Infinite Prompting: Get AI To Think 60x Longer And 5x Better With One Prompt
Why The Future Of AI Isn't Better Prompts—It's Longer Thinking
I recently discovered something that fundamentally changed how I create content and how I use AI more generally.
The breakthrough started with a problem.
I'd just finished teaching a 90-minute class on prompt leadership, and I wanted to turn the transcript into an article. Normally, this means days of organizing, writing, and rewriting. The kind of deep work that sounds romantic until you're actually doing it at 11 PM on a Tuesday. So, you don’t do it at all.
This time, I tried something different.
A few hours later, I had published a high-quality article I was proud of—The Prompt Leadership Revolution: Why The Next Generation Of Thought Leaders Will Create Tools, Not Content.
Here’s what I did,
I gave AI the class transcript and asked it to:
Generate 15 different artifacts analyzing the content from every possible angle.
Create an article based on the insights from all of the artifacts.
Then…
I let Claude think.
Not respond.
Think for 20 minutes.
20 minutes!
Thoughts I didn’t tell it to think.
Ones it decided to.
When it finally started producing the article, what emerged wasn't just a summary. It was something I hadn't expected: a comprehensive exploration that found patterns I'd missed, made connections I hadn't seen, and organized ideas in ways that made more sense than my original presentation.
Three hours of my time editing instead of three days. Not a draft. A finished piece that was genuinely better than what I would have produced alone.
I call this Infinite Prompting—giving AI the time and space AI to think longer and better.
And, for me, as a teacher, this application of Infinite Prompting is profound. I teach three 90-minute classes per week for my AI Thought Leader School, and Infinite Prompting will enable me to turn each of my 120 classes per year into 120 articles that are read by tens of thousands of people each. I estimate that this one application of Infinite Prompting alone will:
Lead to an additional 400,000 published words per year
Generate an extra $100,000 per year for me
Allow me to reach millions more people
But this ability to turn a rough transcript into a solid article is just the tip of the iceberg.
At a more fundamental level, Infinite Prompting points to something much more profound for all of us:
Giving AI the time and space to actually think dramatically improves almost all of its outputs.
It pushes the results from helpful to life-changing.
It does this for many applications—whether you’re turning meetings into reports, ideas into strategies, or research into insights.
With Claude, you can now get AI to think for 20 minutes rather than just 20 seconds.
In this article, I will explain what Infinite Prompting is and provide you with two infinite prompts you can use right now to see the results for yourself. If you test out these prompts, you will see a 50% improvement in the results you get from AI forever.
It’s a bold claim, but it has been true for me and everyone I’ve shown it to so far.
But, before I jump in, it’s critical that you get context on why this is such a big deal and not hype…
The Research-Backed Bitter Lesson About AI That Nobody Is Talking About
In 2019, an AI researcher named Richard Sutton wrote a seminal blog post that most people outside of AI have never even heard of. He called it The Bitter Lesson.
His argument was simple: Throughout AI history, researchers have kept trying to build human knowledge into AI model thinking. While this approach temporarily improves performance, it has always been outperformed by approaches that just used two things:
Better learning
Better search
Chess programs that encoded grandmaster strategies? They lost to programs that just searched more positions.
Speech recognition with linguistic rules? Beaten by systems that just processed more data.
Every time humans tried to be clever, brute computational force won out over time.
Sutton called it "bitter" because it meant decades of clever human work kept losing to ‘dumb’ scaling. But here's what I’ve never heard anyone talk about: this bitter lesson applies to how we prompt AI, not just how we build it.
Most people are doing the same thing with prompts that those researchers did with chess programs. They’re trying to design the perfect prompts. While this works in the short term, the scalable, long-term approach is giving AI more time to think and creating the conditions for it to think better.
The prompt engineering industry is teaching you to be a better chess programmer in a world that rewards computation. It's just teaching you to write better instructions when you should also be creating better thinking conditions. The wrong variable is being optimized.
Try This Infinite Prompt To See It Work For Yourself
Most people treat AI like a smart assistant. Ask question, get answer. Send prompt, receive output. It's transactional. It’s one step.
But what if we treated AI more like a research team?
When you commission real research, you don't expect instant answers. You expect exploration. Multiple angles. Dead ends and breakthroughs. Iteration and discovery. Time.
This is the core insight behind Infinite Prompting method. Instead of asking AI for one perfect response, you ask it to explore a topic through 15 or more interconnected artifacts. Each artifact builds on the previous ones, creating momentum like falling dominoes…
Now, here's what a simplified version of the infinite prompt that generated this post in Claude actually looks like:
Attached is [content/transcript/idea].
My goal is to create a blockbuster paradigm shifting article with a stress-tested idea that will stand the test of time and are so valuable and well-explained that it becomes the dominant prompting paradigm.
Keep in mind that I'm targeting [audience].
