Most people use AI like a search engine with better answers. They ask a question, get a response, done. That is fine, but it is leaving most of the value on the table.
After months of using AI daily — for writing, planning, coding, and problem-solving — I have settled on a workflow that actually moves ideas from raw concept to finished output consistently. No magic. No overthinking. Just a repeatable system.
Here is what that looks like, step by step.
Step 1: Dump the Messy Version First
Before touching AI, I write a rough version myself. Not polished. Not complete. Just whatever is in my head at the moment — half-sentences, bullet points, contradictory thoughts, all of it.
This does two things. First, it forces you to actually think through the idea instead of offloading the thinking entirely. Second, it gives the AI something to work with that is rawer and more honest than what you would ask out loud.
When I skip this step, AI outputs tend to sound generic — like a textbook summary or a LinkedIn post from someone who just learned the topic. When I do it, the AI has real material to sharpen and structure. The rough draft is the signal. Without it, you are asking AI to pull meaning from nothing, and that is when you get the flat, bloodless output people complain about.
The rough draft does not need to make sense. It needs to exist. Open a doc, write for five minutes without stopping, and paste the mess into the AI. That is it. This single habit has done more for my output quality than any prompt engineering I have tried.
A practical way to do this: Set a timer for five minutes and write continuously about your topic. Do not edit, do not delete, do not try to be coherent. When the timer goes off, stop. Paste the result. That is your rough draft.
Step 2: Give AI a Specific Role and Constraint
Vague prompts get vague results. Instead of “help me write a blog post,” I say something like:
“Act as a direct response copywriter. Take the rough draft below and rewrite it for a reader who is skeptical but curious. Keep it under 600 words. Use short paragraphs. No fluff.”
Notice what is in there: role, audience, format, length, tone. The more specific you are, the less revision you will need later. I have wasted hours on AI outputs that were good but wrong because my prompt was too loose. Writing tighter prompts is a skill, and it pays off fast.
Here is a prompt template I return to often:
Role: [who AI should act as]
Task: [what to do with the rough draft]
Audience: [who this is for]
Format/length: [e.g. 400 words, bullet points, email]
Tone: [e.g. conversational, authoritative, casual]
Constraint: [what to avoid, e.g. no jargon, no fluff]You do not need all five elements every time, but the more you include, the more useful the output. If I am writing for my own brand, I always specify tone — otherwise AI defaults to something that sounds like it was written by a content marketing department.
Common constraints to add:
- Word count range (helps prevent both too short and too verbose)
- Audience expertise level (a beginner needs different language than an expert)
- What to prioritize (e.g. “focus on the emotional outcome, not the features”)
- What to avoid (e.g. “no cliches, no buzzwords, no passive voice”)
Step 3: Review and Iterate, Do Not Just Accept
The first AI output is never the final one. I read it like a draft from a junior writer — useful, sometimes solid, but rarely perfect. I look for three things:
- Does it actually say what I meant, or did it smooth over the rough edges I needed? AI often “fills in the gaps” in ways that change your meaning. Read for accuracy, not just polish.
- Is the tone right for the audience? AI tone can drift toward formal or generic. If your audience expects casual, catch where it got stiff.
- Are there spots where I can add something personal that AI cannot replicate? A specific story, a contrarian take, a phrase that only you would say. Those are your fingerprints. Do not let AI erase them.
Then I iterate. I will ask for a second draft with specific adjustments, or I will paste the output back with a note like “make this half as long and twice as direct.” Each round narrows the gap between draft and done.
The iteration pattern I use most:
- First pass: broad revision (structure, tone, audience fit)
- Second pass: tightening (cut by 20–30%, make every sentence earn its place)
- Third pass: personal touches (add voice, specific examples, your own phrasing)
Most things need two rounds of iteration. Complex pieces — like an essay or a business document — might need all three. You do not need to iterate every piece to death. Know your quality bar and stop when you hit it.
Step 4: Ship With a Simple Quality Check
Before anything goes out, I run a quick check:
- Does it sound like me?
- Would I be okay if my name was on this?
- Is the core message clear in the first 30 seconds?
If the answer is no to any of those, I fix it — even if that means rewriting a section manually. AI assists, but you own the output. The moment you publish something that does not sound like you, you have broken trust with your reader, even if they cannot name what is wrong.
I also keep a running document of prompts that worked well. Over time, your best prompts become reusable templates. That is where the real leverage is. Instead of reinventing the wheel each time, you build a personal library of prompts tailored to your specific needs. Add to it whenever you write something you are happy with and remember what prompt got you there.
The prompt document habit: After finishing any piece, paste the prompt at the top of the document and save it in a folder called “Prompt Templates.” Within a few weeks, you will have a collection you can remix and adapt rather than building from scratch every time.
A Real Example
Last month I needed to write a product description for a digital download. I had a vague idea of the core benefit but nothing structured.
- I wrote two paragraphs of messy, rambling thoughts about who the product was for and why it mattered — contradictions and all.
- I pasted that into ChatGPT with the prompt: “Turn this into a product description for [specific audience]. 150 words max. Focus on the outcome, not the features. Conversational tone, like you are explaining to a friend.”
- The first draft was okay but generic. It described the product well but sounded like everyone else is product description. I asked for a revision: “Make it more conversational. Cut the adjectives. One of my readers said this felt sales-pitchy — fix that.”
- I added one line at the end that captured the real reason someone would buy it — something only I could write, drawn from my own experience.
- Published.
Total time: under 20 minutes. Without the workflow, I would have spent an hour staring at a blank page, second-guessing the wording, and eventually settling for something mediocre just to get it done.
Why This Works Better Than Asking AI Directly
The core difference between this workflow and “ask AI directly” is where the thinking happens. Most people treat AI as the thinker. This workflow keeps you as the thinker and uses AI as an amplifier for your ideas, not a replacement for them.
When you dump your rough draft first, you are doing the creative work of identifying what you actually want to say. When you direct AI with specific constraints, you are using your judgment to shape the output. When you iterate and review, you are applying your taste and standards. AI is a very fast, very capable assistant — but it is still an assistant. You stay in the driver seat.
The other reason this works is that it removes the biggest friction with AI: blank-page paralysis. You always have something to work with. The rough draft is not the final version, so there is no pressure to make it good. You just need to get ideas out of your head and into the document.
Start Small
You do not need to restructure your entire workflow today. Pick one piece of writing you are working on this week and try the rough-draft-first approach. Write five minutes of mess, paste it in, and see what comes out. Adjust the prompt. Iterate once. Notice the difference.
That is how this becomes habit — not a productivity hack you try once and forget, but a genuine shift in how you work with AI. It is not about more AI. It is about smarter collaboration.
That is it. Four steps. Not glamorous. But it works every day.
Ready to build more repeatable AI workflows? 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.