I remember the first time I used ChatGPT and felt genuinely underwhelmed.
I typed “Write me a blog post about productivity” and what came back was the most generic, lifeless piece of content I had ever read. Every sentence was a cliche.
Every point was something I had seen in a hundred other articles. Nothing was specific, nothing was interesting, and nothing was actually useful to anyone. I closed the tab thinking that maybe AI was just overhyped and the people raving about it were either easily impressed or trying to sell something.
I was wrong. But not in the way I expected.
The problem was not ChatGPT. The problem was me. More specifically, the problem was my prompt. I had given the AI absolutely nothing to work with — no context, no direction, no specificity — and then been disappointed when it gave me the most average possible response to the most average possible instruction.
What I spent the following months learning, through a lot of trial and error and a lot of frustrating results that eventually turned into genuinely useful ones, is that AI tools are more like a mirror than a magic machine. The quality of what comes back is almost entirely a function of the quality of what you put in. And most people, including me for a long time, are putting in almost nothing.
The prompts I am sharing in this post are the ones I use every single day in my actual work. Not the ones I saw in a YouTube tutorial, not theoretical examples constructed to look impressive. The ones I refined over months of real use because they consistently produce output I can actually use without significant rework.
Some of them came from frustrating moments when a project was going badly and I needed something that worked. Others developed gradually as I figured out what made the difference between a mediocre AI response and a genuinely good one.
The Morning Planning Prompt
Every morning, before I do anything else, I use a version of this prompt. I have a cup of coffee, I open ChatGPT, and I paste in my task list for the day.
The prompt goes something like this: “Here are my tasks for today: [list]. My energy level right now is [high, medium, or low]. I have approximately [number] hours available to work. Some of these tasks are genuinely urgent, some are important but not time-sensitive, and some could wait until tomorrow without consequence. Please organize these into a realistic schedule for today, group tasks that require similar mental energy together, and tell me which task I should start with and your reasoning for that choice.”
The plan that comes back is consistently better than what I would have come up with on my own. The reason is not that the AI is smarter than me about my own work — it is that when I sit down to plan my day, I bring all kinds of emotional weight with me. I am attached to certain tasks, I am avoiding others, I am overestimating how long some things will take and underestimating others. The AI has none of that baggage. It looks at the list the way someone who has never met me would look at it, and the objectivity is genuinely useful.
This prompt has become such a consistent part of my mornings that on days when I skip it, I notice the difference. The mental clarity that used to take me half an hour to arrive at comes in about five minutes.
The “Make It Human” Prompt
Anyone who uses AI for content creation eventually runs into the same problem. The draft comes back technically correct, well-organized, and completely lifeless. It reads like it was written by someone who has read thousands of articles about the topic but has never actually experienced any of it. Which, of course, is exactly what happened.
When I get a draft like this, I do not start over from scratch. I use this prompt: “Read this content carefully. Identify every sentence or paragraph that sounds like it was written by an AI — overly formal, unnecessarily hedged, generic, or lacking any real personality. Rewrite those specific parts to sound like a real person who has actually lived through this subject is speaking. Keep all the factual information exactly the same. The goal is warm, direct, conversational language that a reader would actually enjoy. Do not add filler sentences or inflate the length.”
The result is almost always significantly better than the original. There is something almost funny about using AI to fix AI writing, but the approach works because the second pass has a clear, specific brief — find the flat parts and make them feel real — rather than the vague “write me something” instruction that produced the flat writing in the first place.
The Devil’s Advocate Prompt
This is the prompt I use most consistently before submitting anything important, and it has probably saved me from more professional embarrassments than any other habit I have developed.
Before I send a client proposal, publish a significant article, or pitch an idea to anyone, I use this: “I am about to [send this proposal / publish this post / pitch this idea]. Here it is: [paste content]. Respond to this as a thoughtful, skeptical person who is looking carefully for reasons to push back or say no. What are the weakest points in my argument or presentation? What important things am I not addressing? What assumptions am I making that a reasonable person might challenge? What objections would someone who disagrees with my conclusion raise?”
The feedback I get from this prompt is consistently more honest and more useful than feedback from almost anyone in my life, simply because the AI has no social investment in making me feel good about my work. It points at the weak spots without softening the blow, which is exactly what I need before something goes out the door.
