Learn how to write effective AI prompts for work. Boost productivity with clear, reusable prompts for meetings, reports, emails, and more.
.png)
The rise of AI tools like ChatGPT, Notion AI, Jasper, and Claude has opened up a new paradigm of work—one where professionals no longer just use software, but converse with it. Whether you're drafting a report, summarizing a call, writing cold emails, or debugging code, prompt-writing is the new literacy.
But here’s the problem: most professionals are undertrained in the skill of writing effective prompts. They expect AI to "just get it" — and when it doesn’t, they assume the tool is flawed.
In reality, the power of AI at work comes down to how well you prompt it.
This guide is for knowledge workers, team leads, operators, marketers, recruiters, and anyone who uses AI to get real work done. We'll walk you through the fundamentals, frameworks, examples, mistakes to avoid, and advanced strategies for writing prompts that actually work.
Prompting isn’t just for AI nerds. It’s becoming a critical workplace skill — just like Excel once was.
Here’s what great prompting does:
Whether you’re in sales, HR, legal, operations, or marketing — your results are only as good as your instructions.
To write great prompts, use the CLEAR framework:
Give background. What is the goal? Who is the audience? What role should the AI play?
Bad: “Write a summary.”
Good: “You are a product manager preparing a 1-slide summary of a customer interview for your team.”
Be explicit about tone and format. Should it be professional? Concise? Bullet-pointed? Technical?
“Write in a confident but casual tone, using bullet points where possible.”
Provide examples or previous inputs/outputs when possible.
“Here’s a similar email we sent last month. Use this style and structure.”
Tell the AI exactly what to do. One task per prompt is best.
“Summarize the transcript into three key themes with supporting quotes.”
Add any constraints or specific requirements.
“Do not use the word ‘effortlessly’. Keep it under 100 words. Start with a benefit.”
Here are battle-tested prompts for common work tasks:
Prompt:
“You are a research analyst. Summarize the top 5 trends in B2B SaaS pricing models in 2024. Include 1-2 sentence summaries, company examples, and links to sources. Write in plain English for an executive audience.”
Prompt:
“You are a B2B sales rep targeting mid-sized HR software companies. Write a first-touch cold email offering a demo of our onboarding automation tool. Keep it under 100 words. Add a soft CTA.”
Prompt:
“You’re an operations lead summarizing a 45-minute team meeting. Extract: 1) Key decisions, 2) Action items with owners, 3) Open questions. Write in bullet points.”
Prompt:
“Analyze this Q1 financial report and summarize: 1) Revenue performance vs target, 2) Notable trends or red flags, 3) Any risks highlighted. Write in memo format for an internal finance team.”
Prompt:
“Rewrite this 800-word draft for clarity and conciseness. Use plain English. Break up long sentences. Keep formatting intact. Highlight any missing info with comments.”
Even smart people mess this up. Avoid these:
“Help me write a report.”
AI needs direction. Specify the type, goal, audience, and tone.
“Summarize, rewrite, give feedback, and generate slides.”
Break into smaller prompts.
An internal memo is not the same as a LinkedIn post. Be explicit about who it’s for.
You don’t need to “trick” the AI — just be clear. The magic is in clarity, not cleverness.
Ready to level up? Use these.
Split big tasks into a sequence of smaller steps. For example:
Great for workflows like:
Modularize your prompts for repeat use.
Example:
Tone Block: “Write in the tone of a calm, confident product manager at a Series A startup.”
Structure Block: “Start with a benefit. Then list 3 bullet points. End with a question.”
Combine like LEGO.
Assign the AI a role before asking the task.
“You are a senior recruiter with 10 years of experience. Review this interview transcript and assess the candidate's cultural fit.”
This activates domain-specific logic.
Always specify what you want back.
“List 3 key trends in markdown format:
If you use a tool like Notion AI, Glyph, Zapier, or ChatGPT Team, you can save and reuse prompts in specific workflows. This makes it easy for your team to:
Example:
In a recruitment workflow → add a block:
Prompt: “Extract key strengths, weaknesses, and next-step recommendation from this interview transcript. Use bullet format.”
The beauty of prompting is that it’s accessible. You don’t need to know how to code — just how to communicate clearly.
But like coding, it pays to:
Think of prompting as the new spreadsheet formula — invisible, powerful, and misunderstood until mastered.
Companies are already hiring Prompt Engineers, but soon they’ll hire Prompt Ops — teams who curate internal libraries of reusable prompts for sales, support, ops, and more.
Tools like:
Prompting is not a trend — it’s a foundational skill in the age of AI at work.
Whether you’re automating summaries, improving communication, or scaling yourself with AI, the way you write your prompt determines what you get back.
Start small. Use CLEAR. Build prompt blocks. Share with your team.
The better your prompt, the more AI feels like a teammate — not just a tool.
Nope. Prompting is more like writing a good email than writing Python. Clear thinking > technical skill.
Yes — and you should. Create prompt templates for repeated tasks and document them for your team.
ChatGPT, Notion AI, Glyph, or any tool with a custom prompt field. The tool matters less than how clearly you define the task.
As long as needed for clarity. Some are one sentence. Some are 5–10 lines with context, examples, and format requests.
No. Prompting is becoming part of the modern workplace toolkit — just like search, spreadsheets, or email.
Daniel is the founder of Glyph, a workplace AI platform that helps teams automate internal workflows and search knowledge instantly using prompts. Glyph lets you build custom assistants trained on your own data — no code required.
.svg)
Use GenAI to analyze, draft, summarize, and create—powered by both your company’s internal knowledge and external world data.
.png)