How My 20 Years of 'Useless' Skills Suddenly Became Monetizable

How I Started a Side Hustle After Coast FIRE

I started a side hustle after achieving Coast FIRE (the state where retirement savings are fully funded; specifically, 20 million yen at age 35).
It began with projects worth a few thousand yen per month, and before I knew it, my rates kept rising.
Although I am the same person, using the same computer, and working the same number of hours, the scale of my earnings has completely changed.
Specifically, in my case, it grew into projects worth hundreds of thousands of yen per month in about three months.

I took a step back to analyze what had changed.
To put it simply, it is a story about how my market value was transformed by the "multiplication of five abilities."

There were five abilities whose market value was almost invisible on their own.
One day, they suddenly connected, creating a combination with almost no competition.
I myself haven't changed. The market has.

Auditing the "Useless Skills" Accumulated Over 20 Years

First, let's audit the abilities I've accumulated over 20 years.
Honestly, at the time, I did all of them while wondering, "What is the point of this?"

The "Errors are Information" Intuition from the Assembler/C/C++ Era

Back in my technical college days, I used to run microcontrollers using Assembler and C/C++.
Debugs back then were primitive compared to today. There were almost no helpful error messages; only blinking LEDs or strings output to a serial port were my clues.

Three days before the Kosen Robocon (Robot Contest) finals, a robot I had spent a month building suddenly stopped working.
It worked the day before, but when I turned it on, the motors would growl for a second and then stop.

I had no idea what the cause was.
In situations like this, there is only one thing to do: manually insert debug messages throughout the entire code.
I added serial outputs and LED triggers to every key part of the process.

• Serial.println("STEP1: init done")
• Serial.println("STEP2: sensor read OK")
• Serial.println("STEP3: motor drive start")

This way, I could visualize step-by-step how far the process reached and where it stopped.
After repeating this throughout a sleepless night, I finally identified the cause as a solder crack (the board's solder had cracked due to vibration).
The code wasn't off by even a single byte; it was a hardware failure.

At the time, I thought of this experience only as "gritty, tedious work."
I wasn't building flashy algorithms or learning new technologies. It was just repetitive work, endlessly inserting and removing print statements.

However, this intuition is directly linked to the Claude Code era 20 years later.
When I have AI write code and an error occurs, I can naturally give instructions like, "Insert a print statement here and check the value of the variable at this point." This is because I have internalized the knack for where to place them to identify the cause.

The 20-year-old intuition that "errors are information" and "print statements are visualization tools" was directly converted into debugging skills for the AI era.

Numeric Constraint Thinking Ingrained Through 5 Years of Machining Technology

For the first five years after leaving technical college, I worked in the machining technology department of a machine tool manufacturer.
What was drilled into me there was the mindset of "converting everything into numbers."

An instruction like "make it look nice" wouldn't pass.
Unless you bind it with standard values, such as surface roughness Ra 0.8 or less, or a tolerance of ±0.01mm, the operator doesn't know what to aim for.
Feed rate, cutting speed, tool selection. Only after breaking everything down into numerical conditions can the shop floor operate.

At the time, I wondered, "Why be so obsessed with these tiny numbers?"
It would be easier if "about this much" was enough, I thought.

Quality Control Thinking Ingrained Through 10 Years of Process Management

After the machining technology department, I spent 10 years in the quality control department.
Here, the mindset of change management and process capability became deeply ingrained.

Process capability index Cpk 1.33 or higher, and within the specification limits (USL/LSL).
Processes that cannot maintain these figures will eventually produce defects, no matter how much effort you put in. Furthermore, when a drawing changes in one place, you must circulate a confirmation document for all related processes, jigs, and inspection items. I operated under the premise that "a single change always has a ripple effect on the whole."

At the time, I thought of this as nothing more than "troublesome paperwork." I questioned who it was helping, as it seemed to only increase paperwork without creating anything new.

The Ability to Self-Propel and the Stamina for Trial and Error Cultivated at Robocon

Another thing ingrained in me during the Kosen Robocon was the stamina to never give up until it works.
Faced with something that doesn't work, I learned to form a hypothesis, test it, and if it fails, move to the next. I would repeat this endlessly. This "stamina to self-propel" is something learned not through study, but by getting beaten up on the shop floor.

These four items (skills accumulated in manufacturing) were all things I had spent 20 years accumulating while wondering, "What is the point of this?"

Writing as a Side Hustle Also Belonged to the "Useless" Category

Changing the subject, let's talk about writing as a side hustle.

After achieving Coast FIRE, I had some spare time, so I started a side hustle. It began with writing projects. Writing one piece would earn me a few thousand yen, which was less than minimum wage when converted to an hourly rate.

Honestly, it wasn't worth it. I thought many times, "There's no point in doing this," and I was about to quit.

However, the next turning point came just before I was going to quit.

I realized that, without me noticing, my ability to structure content and my reader-centric perspective had been building up.