The most striking example I can give is a client proposal I almost submitted without running this check. I was proud of it. I had spent real time on it. I thought it was tight and compelling. The devil’s advocate prompt identified three significant gaps in my logic within about thirty seconds. I revised the proposal. The client approved it on the first read.
The Research Accelerator Prompt
A significant portion of my work involves writing about topics where I am not an expert. Client projects cover industries I know nothing about. Blog posts sometimes require me to get up to speed on something quickly. Research used to be the most time-consuming part of any new project.
This prompt changed that: “I need to write about [topic] for [specific audience]. I am not an expert and I need to get up to speed quickly. Please give me: first, the five most important things someone needs to understand about this topic to write about it intelligently. Second, three common misconceptions that people have about this topic that I should be careful not to repeat.
Third, two or three specific, concrete examples that would make this topic feel real and relatable to a non-expert reader.
Fourth, any important nuances or counterarguments that a surface-level treatment of this topic would miss. Do not give me generic information — give me the specific things that would make my writing stand out from the basic articles already out there on this subject.”
The specificity of that last instruction — do not give me generic information — is what makes this prompt work. Without it, the AI defaults to producing the kind of high-level overview you could find in thirty seconds on Wikipedia. With it, you get something that actually speeds up the work.
The Tone Matching Prompt
This is the prompt that has had the single biggest impact on the quality of my client work, and I wish I had developed it much earlier.
When I write content for a client, the most important thing is not that the content is well-written in some abstract sense. It is that it sounds like it came from them — that it has their personality, their voice, their way of talking to their audience.
Generic well-written content that does not match the client’s voice is still a failure.
Here is the prompt: “Here are three examples of existing content written by or for [client name or business]: [paste examples]. Analyze the tone, vocabulary, sentence length and structure, level of formality, use of humor if any, and the overall personality that comes through this writing. Then write [new piece of content] that matches this exact voice. Do not imitate it superficially. Capture the underlying personality and apply it consistently throughout the new piece.”
The examples you provide matter enormously. The more representative they are of the client’s real voice, the better the match. I always try to find examples of the client’s own writing rather than content produced by someone else, because that is the voice the client actually has — not the voice someone else gave them.
The Stuck Writer Prompt
Every writer knows this feeling. You know what you want to say. You have a clear idea in your head. And somehow none of the words coming out are right. The opening is flat. The framing feels wrong. You delete and retype the same sentence four times and none of the versions are it.
This is the prompt I use in that exact situation: “I am trying to write [type of content] about [topic]. I know what I want to say — my main point is [describe it]. The feeling I want the reader to have after finishing is [describe the desired outcome or emotion]. I am stuck on how to begin.
Write me five completely different opening paragraphs, each taking a different approach. The five options should be: starting with a personal story or moment, starting with a surprising or counterintuitive fact, starting with a direct question to the reader, starting with a bold declarative statement, and starting with a specific scenario or scene the reader can imagine themselves in.”
Having five genuinely different options in front of me almost always breaks the block. Sometimes one of them is exactly right. Sometimes two of them suggest something better than either one alone. But the paralysis disappears because I am choosing between real options instead of staring at a blank page.
The Difficult Email Prompt
This prompt came out of a specific situation that I think most people who do any kind of professional work will recognize.
A client sent me a frustrated email. A deliverable had been late, he was not happy about it, and the email made that clear. I needed to respond in a way that acknowledged the situation honestly without being defensive or over-apologetic in a way that would actually make things worse. I needed to be professional without being cold, accountable without being groveling, and forward-looking without seeming like I was trying to rush past the problem.
Here is the prompt I built from that experience: “I received this email: [paste email]. The person writing it is [frustrated / disappointed / angry]. I want to respond in a way that: acknowledges their feelings sincerely without being defensive, takes appropriate responsibility without over-apologizing in a way that seems performative, clearly states what I will do to address the situation, maintains a warm and professional tone that does not feel stiff or scripted, and ends with a specific next step that moves things forward. Please write this response in under 150 words.”
That response saved the client relationship. He replied positively within an hour. I have used versions of this prompt for difficult professional communications ever since, and it consistently produces responses that handle the human side of the situation better than what I would write on my own when I am stressed or defensive.
The Content Repurposing Prompt
Writing a substantial piece of content takes real time and real effort. Getting the maximum possible value out of that time means not treating each piece as a single-use artifact.
