A teleprompter for therapists — one that honours the pause

Psychoeducation content lives or dies on delivery. A video explaining the window of tolerance, what a panic attack actually does in the body, how grounding works, what a first EMDR session feels like — the manner is part of the message. A rushed video about slowing down contradicts itself before the second sentence. And on this material, silence is a tool: you pause after a heavy line because the viewer needs a beat to land it, the same beat you would leave in the room. And the audience is often someone at their most hesitant — a client between sessions, a partner trying to understand a diagnosis, somebody typing “why can I not calm down” into a search bar at two in the morning.

A conventional teleprompter fights exactly that. It scrolls at a set speed, so it keeps rolling through your deliberate pause — and now you are chasing the text or losing your place, and the tool’s tempo has replaced yours in a video whose entire value was your tempo. Structure suffers too. The signposting has to be said, every time: this is education, not therapy; it does not replace your own therapist; if you are in crisis, here is where to go. Improvised endings are where that goes missing. A psychoeducation video without its signposting is not a smaller version of the same thing; it is a different thing, and not one a clinician wants circulating with their name on it.

And there is a quieter problem: the register lives in the eyes. A person deciding whether to book a first session is watching your face to decide whether you feel safe to talk to. Visible reading reads as being processed. Meanwhile, a profession built on confidentiality is rightly wary of tools that send voice to a cloud — and rehearsing this material is itself sensitive. You want honest feedback on how you deliver a suicide-signposting line. You do not want an audience for the rehearsal.

Most therapists improvise from bullet points, because warmth survives improvisation. Structure does not. The five-minute explainer becomes nine, the order of ideas drifts, and by take three the crisis signposting has fallen out without anyone noticing. Scripting and memorising works for a thirty-second reel and collapses at the longer piece — so the longer pieces, the ones that would actually help a client between sessions, never get made. What does get made is the thirty-second version of a five-minute idea, which is how nuanced material ends up flattened into advice-shaped fragments.

The other route is a standard prompter app at a fixed speed, with the pause problem above. Some ask a colleague or a peer-supervision group to review drafts — genuinely useful, but it means an audience for material you are still finding your voice in, and it turns a thirty-minute job into a week of waiting.

The prompter follows your voice, which means it honours the pause. Stop for three full seconds after the line that needs it and the script does not move; speak, and it follows; loop back to say a sentence again more gently, and it keeps up. Your pacing is the input, not the output. The speech recognition doing this runs on the device — nothing you say while rehearsing or recording is uploaded anywhere.

When a steady rhythm is the point, switch to timed scrolling instead: a breathing exercise wants an even pace, not a responsive one, and you can set that pace and let it run, with a progress bar showing the time remaining. Manual scrolling is there for the segments you want under your thumb entirely.

The delivery summary is private rehearsal feedback. Pace in words per minute answers the question you cannot answer from inside your own head — is the grounding script actually slow, or does it only feel slow to you? Filler words are counted, and script coverage tells you whether everything you meant to say was said, the signposting included. All of it is measured on the device from the take itself, and none of it leaves the phone. It is the note a good supervisor would give you, without anyone watching. There is no streak, no score to chase, no feed — a number, once, for you.

Half of psychoeducation is audio anyway. Guided grounding, a sleep exercise, a podcast segment — record audio-only takes with the same prompter and the same pause behaviour, so the delivery you practised for video carries straight over to the recordings clients will actually fall asleep to.

A therapist is recording a five-minute explainer on the window of tolerance for her practice website. The script follows her voice; she pauses twice, deliberately, after the two lines that carry the most weight, and the text waits both times. The summary afterwards reads 118 words per minute with full coverage, crisis line included. Yesterday’s first attempt read 150 — the number confirmed what her ear suspected, privately, with nobody else in the loop.

Then the companion piece: a four-minute grounding exercise, recorded as an audio-only take with timed scrolling at a fixed, slow pace, because in that piece the unvarying rhythm is the exercise. Both recordings sit in the same library, and nothing about either — not the audio, not the analysis, not the script — has left the phone.