Writing a Script an Avatar Can Deliver

What it is for

A script that reads well often looks wrong when an avatar delivers it: long clauses, written-only punctuation and numbers that have to be pronounced. This is how to write for the mouth instead of the page.

The skill Copy

An avatar delivers exactly what is written, with no instinct for breath. Write
for delivery, not for the page.

- **One idea per sentence, 12–18 words.** Long sentences produce a flat,
  hurried delivery with no natural pause, and that is what reads as "fake"
  more than the face does.
- **Write out numbers and abbreviations the way they are said.** "twenty
  twenty-six", "A P I", "dollar sign". A written "2026" or "API" is a coin
  flip.
- **Put a full stop where you want a breath.** Commas do not produce one.
- **Never use a parenthesis.** It produces a swallowed, lower-energy clause
  that looks like a glitch.
- **Avoid tongue-twisters and repeated sibilants.** They are the single most
  common cause of a re-render.

Structure the video the same way:
- First 5 seconds: what this is about. Not a greeting — a greeting from an
  avatar is where people leave.
- Then one point per 20–30 seconds, each with a visual change (slide, B-roll,
  framing) at the boundary. An unbroken talking head past 45 seconds loses
  attention regardless of the script.

Read it aloud once at speed before generating. Anything you stumble on, the
avatar will too.

How to install it Copy

1. Open the tool and start a new conversation, project or template.
2. Paste the text below wherever the tool keeps standing instructions
   — usually called system prompt, instructions, or persona.
3. Save it so it applies to every run rather than a single message.

Did it work for you?

Where it came from