Here is the prompt I use after publishing any significant piece: “Here is a blog post I wrote: [paste post]. Please repurpose this content into the following formats: five short posts for Twitter or X, each making one strong point from the article as a standalone idea; three LinkedIn posts that take a more professional tone and each build a short narrative around one insight from the piece; one email newsletter that summarizes the key takeaways in a personal, conversational tone as if writing to a friend who asked what the article was about; and one outline for a YouTube video that covers the same material in a format that works for spoken delivery rather than reading. Keep my voice consistent across all formats.”
One article becomes nine pieces of content. The total additional time this takes is about twenty minutes. The value of that output — distributed across different platforms and reaching people who would never read a long article — is significant.
The “Explain My Own Thinking” Prompt
This is one of the more unusual prompts on this list, and it took me a while to realize how valuable it was.
Sometimes I have an idea that feels important but I cannot quite articulate it. I know what I am trying to say somewhere in my head, but every time I try to write it down it comes out muddled or incomplete or not quite right. The idea is there — I can feel it — but the words are not cooperating.
Here is what I do: “I have an idea I am struggling to express clearly. Here is my rough, probably disorganized thinking about it: [write whatever you have, even if it is incomplete and messy]. Please help me figure out what I am actually trying to say.
Give me: a single clear sentence that captures my core idea, the strongest argument for why this idea actually matters, the most likely objection a reasonable person would raise, and a simple analogy that would help explain this to someone who had never thought about it before.”.
Using AI to understand my own thinking feels backwards at first. But it works consistently. The clarity I get from seeing my half-formed idea reflected back in organized form is genuinely useful, and it often reveals that the idea is either stronger or more flawed than I realized.
The Feedback Translation Prompt
Client feedback is often one of the most frustrating parts of creative work because so much of it is technically meaningless. Phrases like “it doesn’t feel right” or “I want it to pop more” or “can you make it a little more exciting” tell you almost nothing about what actually needs to change.
This prompt translates vague feedback into specific action: “My client gave me this feedback on my work: [paste feedback]. The work itself was: [describe or paste the work]. Please translate this vague feedback into specific, actionable changes. What are three different interpretations of what they might actually be asking for? For each interpretation, tell me exactly what I would change about the work to address it. Then tell me which interpretation you think is most likely based on the feedback they gave.”
The multiple interpretations are the key part of this prompt. When I go back to the client with clarifying questions that are informed by specific interpretations rather than just “I’m not sure what you mean,” the conversation is much more productive. Usually within one exchange I know exactly what they want, and the revision hits on the first try.
The End of Day Review Prompt
I added this one relatively recently and it has become one of the more valuable parts of my daily workflow.
At the end of each working day, before I close everything down, I use this: “Here is how my day went: I planned to [what you intended to do]. What actually happened was [honest account of what you did, including what went well and what did not]. Please help me with three things.
First, identify what was genuinely productive versus what just felt busy. Second, point out any patterns in how I worked today that might be quietly limiting my effectiveness. Third, suggest one specific thing I could do differently tomorrow that would make the day meaningfully better.”
The distinction between genuinely productive and just feeling busy is the most valuable part of this prompt for me personally. It is surprisingly easy to spend an entire day feeling like you are working hard while actually accomplishing very little, and equally easy to have a quiet, focused two-hour session that produces more than the entire busy day. This prompt helps me tell the difference.
What Makes These Prompts Actually Work
Looking at all twelve of these, there are a few things they have in common that I think are worth naming directly.
Every one of them gives the AI a specific role to play rather than just an instruction to follow. “Act as a skeptical reviewer” or “think like someone trying to say no” gives the AI a frame that shapes the entire response, not just the content.
Every one of them includes enough context that the AI understands the situation — who the audience is, what the goal is, what constraints exist. The AI cannot guess these things. Every detail you leave out is a place where the response defaults to something average.
Every one of them has a specific output structure. Not just “help me with this” but “give me five options” or “tell me three things” or “keep it under 150 words.” Specificity about what you want back is just as important as specificity about what you are putting in.
These are not tricks or hacks. They are just good communication — the same things that make a good brief, a good set of instructions, a good question. AI responds to good communication the same way thoughtful people do. The better you get at explaining what you actually need, the better everything that comes back will be.