Have you ever typed a quick request into Gemini or another AI tool, got back something generic, and then spent the next 15 minutes going back and forth trying to fix it? You're not alone.
Most of the time, the problem isn't the AI. It's a lack of context. AI can only work with what you give it. A one-line request gets a one-size-fits-all answer.
The fix is to structure your request up front. Here's the four-part framework I use and teach for writing AI prompts, including Gemini prompts.
The 4-part framework for better Gemini prompts
1. Persona
Tell AI who you want it to be. A travel planner, a marketing strategist, a web developer, an HR manager. The persona shapes the tone, vocabulary and level of expertise in the answer.
2. Context
Give it the background on your project. Who is this for? Why does it matter? What does success look like for you? This is the part most people skip, and it's the part that makes the biggest difference.
3. Task
Be specific about exactly what you want done. "Write a research report on…" or "Create a Google Sheets formula that…" is far better than "Help me with…"
4. Format
Tell it how you want the answer delivered. A bulleted list, a 200-character LinkedIn update, a 500-word blog post, a table. If you don't say, AI will guess, and it often guesses wrong.
Putting it together: a Gemini prompt example
Here's an example you could paste into Gemini in Google Workspace:
You are an experienced HR manager. We're a 60-person company moving our team to a hybrid schedule next month, and staff are worried about how meetings will work. Write an announcement email explaining the new schedule and meeting guidelines. Keep it under 250 words, friendly in tone, with three short bullet points.
Persona, context, task, format. Four elements, one prompt, and a far better first draft than "write an email about hybrid work."
Make it a habit
The first few times, the framework can feel like extra effort. It isn't. A minute spent structuring your request saves you the 15-minute back-and-forth that follows a vague one. Once your team gets used to it, it becomes second nature. Clear prompts are also the first step in letting AI do the heavy lifting on repetitive work.
Quick recap
- Persona: who should AI be?
- Context: who is it for, why does it matter, and what does success look like?
- Task: exactly what do you want done?
- Format: how should the answer look?
Save the framework somewhere your team can see it, and try it on your next request. If people are pasting prompts into unapproved tools, read our guide to taming shadow AI.
If you'd like your team to get more out of Gemini, our Google Workspace training covers practical prompting like this, using the tools your people already work in.
