Skip to content

Platform fluency

Every engine speaks a dialect.

The rules are universal, but the phrasing is not. Here is how each major AI platform likes to be prompted, with a working recipe for each.

Claude

Text

Rewards rich context and explicit reasoning. Give it the background, define the role, and ask it to think step by step before answering.

You are a senior editor. Context: [brief].
Task: tighten this paragraph.
Think through the changes, then give the final version.

ChatGPT

Text

Responds well to a clear system role and step-by-step framing. Ask for a specific output format and it will hold to it.

Act as a market analyst.
Summarize this report as 5 bullets,
each under 15 words.

Gemini

Text

Favors concise, well-scoped tasks and handles long documents well. Keep the ask narrow and the context clearly delimited.

From the text below, extract every date
and return them as a JSON array.
---
[document]

Midjourney

Image

Loves dense, comma-separated descriptions and parameters. Lead with subject, then style, then camera, then flags like aspect ratio and stylize.

a lone lighthouse at dawn, storm clouds,
cinematic, 35mm, dramatic side light
--ar 16:9 --stylize 250 --seed 42

DALL·E

Image

Prefers natural, complete sentences over keyword soup. Describe the scene as you would to a person, including the mood and the style.

A watercolor illustration of a cozy
bookshop on a rainy evening, warm light
spilling onto the wet pavement.

Stable Diffusion

Image

Gives you the most control: positive and negative prompts, weights, seeds, and samplers. Use negatives aggressively to remove artifacts.

Positive: portrait, soft light, 85mm
Negative: blurry, extra fingers, text
Seed: 12345  CFG: 7

Sora

Video

Describe the scene, the subject action, and the camera movement in one clear paragraph. Specify pacing to control how much happens per second.

A slow dolly-in on a woman reading by a
window, dust drifting in the light,
calm pacing, warm afternoon tones.

Runway

Video

Works best from a starting image plus a short motion prompt. Keep the motion description simple: one camera move, one subject action.

Start image: hero.png
Motion: gentle push forward,
hair moving slightly in the breeze.

Kling

Video

Handles longer clips and complex motion. Chain reference frames between clips to keep a character consistent across a sequence.

Clip 2 start = Clip 1 end frame
Motion: character turns and walks
toward camera, steady tracking shot.

Ready to put the rules to work?

Start at module one and build up to full image-and-video workflows.

Start the course