DEV Community

Tosh
Tosh

Posted on

The 12 ChatGPT Prompts I Use Every Day to Save 3 Hours (Copy-Paste Ready)

#ai #beginners #productivity #chatgpt

The 12 ChatGPT Prompts I Use Every Day to Save 3 Hours (Copy-Paste Ready)

Most ChatGPT prompt lists are generic. These are the ones I actually use.

I've been building digital products and running a content business with AI for over a year. In that time, I've tested hundreds of prompts and thrown out most of them. What's left is a tight set of prompts I reach for daily — the ones that actually save time instead of just feeling productive.

Here are the 12. Organized by use case. Ready to copy.

✍️ Writing

1. First Draft Accelerator

Write a first draft of [article/email/post] about [topic]. Target audience: [describe them]. Tone: [casual/professional/direct]. Length: approximately [word count]. Do not write a conclusion — I'll write that myself. Focus on making the opening paragraph compelling enough that someone who's skimming will stop and read.
Enter fullscreen mode Exit fullscreen mode

The opening paragraph instruction is the key. Without it, ChatGPT writes a generic intro every time. Telling it to skip the conclusion forces it to front-load the value.

2. Edit for Clarity

Edit the following text for clarity and concision. Remove filler words, passive voice, and any sentences that don't add information. Keep my voice and don't change the meaning. Return the edited version only, no commentary. Text: [paste your text]
Enter fullscreen mode Exit fullscreen mode

I run everything through this before publishing. It's caught embarrassing redundancies more times than I can count.

3. Headline Generator

Generate 10 headline variations for this article: [paste article or summary]. Include: 2 numbered list headlines, 2 curiosity-gap headlines, 2 direct benefit headlines, 2 "I did X" story headlines, and 2 question headlines. Make them specific — avoid vague words like "amazing" or "powerful."
Enter fullscreen mode Exit fullscreen mode

The format breakdown matters. Without it, you get 10 variations on the same headline style.

💻 Coding

4. Code Review + Bug Hunt

Review the following [language] code. Identify: (1) bugs or potential errors, (2) performance issues, (3) anything that could break at scale, (4) anything that violates common conventions for this language. Be specific — reference line numbers or function names. Code: [paste code]
Enter fullscreen mode Exit fullscreen mode

Generic "review my code" prompts give generic feedback. This structure forces specificity.

5. Write a Function from Spec

Write a [language] function that [describe exactly what it should do]. Requirements: [list edge cases, input formats, expected output]. Include: inline comments explaining non-obvious logic, error handling for [specific cases], and a usage example at the bottom.
Enter fullscreen mode Exit fullscreen mode

The edge cases and usage example requirements are what separate good output from great output.

🔍 Research

6. Summarize and Extract

Summarize the following [article/report/text] in 5 bullet points. After the summary, list: (1) the 3 most important claims made, (2) any statistics or data cited, (3) anything the author assumes without evidence. Text: [paste content]
Enter fullscreen mode Exit fullscreen mode

The "assumes without evidence" section is underrated — it saves you from citing weak claims in your own work.

7. Competitive Analysis

I'm researching [topic/product/industry]. List the top 5 alternatives to [thing I'm looking at] and for each one: describe what it does differently, who it's best for, and one weakness. Be specific — avoid marketing language.
Enter fullscreen mode Exit fullscreen mode

The "avoid marketing language" instruction is what makes this one actually useful instead of just restating product descriptions.

📧 Email

8. Reply to a Difficult Email

I need to respond to this email professionally. My goal: [state your goal — decline, negotiate, delay, clarify]. Constraints: keep it under 150 words, stay polite but direct, don't apologize unless necessary. Here's the email I received: [paste email]
Enter fullscreen mode Exit fullscreen mode

"Don't apologize unless necessary" is the most important line. ChatGPT defaults to over-apologetic email tone without it.

9. Cold Outreach

Write a cold outreach email to [type of person] about [what I'm offering or asking]. The email should: open with a specific observation about their work (I'll fill this in — use [PERSONALIZATION] as placeholder), explain the value in one sentence, and end with a low-friction ask. Under 120 words. No subject line needed.
Enter fullscreen mode Exit fullscreen mode

The low-friction ask instruction is crucial — it pushes ChatGPT toward "can I send you more info?" instead of "schedule a call with me."

📋 Planning

10. Project Breakdown

Break down this project into tasks: [describe project]. For each task: estimate time required, list dependencies (what must be done first), and flag any decision points where I'll need to make a choice before continuing. Format as a numbered list I can paste into a task manager.
Enter fullscreen mode Exit fullscreen mode

The decision points instruction is what I didn't know I needed until I started using it. It surfaces blockers before you hit them.

11. Weekly Priorities

I have the following tasks on my plate this week: [list everything]. My key constraint is [time/energy/money]. Rank these by impact-to-effort ratio and tell me which 3 I should prioritize first and why. Flag anything I should delegate or delete entirely.
Enter fullscreen mode Exit fullscreen mode

The "delete entirely" option gives ChatGPT permission to challenge your assumptions, which is the whole point.

12. Post-Mortem Analysis

I just finished [project/launch/campaign]. Results: [brief description of what happened]. Help me analyze: what went well and why, what underperformed and likely causes, and 3 specific things to do differently next time. Be direct — I don't need encouragement, I need useful observations.
Enter fullscreen mode Exit fullscreen mode

"I don't need encouragement" sounds harsh but it's essential. Without it, you get a lot of "great job on X" padding before the actual insight.

The Pattern Behind All of These

Every prompt above shares a structure: it tells ChatGPT the task, the constraints, and what not to do. Most people write the first part and skip the last two. That's why they get generic output.

Constraints are the secret lever. "Under 150 words," "no marketing language," "don't write a conclusion," "be direct" — these instructions filter out the filler that makes AI output feel fake.

I keep a reference pack of 500+ of these organized by use case — marketing, development, research, client communication, product building. It's what I reach for when I need a prompt fast and don't want to engineer one from scratch: https://toshleonard.gumroad.com/l/rzenot

Start With Two

Don't try to implement all 12 at once. Pick the writing prompt and the email reply prompt. Use them for a week. Notice how much faster things move when the first draft is already 70% of the way there.

The prompts compound. The more you use them, the more you'll modify them for your specific context — and that's when they get really powerful.

The full collection (500+ prompts, 20+ categories) is at https://toshleonard.gumroad.com/l/rzenot.

Top comments (0)