Designers already know this one
“Make it pop” is a prompt. The brief is the context. Anyone who has received the first without the second knows what comes back: a confident guess.
AI works the same way. The prompt points. The context decides what there is to point at. Designers: get the free prompt kit that starts from the brief.
Prompt: “Write a follow-up email to my client.”
Without context: a polite template that could go to anyone.
With context: the session notes, two emails you wrote yourself, the resource you promised and a 120-word limit.
Same prompt. Now the draft sounds like you and remembers the promise.
Can a better prompt fix missing context?
Rarely. The failure that changed how I build came from a perfectly good prompt.
My daily review reads a board of all 13 of my social accounts. One morning that board was 25 hours old.
So the review reached for an older source that only knew about two accounts, and wrote a clean, confident report about two accounts. I read it and believed it.
Same prompt as every other day. Wrong context.
My rule since then, in my words at the time: “I don’t want fall backs on ANYTHING.” If a source is stale, the report says so and leaves the number out.
Where should instructions you reuse live?
Not in your head, and not retyped every time.
Anything you’d paste twice belongs in the context: project instructions, a saved file, a skill. Keep the prompt for what’s new today.
The first prompt line I teach is really about context: “Before you answer, ask me 3 questions.” It makes the AI collect what it’s missing before it guesses.
One trap: the words you didn’t want
A model notices every word you give it, including the ones you’re trying to ban.
On six of my app’s TikTok accounts, the writing instructions said, in brackets, “NO link en la bio”.
The tool printed it on the last slide. The warning became the content. Say what you want instead, like “end with a question”.
The wider practice of choosing what the model sees is context engineering. How much it can see at once is its context window.
AI for Designers is built on this difference: a clearer brief in, a direction you can explain out.