of documentation burden in private practice is self-imposed — notes written to a standard well beyond what payers require or what clinical utility justifies, often driven by anxiety about audit or litigation rather than by any actual requirement. Automating an over-long note produces an over-long note faster.

Before buying software to write your notes, it is worth checking what your notes actually need to contain.

Documentation is one of ten operational domains where practices leak time and money. It is the one clinicians talk about most, which is not the same as the one costing them most.

FlowGap's free scribe guide walks through vendor selection without requiring a signup. If you want to know whether documentation is genuinely your largest gap before you shop, the full Flow Factors assessment scores it against the other nine. Take the scribe quiz · Start the free assessment