Do AI scribes still work after the first few weeks?

For most clinicians who stay with one past the first month, yes. The ones that last share three traits. The note lands where you sign it in one tap, it is written in your own template, and it asks less of you each week rather than more. The ones that fail pick the wrong codes or forget the orders you promised in the room.

What changes after week three?

The novelty ends and the arithmetic starts.

In the first week you are comparing the draft with nothing. By week three you are comparing it with the note you would have written, which is a much harder test. That is where most clinicians quit. It is also where the ones who stay report the number that counts: the charts stopped following them home.

A nurse practitioner put it plainly. Since the scribe, almost no charts at home. That is the test. Not accuracy scores, not model names. Did you close the laptop before dinner.

Why does your template matter more than the model?

Because a note in the wrong shape has to be moved, and moving it is manual.

Clinicians in these threads rarely complain about the words. They complain about the journey. Copy here, paste there, fix the headings, split the sections. One physician was frustrated enough to write his own hotkey tool to move text from the scribe into the place he signs.

If a scribe cannot produce the note in your shape, in the place you sign it, it has not saved you the work. It has moved it.

Does a scribe put in orders for you?

Mostly not, and this is the sharpest edge.

One nurse practitioner described the failure that matters. The scribe forgot orders promised in the room and picked the wrong codes. A missed order is not a typing problem. It is a patient waiting for something that was never sent.

Until a tool is reliable here, treat the order list as yours. Say it out loud in the room so it is recorded, then check it against what you actually sent before you sign.

Is a free scribe good enough for a practice you own?

It depends on who is paying for the free one.

If your employer bought it, use it. Someone with a budget decided it was worth having. If you own the practice and you are weighing up a free tool, ask who pays for it. Free tools have a customer. If it is not you, you are the audience.

That is not a reason to refuse one. It is a reason to know what you are trading, and to be careful about what leaves the room.

How do you test one in ten minutes?

Take three patients, not thirty.

Record the visit. Let the tool write the history only. Dictate your own assessment and plan. Then time yourself from the patient leaving to the note being signed.

Do that on three ordinary visits, not your easiest ones. If the number is no better by the third, the tool is not for you, and it cost you ten minutes to find out.

The short version

Yes, for the clinicians who get past week three, and what carries them is where the note lands rather than how clever it is. Solanto is being built around that: your own template, in the place you sign it, in Cloud mode or in Private Mode, where the note never leaves your device.

Solanto has not shipped yet. The price is on the page rather than behind a call, and early access is open: solanto.ai/pricing

Forward this to the colleague who quit a scribe in week two. Ask whether it was the words or the journey.