Most people try AI at work once, get a mediocre result, and decide it doesn’t work. The problem isn’t the AI. It’s the prompt.
A vague prompt gets vague output. A structured prompt gets results you can actually use. The difference between the two takes about 10 minutes to learn — and once you know it, you’ll start using AI for real work every day.
Here’s how to write prompts that actually work.
Why Most Prompts Fail
The biggest mistake is asking AI to do everything at once. “Help me with my project” is not a prompt — it’s a wish. The AI has no context, no constraints, no sense of what success looks like. So it gives you something generic that sounds reasonable but doesn’t actually fit your situation.
Good prompts do three things: they give context, they define the output format, and they set constraints. That’s it. Once you understand that structure, you can apply it to any work task.
The Prompt Structure That Works
Think of a prompt as a short brief — the same kind of brief you’d write for a new team member. You wouldn’t hand someone a project with no background, no deadline, and no idea what done looks like. Don’t do that to your AI either.
1. Context — Who’s this for and what’s the situation?
Give the AI enough background to frame the response correctly.
Bad: “Write an email to my team about the project delay.”
Good: “I’m a project manager sending a status update to my 8-person team. We’re 3 days behind schedule on a client deliverable due Friday. The delay is due to unexpected scope from the client, not our team’s performance. Write an email that informs them of the new timeline without sounding like we’re making excuses.”
The second version tells the AI who it’s talking to, what the situation is, and what tone to strike. You’ll get something you can actually send.
2. Task — What exactly do you need?
Be specific about the action, not the topic.
Bad: “Help me with my presentation.”
Good: “Create an outline for a 10-minute presentation on Q2 budget performance for a leadership team that doesn’t work in finance. Focus on the three biggest wins and one area that needs additional budget approval.”
Specific tasks get specific results.
3. Format — How should the output look?
Tell the AI how you want the result structured.
Bad: “Write a summary of this meeting.”
Good: “Summarize this meeting in three bullets: (1) decisions made, (2) action items with owners and deadlines, (3) open questions that need resolution. Format it as a table.”
Format instructions are powerful. They turn a generic paragraph into something you can paste directly into your meeting notes or forward to a colleague.
4. Constraints — Any limits or requirements?
Set boundaries for tone, length, audience, or anything else that matters.
“Keep it under 300 words.” “Write for a non-technical audience.” “Use a confident, direct tone.” “Don’t use jargon.” These constraints shape the output in ways that make it actually usable.
Five AI Prompts You Can Use Today
Copy these frameworks, plug in your own details, and use them right now.
Email response:
“I’m a [your role] who needs to reply to an email from [description of sender and situation]. The sender is [emotional state]. Write a short, professional reply that [what you need the email to accomplish]. Keep it under 100 words. Tone: [formal/casual/ assertive].”
Meeting agenda: For more on running effective meetings, see [How to Write a Meeting Agenda That Actually Gets Results](/how-to-write-meeting-agenda/). Then use: “Create a meeting agenda for a [type] meeting with [number] attendees. The meeting goal is [specific outcome]. We have [duration]. Include: context setting, [number] discussion topics, and a decision or action item at the end. Flag any topics that might run long.”
Status update:
If you want a full weekly review system to go with your status updates, check out [How to Create a Weekly Review That Takes Under 30 Minutes](/how-to-create-weekly-review-30-minutes/). Also consider: “Write a weekly status update for a [role] Include: what was completed this week, what’s in progress right now, any blockers with what’s needed to unblock them, and what comes next. Keep it to 5 bullets. Format as: Done / In Progress / Blocked / Next.”
Feedback:
“Write feedback for a [role] who [what they did well] and [specific area for improvement]. The feedback should be direct but not harsh, specific enough to be actionable, and under 150 words. Include a concrete example of the behavior.”
Task breakdown:
“I’m a [role] who needs to [overall task]. Break this into 5-7 steps in priority order, with each step taking no more than 2 hours. Flag any steps that require input from others or external approval.”
How to Refine Prompt Output
The first output is rarely the final output. Think of it as a first draft, not a finished deliverable.
When you get something back, ask yourself: Is this specific enough? Does the tone fit? Is the format right? Then use follow-up prompts to refine it.
“Make this more direct.” “Add an example to the third bullet.” “Shorten this to 3 sentences.” “Rewrite this for a non-native English speaker.” These refinements take seconds and turn a rough draft into something polished.
The One Rule That Changes Everything
The first time you use AI at work, don’t try to replace a task. Try to accelerate one. Pick something small — an email draft, a meeting agenda, a status update — and use AI for that specific piece. When it works, you’ll see the value immediately. When it doesn’t, you’ll know exactly what to adjust.
The goal isn’t to automate your job. It’s to handle the 20 minutes of low-leverage writing you do every day so you can spend that time on the work that actually matters.
For more on building AI into your daily workflow, see The Complete Guide to Building Your First AI Workflow.
Use tested prompts for recurring workplace writing and planning tasks. The AI Productivity Prompts Pack includes 50+ tested prompts for meetings, emails, documents, reports, proposals, briefs, deep work, workflows, and administrative tasks. Get instant access — $12.