Match the style of my writing on my newsletter ([URL to archive of your content]) so it is authentic to me.
Do no make up any hypothetical stories.
To most effectively accomplish the goal of the article, generate 15 artifacts. Then output the final updated article as a final artifact.
The goal of each artifact should be to generate synthetic data that would help you accomplish the goal better.
Guidelines:
- Have the artifacts in a sequence that builds on each other and generates momentum
- Use multiple reasoning methods, paradigms, frameworks, and/or perspectives
- Think with deep rigor and attention to detail
First, it planned everything. Notice how it’s surfing the Internet and following multiple steps.
Then the artifacts started appearing. Not random outputs—connected explorations that reference and build on each other. An analysis leads to a framework, which leads to a case study, which leads to a deeper insight.
It's like watching a research team work in fast-forward.
Important Caveat: If you have a free plan with Claude, you likely won't be able to produce as many artifacts at one time, and you will hit your usage limit fast. Because of how good Infinite Prompting is, I upgraded to the Claude Max plan, which gives you 20x the usage as the basic $20/month plan. I have never hit a usage limit, and I create hundreds of artifacts a day.
The Synthetic Data Gold Mine
Here's where things get interesting.
Every artifact the AI produces becomes "synthetic data"—AI-generated content that serves as high-quality input for future AI interactions. This isn't throwaway output. It's intellectual capital.
Think about what this means. Traditional prompting creates disposable outputs. You generate something, use it, and move on. But with the Infinite Prompting Method, every output becomes an asset that makes future outputs better.
I now have a library of AI-generated analyses, frameworks, and insights that I can reference in future prompts. Each new project builds on the synthetic data from previous ones. It compounds.
Your prompt library? That's linear value. A synthetic data library combined with a prompt library? That's exponential. I write more about creating your own synthetic data in putting it in an AI second brain in Why Top Performers Are Building Third Brains With AI.
Not only that, what I also love about watching Claude think is that it gives me new ideas on how I can think better in my own life.
Why 15 Artifacts Change Everything
You might wonder: why 15 artifacts? Why not 5? Or 50?
I've experimented with different numbers, and 15 hits a sweet spot. It's enough to explore a topic comprehensively, but not so many that the AI loses coherence or times out. It forces both breadth and depth.
With 5 artifacts, you get surface-level exploration. The AI plays it safe, hits the obvious points.
With 15, something different happens. The AI has to dig deeper. It can't just rehash the same ideas. By artifact 10, it's finding connections and insights that surprise even me—and I provided the original content.
It's the difference between a 30-minute consultation and a week-long research project. Same AI, completely different results.
As Claude increases its capacity and context window, I can’t wait to see what a 100-artifact output would look like.
Beyond Content: Where This Gets Interesting
I've focused on content creation because it's the most relevant to what I do at the AI Thought Leader School. But the Infinite Prompting Method applies to any complex thinking task:
Strategic Planning: Tell AI information about your goals, your company, and your industry. Then watch it collect market research and do competitive analysis. Then, watch it generate strategic options you hadn't considered.
Problem Solving: Give it a complex business problem and let it explore solutions from 20 different angles—technical, financial, organizational, and cultural.
Research Synthesis: Upload multiple papers or reports and let it find connections and patterns across disciplines.
Innovation Development: Start with a basic concept and let it iterate through possibilities, constraints, and applications beyond what any human has ever written about before.
Framework development: Start with a goal and let it create a framework. For example, I created a virality framework. Now, I use that framework as synthetic data to improve my articles.
More broadly, you can improve practically any deep question by simply adding the following text at the end:
To most effectively respond to my prompt, generate 15 artifacts first. Then output the final response as an artifact.
The goal of each artifact should be to generate synthetic data that would help you accomplish the goal better.
Guidelines:
- Have the artifacts in a sequence that builds on each other and generates momentum
- Use multiple reasoning methods, paradigms, frameworks, and/or perspectives
- Think with deep rigor and attention to detail
The pattern is always the same: Infinite Prompting skills plus more thinking time lead to dramatically better outputs.
Not only that, there are many variations of Infinite Prompting you can use to get better and better outputs once you master the basics.
To demonstrate, let me show you exactly what happened with that prompt leadership class in more detail…
Case Study: The Class That Became an Article —Part 2
I used the following prompt to improve the draft that Claude created:
Look at the article I uploaded and the consider that the goal of the article is to create a blockbuster paradigm shifting article with a stress-tested idea that will stand the test of time and is so valuable and well-explained that it becomes the dominant paradigm.
1. Do you think the goals were accomplished?
2. How specifically did you come to this conclusion?
3. If they weren't reached, how would the article need to be updated to reach them?
4. Create 5 artifacts to help you iteratively improve the article and then publish an updated version of the article as the 6th artifact.
