4 prompting tricks that will take your AI from basic to brilliant

Laura Gagliardi is a UX strategist and AI comms expert in the intersection of tech, design and language. She is also an advisor to the Center for AI Strategy.

Prompting has gone from novelty to operational skill, and the stakes are even higher in Q4 as AI powers executive briefs, internal messaging and stakeholder comms. But often, AI language is generic and dull, dull, dull. How to fix it? By knowing how to direct AI with precision.

1. Start with what AI must understand, not what you want it to do

Most weak prompts jump straight to the task (“Write an email…”). The strongest prompts open with context-first framing because AI can’t prioritize what it doesn’t understand.

Structure to use every time: context → goal → format → constraints

Example:
“You are helping me prepare a vendor update email. The vendor is late on deliverables, and tone must stay firm but respectful. Write two versions in fewer than 120 words.”

Internal or external — this is the single biggest performance unlock.

2. Always define tone by audience, not adjective

“Friendly but professional” means nothing to a model. “C-level healthcare exec with zero time for fluff” means everything. AI responds best to audience mindset, not mood.

Instead of “Make it inspiring,” use: “Write this as if you’re speaking to a skeptical board that wants proof, not energy.” This is especially useful for crisis comms and executive comms.

3. Force comparison — don’t accept the first output

AI is a comparative thinker. Asking for variants coerces it to surface thinking you didn’t know to ask for.

“Give me three sharply different approaches: one conservative, one bold risk, one neutral middle ground. Label them.”

Perfect for strategy proposals, email subject lines and story angles.

This isn’t “more options.” It’s structured expansion, and it leads to faster alignment.

4. Add constraints that mirror real business pressures

AI performs best inside real-world limitations, especially business ones. Examples of high-performance constraint prompts are:

• “Rewrite under 100 words and assume reader is multitasking.”

• “Make it legally safe for healthcare, but still warm.”

• “Summarize as a Slack update for leadership, one scroll max.”

AI sharpens when pressure is explicit. Vague instructions lead to vague output. Q4 demands speed, clarity and risk awareness.

These prompt structures help AI behave like a strategic collaborator, not a content vending machine. Teams that apply them now will move faster and sound sharper before Q1 2026 begins. Good luck!

Learn more with the Center for AI Strategy.