← All posts

From idea to video, with your clone at the prompter.

2026-10-24 · 7 min read · by Nabil BA-MOH

Founders do not run out of ideas for videos. They run out of time between "that would make a good video" and pressing record. The idea is free. The script, written the way you talk, cut up, loaded and ready, is what costs the evening.

This article is about closing that gap with a clone: an assistant that already knows how you talk, and that can now put the script on your teleprompter for you.

Why ideas die before the camera

Look at what has to happen between the idea and the take (one recording attempt, from start to stop). Someone has to write a script that sounds like you speaking, not like you writing. Someone has to cut it into chapters, short sections you can read off a screen and restart from. Someone has to paste it into the prompter, load it and, after every mistake, find the right place again. When you work alone, that someone is you, every time.

A generic AI assistant helps with the first step, then hands you a new job. It writes fast, but it writes like a stranger. It does not know your clients, the positions you have already taken in public, or the words you would never use. So you spend the session rewriting its draft until it sounds like you, and then you still paste and load it by hand.

A clone writes like you because it knows you

Klyr is a clone of you. It starts from the identity you write yourself, your DNA: who you are, what you are aiming for and how you want it to work with you. And it keeps a memory of your work across sessions: your decisions, your clients, the way you explain things. When it drafts a script, it starts from that, not from a blank page.

That does not make the first draft perfect. You still read it out loud, and some sentences will not sound like you. But the draft starts closer to how you talk, and the clone keeps your corrections. It learns something once, so you do not have to repeat yourself: the next script starts from what you fixed in this one.

The missing link: the clone can now work the prompter

Writing like you was only half the problem. The other half was that the assistant lived in a chat window and the teleprompter lived in another app, with you copying and pasting between them.

Mirrento is the teleprompter for Mac that I develop and record with. Your Klyr clone can drive it. All you have to do is ask, once Mirrento is connected. It writes the script into the prompter, loads it and moves between chapters while you record. Mirrento is sold separately: it is not included in Klyr, and it has its own 7-day trial on mirrento.com. If you want the details, they are on the Mirrento blog: what an assistant can and cannot do in Mirrento.

Just as important is what the clone cannot do there. It cannot record, cannot turn on the microphone, and cannot delete a script or a take. It cannot change the script while you are recording it, and it cannot switch to another script during a take: changes land between takes. It prepares the words. You press record.

What a session looks like

Here is the loop, with Mirrento open:

1. Give it the idea. A post you already published, a voice note you transcribed, a few bullet points. Ask for a short script to read out loud, one idea per chapter.
2. It writes the way you talk. It drafts from what it knows about you and saves the script in Mirrento. Your prompter keeps showing what it showed until you ask for the new script.
3. Read it once, out loud. Say what sounds wrong, in plain words: "too long", "I would never say that", "open with the client story". It edits the script, and if it is already on the prompter, the new text appears on the same chapter.
4. Record. Ask it to load the script. You start the take. After a mistake, ask for chapter three instead of scrolling for it.
5. Go over the take together. Mirrento notes where you slowed down, paused or started a sentence again, chapter by chapter. It keeps no transcript of what you said. Ask the clone which chapter dragged, and to rewrite it before the next take.

The point is not that the clone does more. It is that the chores that broke your focus now take one sentence from you instead of a task.

Where you stay in charge

A clone that can work your prompter raises a fair question: what does it see, and what else will it touch? Mirrento records the sound of each take, not the picture: you film with your own camera or recording app. After a take, it keeps that audio on your Mac, and next to it a note of where you slowed down, paused or went back, chapter by chapter. The note holds positions and timings, never your words. To know where you are in the script, Mirrento follows your voice with the speech recognition built into your Mac, and it throws the words away when the take ends. That note is all the clone can read about a take: it never gets the audio. And it does not record, does not delete, and does not touch the script you are recording. Klyr follows the same rule everywhere: it proposes, you decide, and it never acts behind your back. You choose the idea, you approve the words, and the camera only ever sees you.

FAQ