This AI prompt cheat sheet solves 4 of work’s big everyday problems

You’ve heard the gospel: AI is going to change everything. Good, great, grand.

But when you’re staring down a deadline and 80 unread emails, you don’t need philosophy, you need a cheat sheet.

The fastest way to master AI isn’t by watching lectures, it’s by finding a way to replace an hour of your grind with a 10-second prompt. Here are five specific, repeatable ways to automate your most time-consuming professional tasks.

Grab your chatbot of choice (Gemini, ChatGPT, Claude, Copilot—whatever floats your boat) and let’s get to work.

Writing

Staring at a blank page. Tedious, formulaic first drafts. Enough.

You are a professional. You shouldn’t be spending an hour drafting a boilerplate email to a client or writing the first three paragraphs of a report. That’s grunt work.

Instead, master “constraint-based prompting.” This is where you tell the AI exactly what to write and how to write it, forcing it to follow your specific, professional rules.

Here’s a prompt example:

“You are a [job title]. Draft a [document type (memo, email, etc.)] to [target audience]. The tone must be [tone]. The three key takeaways are [list three specific bullet points]. The final memo should be around [length in words] and include a subject line.”

Post-meeting action items

Sifting through long transcripts and meeting notes for action items?

You’re doing it wrong. Let the AI do the heavy lifting of synthesis.

It’s time to leverage “deliverable-based prompting.” Instead of asking for a summary, ask the AI to produce specific, structured outputs from a large body of text, such as a meeting transcript or a dense PDF.

For example:

“Analyze the following [meeting transcript/document]. Do not summarize the entire text. Instead, produce three distinct outputs: 1) A table listing all action items, the person responsible, and the deadline mentioned. 2) A list of three open questions that were not resolved. 3) A short, two-sentence email subject line for the follow-up.”

In less than a minute, you can transform raw data into a clean, actionable task list.

Research

To turn generative AI into a true, trusted research assistant that can search and cross-reference information scattered across multiple work files requires using tools that let you upload your own content, such as Google’s NotebookLM or similar features in other platforms.

This is called “contextual grounding,” and it involves uploading a handful of annual reports, project documents, or extensive research files. Check with your organization first to see if there are any rules against this.

Here’s a prompt you can use:

“Based only on the uploaded documents, what is the biggest discrepancy between the Q4 2024 revenue projection [from Document A] and the actual Q1 2025 marketing spend [from Document C]? Explain the gap in three bullet points, referencing the specific document where the information was found.”

This lets you stop relying on the AI’s general knowledge and start using it as a hyperefficient analyst for your own private data, generating insights that would take hours to gin up on your own.

Brainstorming

Thanks to AI, hitting a creative wall or falling victim to groupthink during brainstorming is nothing like it used to be.

While your brain thinks linearly, AI can think exponentially—but you have to force it to show its work.

Employ “critical reasoning prompting,” also called “chain-of-thought.” This forces the AI to debate, critique, and explore alternatives before settling on an answer.

A sample prompt formula:

“I have an idea for a new product feature: [describe the feature]. Before you propose a name for it, I need you to first: 1) Act as a skeptical customer and list three reasons why this feature is useless. 2) Act as a competitor and list three ways they could easily copy and neutralize the feature. 3) Only after those two steps, propose three distinct, benefit-driven names for the feature.”

This forces the AI to act as a constructive adversary, getting you to a better, more robust idea much faster.

View the full article