After 5 Years of Writing Online, This Is How AI Helps Me Write Faster, Think Better, and Publish More

Learn how I use AI to speed up my writing, improve my thinking, and publish more.

Let me be honest, most people still use AI like a copy-paste machine.

They type one lazy prompt into ChatGPT, copy the answer, paste it into their blog, and then complain that “AI content does not work”.

I do the opposite.

For me, AI is not a single tool. It is a small team that works behind the scenes every single day:

• One AI tool that finds ideas.

• One that researches like a PhD student.

• One that challenges my thinking.

• One that cleans up my rough first drafts.

• One that turns words into infographics, carousels, and visuals.

• And one that helps me repurpose everything across platforms without losing my voice.

And in this post, I will walk you through the actual workflows that help me:

• Write faster without sounding like a robot

• Think better instead of blindly trusting AI

• Publish more often than most people while still keeping the quality high

Note: I have tried to make this post super informative and highly practical, so read it completely, bookmark it if you need to, share it with your friends, and take whatever fits your style.

And I honestly planned to put this post behind the paywall, but then I thought, why hide something that can genuinely help people? So I decided to keep it free for everyone. And if you enjoy this kind of no-nonsense content, I’m sure you’ll want to support me and eventually become a paid subscriber.

1. Turning my messy brain into an “always full” idea inbox

You know, I never sit down in front of a blank Substack editor and ask, “What should I write today?”

By that time, it’s already too late.

Instead, all day, I capture tiny sparks:

• While learning about AI or reading posts

• Comments readers leave on my work

• Interesting topics from books, tweets, or YouTube videos

• Content related to tools or workflows I’m testing

• Questions people DM me about AI, money, or writing

All of these go into a single “Idea Inbox” note in my Notion app.

The important part is what comes next: letting AI clean everything and help me find the best topic from all that chaos.

For that, once or twice a week, I paste a big chunk of these messy notes into ChatGPT (or Gemini) and ask:

You are my editorial assistant. Clean this list of random notes and group them into 10 - 15 potential article ideas.

For each idea, give:

• a working title

• 3 angles or hooks

• 3 problems the reader is struggling with

• who exactly this idea is for (beginner, advanced, creator, developer, etc).

This simple step does a few powerful things:

• Turns raw noise into clear topics

• Shows me what my audience actually cares about (because all ideas come from real conversations)

• Gives me hooks that are already close to “publish ready”

I never ask AI, “Give me viral ideas about AI”. I let my real life generate the raw ideas and let AI organize them.

Of course, I then add my own expertise to refine the idea further and make it worth writing about.

---

2. Using AI to choose the “right angle”, not just a good topic

You know, most writers pick topics and start writing whatever comes to their minds.

Let’s say my idea is: “How I use AI to write faster”.

Well, that topic is boring, and everyone else has already written content related to that, but I want to be more practical, pick better angles, and solve issues that most writers are struggling with.

And based on that, here is how I use AI to make it specific:

I paste 3–5 of my past posts that performed well, plus my rough idea, into ChatGPT and ask:

Study my writing style from these posts, and now take this idea: [idea].

Give me 10 angles that:

• feel controversial or surprising

• are practical, not motivational

• connect directly to money, results, or time saved

Format:

• Angle

• Why this angle will make someone stop scrolling

• One real example or story I can tell from my life or through the internet

You see, I am not asking AI to invent some fake story. I am asking it to surface angles I already have but cannot see clearly. From there, I just pick 1 or 2 angles that feel true, and I’m ready to outline.

And it gives me enough ideas about what I can write about, along with what I can include.

---

Just so you know:

Everything you just read is something I use every single day to run my business.

But a single tip can only take you so far. The real breakthrough happens when you stop using AI for “tasks” and start using it for “workflows”.

That shift is exactly what I packaged into “The (Unfair) AI Workflow Bundle”.

This is the complete blueprint of my day-to-day operations. It removes the guesswork and gives you the plug-and-play systems I use to:

• Cut execution time in half.

• Automate repetitive drudgery.

• Deliver work faster than what seems “normal”.

You can spend months figuring this out on your own, or you can steal my entire playbook right now.

---

3. My research stack: Perplexity + NotebookLM + ChatGPT

Let me be honest: most people still open 20 tabs and get overwhelmed by huge amounts of information.

