NewAI video is here. What should creatives actually use it for?
AI

Prompting for designers: think in briefs, not keywords

The skills that make a great creative brief are the same ones that get great results from image and text models.

Creative Pouch · · 1 min read

A pen resting on an open blank notebook
Photo: Mike Tinnion on Unsplash

Most bad AI output comes from bad briefs. Designers already know how to write good ones. They just don’t always notice that a prompt is the same thing in a smaller box.

Start with intent

A keyword list like “minimal, modern, blue, logo” gives a model almost nothing to work with. A brief does. It says who the work is for, what it should feel like and what it must avoid.

Compare these two:

  • “Minimal modern logo, blue.”
  • “A logo mark for a small Baltic bakery that sells rye bread. It should feel warm, handmade and a little old-fashioned. Use one colour, and avoid wheat icons and anything that looks like a tech startup.”

The second is longer, but every word narrows the space the model has to search.

Give constraints

Aspect ratio, palette, typography direction, what’s off-limits. Constraints aren’t limits on creativity. They’re what make the output usable. Models, like junior designers, fill any gap you leave with the most average answer they know.

Show, don’t only tell

Many tools now accept reference images. One good reference often says more than a paragraph. Use it to set mood, composition or texture, and say in words which part of the reference matters.

Iterate like a critique

Treat each result as a draft in a review. Name what works, name what doesn’t, and change one thing at a time. If you rewrite the whole prompt every round, you’ll never learn which change made the difference.

Keep a prompt library

When something works, save it with the result. Over time you’ll build a set of house briefs for recurring jobs like social graphics, moodboards and product shots. That’s your studio’s style, written down.

A prompt is a creative brief with a very literal reader.

  • #prompting
  • #generative-ai
  • #workflow

More in AI

AI · 2 min

Multimodal AI, explained for designers

Models that can see, read and generate across images, text and sound are changing how creative tools work. Here's what that means in practice.