From there, I spent time editing based on:
Insights from my prior research on the topic
Important details I didn’t mention in class
Visuals I found online
My personal story
My taste
Fixing AI errors
During this editing phase, I also used infinite prompts to:
Refine the title and subtitle
Refine the intro
Refine the conclusion
Do a final pass of proof reading
Explore the implications of the ideas in the article
This case study is illustrative because it also shows that there can be multiple types of infinite prompts that can be chained together. In this case, the first infinite prompt created the draft. Then the second prompt iterated on it.
This case study just gives you a taste of what I’ve learned…
Advanced Infinite Prompting: How World-Class Infinite Prompters Get More Done
When you run an Infinite Prompt, not every artifact will be gold. Sometimes the AI will go down rabbit holes. Sometimes it will repeat itself. Sometimes artifact #13 will be brilliant and artifact #14 will be garbage.
This is a feature, not a bug.
The current AI models are still limited by their thinking time limits, memory, and reasoning level. And, it’s still at a stage where the best prompts are done when you use human-generated interventions in combination with independent AI thinking. Therefore, the bitter lesson doesn’t fully apply…yet.
Thus, your unique ability to create conditions for better AI thinking is your competitive advantage.
You can get better outputs from Infinite prompts when you:
Employ specific reasoning methods or frameworks in some of the prompts.
Specify some of the sequencing of the artifacts.
Determine the topic of some of the prompts.
Generate your own evaluation metrics so AI can judge the quality of responses and iterate on its responses better.
Upload your own synthetic data.
Furthermore, the additional skills you need include:
Recognizing when an insight is genuinely novel vs. just rephrased
Knowing when to let the AI continue exploring vs. when to redirect
Building intuition for which artifacts to develop further
Creating good initial conditions with your source material
Storing valuable synthetic data in your AI second brain so you can use it again
Understanding where current AIs are limited and providing crutches
It's in these bullet points where world-class Infinite Prompters will get their edge in the coming year or two.
ANNOUNCEMENT:
I teach advanced Infinite Prompting in the AI blockbuster track of my AI Thought Leader School on Wednesdays at 11:00am EST. Visit the course page to learn more and see if it's a fit for you.
The Final Moments
We're living through the final moments before thinking itself transforms. Like scribes watching Gutenberg's first prints, unaware their world had already changed.
The brutal truth? Most people will miss it. They'll perfect their prompts while pioneers build thinking systems. They'll save hours while others multiply minds.
Twenty minutes of AI thinking feels like forever when you're used to 20 seconds.
Yet those twenty minutes are nothing compared to what's coming.
On a recent Stratechery podcast, Jensen Huang, the founder and CEO of Nvidia, the company powering the AI computer chip revolution, said something that stopped me in my tracks:
Five years from now, [there] could be 100,000 times faster agentic models.
Sit with that. An AI thinking in 2030 for 20 minutes will equal an AI from today thinking for four years. Not four days. Not four months. Four years of thought in the time it takes to drink coffee.
Those who master Infinite Prompting today will be ready to harness each exponential leap. Like learning to surf on smaller waves before the big sets arrive. People who wait will be more likely to be overwhelmed or miss the opportunity.
Yes, it's overwhelming. Yes, it's thrilling. Yes, it’s scary. But here's the thing—you're curious enough to be here, reading this, which means you're exactly the kind of person who thrives when everything interesting happens at once.
We've been using AI like a vending machine when it's actually a thinking partner in its infancy, growing stronger every day—and taking those who engage with it along for the ride. At least for now.
Tonight, take your messiest problem—the one you're avoiding. Give it to Claude with the Infinite Prompting template. Then, watch fifteen artifacts unfold before your eyes like an exponential wave.
On the other side of your next twenty minutes is a different way of working forever.
Your Move
If you’re interested in an introduction to the fundamentals of Infinite Prompting, I recommend going through my Infinite Prompting course, which you can access as a basic paid member ($20/month). Along with the on-demand video course, you a get a full transcript of all of the classes, which you can copy and paste into NotebookLM to get personalized tips and support.
If you want to go deep on the advanced Infinite Prompting workflows necessary to create blockbuster articles and get 10x leverage, then I recommend the AI blockbuster track in my AI Thought Leader School. In this track, we’ll provide the ultimate set of Infinite Prompts for your AI to:
Research a topic thoroughly from multiple perspectives
Generate paradigm-shifting ideas, frameworks, and hacks no other human has shared before
Create Trademark Ideas (frameworks, protocols, hacks, etc) that you become known for
Produce blockbuster articles about the Trademark Ideas that spread virally
Furthermore, we’ll cover how you can chain Infinite Prompts together in order to achieve results that you didn’t think were possible with today’s tools.
I am both super excited and nervous. Feeling very fortunate to have this opportunity to learn more in depth from you Michael. Leaning in and letting go of my self limiting beliefs.
This might be the best thing I have read today 🤌