That’s where I use a simple 3-layer stack.

Layer 1: Perplexity for fast, source-backed overview

When I have a clear angle, and know what I want to search, I go to Perplexity and ask very specific questions:

• What are the most common mistakes people make when using AI for writing?

• What does the research say about spaced repetition and learning?

• Compare Nano Banana with other visual AI tools for infographics.

You see, I am not looking for generic explanations.

I want edge cases, concrete numbers, studies or real examples I can link to, and more.

My prompts look like this:

Give me a concise breakdown with sources.

• Start with a 5-bullet ‘What most people get wrong’ section

• Then a short explanation

• Then links to original studies or articles

Keep it short and precise. No generic content please.

This way Perplexity become my “fast researcher” that keeps everything grounded.

---

Layer 2: NotebookLM for deep dives into my own files

When I really want to go deep, I open NotebookLM and create a notebook that includes:

• PDFs of research papers

• Info from Perplexity

• Screenshots of product pages

• My old posts on the same topic

Then I ask:

You are a researcher who only cares about what is inside these sources.

1. Pull out every quote, stat, or idea that would be interesting to an online creator.

2. Show me where my past thinking (from my old posts) was incomplete or wrong based on these new sources.

3. Suggest 5 ‘spicy’ takes I can use that are still honest and backed by the files.

This way NotebookLM makes sure my writing is not just “what AI thinks”, but what my own library says.

Further, I also create a podcast, a video overview, a mind map, and use more NotebookLM features to understand the topic clearly.

If you want to learn more about how to use NotebookLM in your workflow, you can read this post.

---

Layer 3: ChatGPT as my thinking partner

Once I have notes from Perplexity and NotebookLM, I paste them into ChatGPT or Gemini, and do something most people never do:

I ask it to fight me.

Here’s the prompt I use:

You are my harsh editor. Based on these notes, here is the argument I want to make: [write my rough argument in 5 - 10 bullet points].

Your job is to:

• List the 5 strongest objections a smart reader will have

• Show me where I am being vague, cliche, or emotional without proof

• Suggest what data, examples, or stories I should add to make this airtight.

This is how AI helps me think better, not just write faster.

---

4. Generating outlines that feel like me (not like AI)

To be honest, I usually don’t generate outlines for my posts because I can write better outlines myself than what AI gives me.

But sometimes, once my thinking is clear, I focus on creating an outline that feels like my style.

I rarely ask AI, “Give me an outline for X”.

Instead, I paste a high-performing article of mine and ask:

Create a structural blueprint of this article:

• Introduction pattern

• How you transition into the main idea

• How many sections, and in what order

• How you close

Make it abstract so I can reuse this as a template for future posts.

You see, now I have reusable “post skeletons” that already match my voice.

And when I want a new outline, I say:

Using the above blueprint, create an outline for this new angle or title: [add your angle or title].

Constraints:

• No generic content

• At least one personal story per major section

• At least one concrete prompt or example per section

• Keep the tone conversational like my earlier posts.

This gives me an outline where 70 percent of the work is done, but the thinking and personality are still mine.

---

5. Drafting: how I use AI without losing my voice

I know this is the part everyone is curious about, and many of you may be wondering if I let AI write the whole article. The answer is a big “NO”.

Here is my actual process.

Step 1: I write a “dirty” first pass

I put on a timer for 60 - 90 minutes and write the entire piece in ugly form:

• Broken sentences

• Repeated ideas

• Random thoughts that may or may not fit

• Notes to myself inside brackets like [add example of the headshot story here]

The goal is speed and honesty, not perfection. I just write whatever comes to my mind.

And since I’ve been writing online for over 5 years, I have the experience and the habit to write continuously for at least an hour.

---

Step 2: I ask AI to rearrange, not rewrite

Once I have that raw draft, I paste it into ChatGPT and say:

You are my structural editor. Do not rewrite in your own words, and keep my tone, my phrases, my stories.

Your job is to:

• Remove obvious repetition

• Suggest a better order for sections if needed

• Highlight any paragraph that feels confusing or weak

• Point out claims that need proof or examples.

