Work that used to take 3 weeks now finishes in 1 hour.
This is not an exaggerated advertising slogan. This is a story told by Kei Tanaka, who served as the Co-Head of the Investment Division in Japan at Goldman Sachs, about what actually happened in his own work. Complex research and thick document preparation. A series of tasks that once required three weeks can now be handled in about an hour by combining them with AI.
Most people react by saying, "That's because he's a former elite and he's smart."
But when you actually follow his usage, it's surprisingly simple. What he is doing is neither talent nor a secret trick. It was a matter of a down-to-earth "template" that anyone can imitate starting today.
Nice to meet you, I'm tatsuki. I support AI utilization for small and medium-sized enterprises and help with the business implementation of Claude and Codex. I am also someone who runs Claude Code all day long. That's why when I see people who stop at "AI is a convenient tool for looking things up," I want to chime in and say it's not the tool's fault, it's how you use it.
In this article, I will break down the AI utilization templates seen in Mr. Tanaka's statements and public interviews into five forms that you can run immediately. I've also included prompts for each method that you can copy and paste.
One thing first. At the end of the article, I am giving away a free "Design System Kit" to make AI design smoother. I'll put the instructions for receiving it at the end, so please enjoy the main story first.
Why do most people stop at "convenient tools"?
First, let's look at where the difference lies between Mr. Tanaka and ordinary people.
Most people open AI as a "smart search." They throw a question. They get an answer. The end. In this case, AI is just an information vending machine. It is convenient. But that is a way of using it that anyone else can do.
That is not what Mr. Tanaka sees in AI. He leaves everything from deep research to hypothesis building, decision support, and automatic document creation to AI as a single flow. In other words, he doesn't place AI as a "tool to rely on occasionally," but as a "partner that reconfigures the very foundation of work."
There is a world of difference between those who use it instead of search and those who use it as a device to speed up judgment.
I believe Mr. Tanaka's usage can be narrowed down to two things. One is "templating." Turning the judgment process in your own head into language and porting it to AI. The other is "compression." Ordinary people multiplying their time through discipline and efficiency. Keeping these two axes in mind, let's look at the five methods.
Method 1: Porting the "Template" of Judgment to AI
The first one is the most effective. And it's a move that almost no one is doing.
Mr. Tanaka decides on the "template" used for investment decision-making first and has the AI learn it. On top of that, he creates a state where a draft for judgment is returned just by inputting the collected information and his own hypothesis. It's fast because he doesn't think from scratch every time.
This isn't just for investors. The more judgments you repeat in your work, the more they can be templated. "Should I continue this side hustle idea?" "Should I release this product?" Take that judgment you're always vaguely worried about, verbalize it once, and give it to the AI.
Let's try it from zero to completion. For example, suppose you are wondering, "Should I release a paid note summarizing how to use AI?"
First, pass the judgment template like this:
You will help me with my judgment from now on. I will give you a judgment template, so please remember it. 1. Demand (Are there people who want it?) 2. Reason for winning (The meaning of me doing it) 3. Worst-case scenario (What is lost if it fails) 4. First step (The smallest step that can be taken this week). Please evaluate each item in 3 lines or less based on this 4-item template.
Next, pour in the idea you're hesitating about as is:
Idea: Release a paid note (980 yen) summarizing AI utilization for side hustles. Readers are side hustle beginners. My materials are my actual experience of touching AI every day. I haven't released a single paid article yet. Evaluate it with this template.
Then, the demand, reason for winning, worst-case scenario, and first step will be returned in a filled state. What I do there is add a correction like, "Make the 4th step smaller. Make it a granularity that can be done in 30 minutes today." This is where the judgment finally moves forward.
Let the AI go from 0 to 1, and the human goes from 1 to 10. Do not reverse the order.
- ✗ "Do you think I should release this note?" (AI will only return harmless generalities)
- ◯ Pass the "judgment template" first as shown above, then pour in the idea (A judgment specific to you will be returned)
Method 2: Reducing Writing Time to 1/10 with Voice Input
Once you've sped up judgment with templates, next you cut the time spent "writing."
Mr. Tanaka makes full use of AI transcription for document creation. He speaks about 2 minutes' worth of ideas for 5 pages of A4 paper. The AI instantly turns that into text and organizes it. He says this has reduced the time spent typing text to about 1/10.
The point is not to stop your thinking. Don't try to match the speed of your head to the speed of typing on a keyboard. Think while talking, and leave the tidying up to the AI.
Pass the "messy monologue" spoken via smartphone voice input like this:
Below is an unorganized memo I spoke via voice input. Since it's in spoken language, please organize the order without changing the meaning, and summarize it into 3 bullet points and short sentences explaining them. Please keep my tone of voice.
The sentence "Please keep my tone of voice" is effective. Without this, the AI will overwrite it with its own uniform writing style, resulting in that "AI-like writing."
I also dictated the framework of this article by voice while walking. Let the AI take over the heaviest first step of sitting at a desk and saying, "Okay, I'm going to write." That alone drastically lowers the hurdle to starting.
Method 3: Write Prompts "As if Teaching a Middle Schooler"
Even if you know how to use the tool, if the instructions are sloppy, the output will be sloppy. Mr. Tanaka's words hit home here.