Output:

• My original text with suggested headings

• Comments like an editor, not a new version

Then I manually apply the suggestions I like. Of course, I can also add or modify content based on my own experience.

You see, I am not fully dependent on AI. I just combine the AI workflows with my expertise to get the results I want in the shortest time possible.

---

Step 3: I use AI for micro-improvements, not to generate full paragraphs

When a paragraph feels off, I paste just that chunk and say:

• Improve the clarity and flow of this paragraph.

• Do not make it formal.

• Do not remove the emotional tone.

• Keep it sounding like a human complaining to a friend.

I never let AI overwrite my entire draft or rephrase it in different tone since it will make it worse. I just let it polish the rough edges.

---

6. Turning walls of text into visuals with different AI tools

You know, as a reader, I do not want to read content that is filled with only theory.

So I try to write content that provides real value about AI and shows how to use it practically inside your workflow. To add even more value, I have started including infographics.

If you have subscribed to my newsletter AI Made Simple, you may know that I was using Napkin AI to generate visuals earlier.

But now, I am using several AI tools that help me create visuals, infographics, and more in just a few seconds.

Here is my workflow:

a) Nano Banana for infographics and carousels

When I have a finished draft, I simply upload it to ChatGPT or Gemini and ask:

• Which part could be a one-page visual explanation?

• Which section could be a LinkedIn carousel?

• Which process could be shown as a flow chart?

Then I take that specific part, write it as bullets, and feed it to Nano Banana:

Turn this into a clean, modern infographic: [paste the content, steps or bullet points]

[with some more conditions….]

And to be honest, Nano Banana suggests layouts I would never think of.

Then I usually refine the text inside those layouts using Nano Banana Pro or by editing it in Canva.

These visuals end up in my Substack posts, LinkedIn carousels, and sometimes as standalone images on X.

If you want to learn how to use Nano Banana Pro, you can read this post where I shared 7 advanced, profession-specific use cases that will actually blow your mind.

---

b) Recraft to finish the actual design I want to create

You know that even after generating an image, you usually have to upload it into Canva to modify the background, add text, change colors, and so on.

That is where I started using Recraft. It lets me create visuals exactly the way I want and edit them without switching between multiple apps.

To be more specific, I use Recraft to:

• Generate and edit images without leaving the tab

• Create consistent custom styles for my brand

• Generate product mockups and logos in my own brand style

• Design T-shirts and sell them on Etsy

• Create book covers or digital product covers

I have even written a detailed post explaining how I use Recraft inside my workflow, along with several practical use cases for real work. Here is the post if you want to read it.

---

c) Leonardo AI or other image generators for hero images

If you have been reading my posts, you may know that I’ve tried a number of AI image generators and editors, and I am a big fan of Leonardo AI.

I have been using it for years because it generates realistic images and offers a useful free plan.

I mainly use Leonardo AI to create thumbnails or hero images for my posts.

For thumbnails or hero images, I write specific prompts tied to the article:

A realistic workspace where a human writer is sitting at a desk, laptop open with glowing AI interface, papers and sticky notes everywhere, late night, warm lighting, mood: focused and slightly chaotic, not sci-fi.

And here’s another prompt that I use most often:

A minimalist concept art depicting a lone real, beautiful, female programmer in her early 30s with gentle facial features and warm skin tone, using mac Laptop, sitting at a classic wooden desk in a cozy, sunlit room with soft, creamy walls and a large window that allows a flood of warm, golden light to illuminate the space, casting a cinematic glow on the writer’s focused face, warm colors dominate the palette with shades of beige, ochre, and sienna, ultra-detailed textures and high contrast lighting create a sense of depth and dimensionality, the overall mood is inspirational and contemplative, inviting the viewer to step into the serene and creative atmosphere of the writer’s sanctuary.

Based on the prompts I give it, here are the types of images it can generate:

And I mainly generate thumbnails to keep the image close to the real feeling of the article, not some random futuristic city shot.

---

7. Publishing more by automating the boring parts

Writing is hard. Hitting “publish” should not be.

That’s where, I use AI to handle the repetitive tasks around publishing so I can focus on the actual thinking and drafting.