He says that even for complex things, he gives specific and clear instructions as if teaching a middle or high school student. Instead of throwing an ambiguous word and saying "do it nicely," he writes on the premise that the other party knows nothing about the background or conditions.
The difference appears in this one point.
- ✗ "How to start a blog as a side hustle?"
- ◯ "An office worker wants to start a blog as a side hustle. One hour a day on weekdays, and hasn't decided what to write yet. Under these conditions, think together about what should be done in the first month, with priorities."
Even with the same AI, what comes back is completely different. Treat it like a smart junior and pass it carefully starting from the background. That alone will make the accuracy of the output jump.
If you're unsure, it's easy to add this sentence to the beginning of your instructions:
To do this task well, please list 5 questions you should ask me. Start the task after I answer them.
Let the AI ask questions first. Then, the premises you forgot to pass will be naturally filled in.
Method 4: Possess a Role Model in the AI
Once you've improved the accuracy of your instructions, next you increase the number of perspectives. This is a way to go beyond the limits of thinking alone using AI.
Mr. Tanaka has the AI learn the thinking and expression styles of role models in various fields and has it analyze from multiple perspectives from those viewpoints. Get opinions from angles that your own head cannot reach.
The weakness of solo brainstorming is that your vision doesn't go outside yourself. You fill that gap by having the AI possess a different persona.
From now on, please give me opinions on my project from the positions of three people in order. 1. A cautious investor (pointing out naive premises) 2. A lazy reader (whether they feel like continuing to read) 3. Myself one year from now (whether this has become an asset). For each, give one good point and one harsh criticism.
You'll be in a state of running a meeting with multiple people by yourself. You can then ask further questions like, "How would you change the first 3 lines to convince reader #2?" Just by going back and forth a few times, the holes in the project will be filled surprisingly well.
However, do not swallow the AI's opinion whole. It is strictly a device for identifying points of contention. You hold the final judgment.
Method 5: "Content-ize" Yourself and Feed it to the AI
Finally, this is about raising the accuracy of the AI itself to be dedicated to you.
In a dialogue, Mr. Tanaka talks about the idea of "writing a diary" to master AI. Keep a record of your thoughts, judgments, and words, and keep them in a state where they can be passed to the AI. In other words, content-ize yourself.
This is plain, but it works. The AI knows nothing about you. That's why it returns an average answer like a stranger every time. But if you pass your thought logs, the AI will start returning answers that are more like you.
You don't need a grand diary. 3 lines a day is fine.
Below is a memo of things I've thought about recently. From this, please extract my values, judgment habits, and frequently used words, and summarize them in bullet points. Reflect these characteristics when writing sentences for me in the future.
Once you do this, the AI's output from then on will become "personal." For me too, just by passing a single memo summarizing the tone of my posts, the corrections to the returned text have decreased significantly.
To those who thought, "It's because he's an elite"
After reading this far, some people must have felt this way: "He's former Goldman, his natural intelligence is different."
It's the opposite.
None of these five methods require special qualifications or expensive tools. Templating judgment. Writing with voice. Giving instructions as if teaching a middle schooler. Possessing different perspectives. Recording yourself. All of these are actions you can start today with free AI.
The reason Mr. Tanaka's usage is superior is not because he has talent. It's because he is calmly and thoroughly implementing the principle that anyone can imitate: "Turn the inside of your head into a template and move it to AI." Rather, I see him as a person of the "ordinary person's template" who saw the bottom during the Lehman shock and built up from there.
Summary
As long as you open AI as a "convenient tool for looking things up," you aren't even drawing out half of its power.
- Pass the "template" of judgment first, then pour in the idea (let it go from 0 to 1)
- Reduce writing time to 1/10 with voice input
- Give specific instructions starting from premises, as if teaching a middle schooler
- Possess role models and run a multi-perspective meeting by yourself
- Record your own thoughts and make the AI dedicated to you
If you end with "Oh, I see" after reading, tomorrow you will also be holding everything in manual labor as usual. Most people do that. Will you try passing one judgment as a template to the AI just once? Or will you worry about everything on your own again today? This is the turning point.
First, try asking the AI today, "If you were Kei Tanaka, what would you say about my current side hustle?" That will be your first template.
I am distributing the "Claude Design System Kit" for free on LINE.
▼Register on LINE to receive the "Claude Design System Kit"
https://t.co/90omRA4UQ7
Please join the Open Chat (LINE Open Chat) and then from the official LINE listed in the note
**
Design System
If you send that, you can receive the benefits.
※Please refrain from sending keywords within the Open Chat
Reference: Mr. Kei Tanaka's statements and career are based on the following public information.
・The Keyperson Interview (Specific AI utilization methods, "3 weeks → 1 hour", etc.)
・PIVOT "5 Rules of Former Goldman Sachs Kei Tanaka"
・Dialogue with Kazuyo Katsuma (AI and asset management, content-izing oneself)
・Book: "People up to 100 Million, People from 100 Million: The 'Trillionaire' Mindset Revealed by an Investor with 17 Years at Goldman Sachs" (Tokuma Shoten)