But how? Well, here are the steps I follow:

Step 1: Headlines and subheads for every platform

Once my post is ready, I ask:

Based on this article, generate:

• 10 Substack-style titles (informal, honest, value-first)

• 10 YouTube-style titles (more curiosity, still honest)

• 5 email subject lines (short, punchy)

Rules:

• No clickbait

• No ‘ultimate guide’ cliches

• Use words my readers actually use, not marketing jargon.

What next? Well, I rarely copy them as is, but they give me strong starting points.

---

Step 2: One article, many formats (with smart prompts)

After publishing on Substack, I feed the article back into ChatGPT or Gemini and say:

For LinkedIn

Turn this article into a LinkedIn post in my writing style and tone.

Structure:

• 2 - 3 line hook

• 3 - 5 short paragraphs explaining 1 main idea from the article

• 3 bullet ‘What you can do today’

• Soft call to action to read the full post.

Keep my tone and personality from the original.

In a similar way, I generate content for Twitter, Substack Notes, and even email. I then proofread it once and make adjustments based on my expertise if needed.

Also, note that I do not treat these as separate new pieces. They are simply different doorways into the same idea.

---

8. How I use AI to keep my quality high while publishing often

I’ve learned that publishing more only matters if the quality stays at a high baseline.

Here is how I stop myself from getting lazy.

Step 1: The “brutal reviewer” workflow

Before I hit publish, I paste the full draft into ChatGPT and say:

You are a rude but honest editor who cares only about the reader.

Your job:

• Highlight any paragraph that is generic, vague, or could appear in any blog post on the internet

• Point out every cliché phrase

• Tell me where a busy reader will get bored and close the tab

• Suggest what I should cut completely.

Do not rewrite the article. Just give me a numbered list of harsh comments.

This hurts a little but it always makes the post better.

---

Step 2: Forcing myself to add proof, numbers, and more

I also ask:

List every claim in this article that would make a skeptical reader say ‘really’?

For each claim, suggest:

• What kind of example, story, infographic, or number I could add

• Where I can find that (data, experience, reference)

This is where I plug in:

• Personal income examples

• Subscriber growth screenshots

• Real screenshots of tools I use

• Actual time saved

Thanks to that, my posts feel real, valuable, and super practical, something readers actually want to read.

And the best part? They do not sound like an AI wrote them with vague advice or generic content.

---

9. My “daily AI writing routine” in 2-3 hours

To make this more concrete, here is what a typical writing day looks like when I am creating content.

1. Idea selection (10 minutes): I open my Idea Inbox, pick one or two topics, and use AI to refine the angle.

2. Research and thinking (20 to 30 minutes): I use Perplexity for quick facts, NotebookLM to review my notes, and ChatGPT to challenge and improve my arguments.

3. Drafting (40 to 60 minutes): I write the rough draft myself, then use AI to suggest structure improvements and help with small edits.

4. Visuals and repurposing (20 to 30 minutes): I use Nano Banana Pro to create infographics, and then create Thread and LinkedIn versions using the prompts I mentioned above.

Some days I cannot finish all the steps, and on most days I spend around four to five hours to write something valuable and practical.

Still, because the system is consistent, I can publish more often without burning out.

---

Let’s wrap up

If you look closely, you will notice something important:

• AI does not decide what I believe.

• AI does not invent my stories.

• AI does not replace the hours I spent learning, testing tools, failing, and figuring things out.

What it does is remove the friction by cleaning my messy ideas, checks my thinking, finds better angles, generate visuals, and handles the boring publishing parts.

That is how I can write faster, think better, and publish more than most people while still sleeping at night knowing the work is mine.

If you want to start using AI like this, do not begin with tools.

Begin with this question: Which part of my creative process feels heavy, slow, or confusing right now?

Then use AI into that exact spot, just like I did.

And to be precise, the unfair advantage is not that “I use AI”.

The unfair advantage is that I turned AI into a quiet team that works for me every single day, while most people are still arguing on the internet about whether it is cheating.

And if this post helped you, you already know what I am going to do next: turn it into a Nano Banana infographic, a LinkedIn post, a Twitter thread, and probably a YouTube script.

All from the same idea, and that is the real game.

---

Hope you like it.

That’s it—thanks.

Before you go, if you found this post valuable or learned something new, consider becoming a paid subscriber to this newsletter. It’s the best way to support my work and keep this kind of content coming.