• Put the conclusion first
• Keep sentences short
• Put concrete examples first, abstract theories later
• Organize information in the order the reader wants to know it

These were things I had picked up naturally while completing dozens of writing projects. Although I had thought of it as a "dull task for a few thousand yen a month," it seemed like something else was accumulating in my skills.

The Moment I Acquired AI Skills, Everything Connected

This is where we get to the main point.

When I started getting into Claude Code and "vibecoding" in earnest, my initial perception was that it was just "a task to verify the code written by AI."

However, as I used it more, I realized something:

My past "useless" skills and my side hustle writing, all of it, were being transferred here.

Let me describe specifically how they were transferred.

Numerical Constraint Thinking -> Specifying Prompts

Instead of saying "write it softly," I can write prompts like "no more than two consecutive sentences with the same ending, kanji ratio under 60%, and an average of less than 50 characters per sentence."

It is the exact same structure as the sense I used when constraining the shop floor with Ra0.8 or tolerances of ±0.01mm. I do not pass abstract expressions to the AI; I always convert them into numerical constraints. This way of thinking was already ingrained in me.

Change Management Thinking -> Designing CLAUDE.md

I separate project rules into CLAUDE.md files.
"List the scope of influence before making changes," "limit functions to 50 lines," "clearly state forbidden areas for modification."

This is the exact same structure as the "drawing change workflow" in quality control.

Debugging Intuition from Robotics Competitions -> Error Isolation in the AI Era

When an error occurs in code written by AI, I can naturally instruct it, "Insert a print statement here to check the value at this point."

I can make an educated guess about where to insert it to get closer to the cause. The sleepless nights I spent filling robots with serial output right before a competition turned into a monetizable skill 20 years later.

Writing Structure -> Prompts, Applications, Zenn Articles, and Client Explanations

My structural writing ability, which I thought was not worth the few thousand yen I earned, is effective in all directions.

Giving instructions to an AI is structurally identical to writing for a reader. The persuasiveness of an application, the clarity of an explanation to a client, the readability of a Zenn article—everything was an extension of what I had accumulated through the writing I was about to quit.

Before I knew it, 20 years of "uselessness" and the "unprofitable experience" of my side hustle had connected into a single line.

The Moment Multiplication Became Possible

Here, I will analyze what happened in terms of structure.

Evaluating Each Element Individually

• 15 years in manufacturing alone: A standard skill within my main job; its market value was hard to see.
• Writing alone: A few thousand yen per month. I was about to quit.
• AI skills alone: Everyone started having them as of 2026, so it's not a differentiator on its own.

None of these stood out in the market individually.

The Moment Multiplication Happened

"Able to use AI, verbalize on-site issues, and deliver with clear writing."

Once this combination was formed, there was almost no competition.

• There are more engineers who can use AI.
• There are a certain number of people with on-site experience in manufacturing.
• There are people who can write clear text.

But there were almost no people in the market who could handle these three axes alone.
Moreover, despite its high scarcity, the time I spend working hasn't changed.

I Haven't Changed, the Market Has

This is the most important structural analysis.

I have been the same person for 20 years. I didn't suddenly acquire new abilities. It was the market that changed.

With the advent of AI, a market was born to monetize skills that were previously outside the evaluation axis.

• The modest ability to "read errors as information" becomes direct profit when paired with AI.
• The on-site sense of "constraining with numerical specifications" is valued in prompt design.
• The mindset of paperwork, like "estimating the scope of impact of changes," functions in CLAUDE.md design.

These abilities existed 20 years ago. The market just didn't have a mechanism to purchase them. AI created that market.

A Proposal to Readers

After reading this, some of you might think, "I don't have any material to multiply."
But it's worth inventorying your skills a bit more.

The more you think, "I've only accumulated useless skills," the more you should try an inventory.

There is surprisingly little waste in what you have done for 20 years. It just wasn't useful under the evaluation axes of the time; it looks different under the evaluation axes of the AI era.

In particular, the following three have become rare assets today:

• Ability to convert abstract to concrete (standards, numerical constraints, verbalization)
• Ability to read errors as information (cause isolation, hypothesis testing)
• Ability to estimate the scope of impact of changes (impact confirmation, documentation)

Even if their market value is not visible individually, they create scarcity the moment they are multiplied with AI skills or writing.

Side hustles are the cheapest way to measure the market value of these multiplications. It might only be a few thousand yen a month at first. Even if you feel it's not worth it, it's worth trying for a few months. You won't know where your multiplication hits the market until you try.

And Coast FIRE serves as a foundation to carry out this trial and error without rushing. When you start a side hustle without the anxiety of lacking retirement funds, you can spend time "testing the multiplication" rather than focusing on the immediate unit price. In a state dependent solely on your main job, this experiment is too risky to step into.

The "useless things" you have accumulated for 20 years can become monetizable at a certain point. That phenomenon is happening with the arrival of AI.

I invite you to inventory your past once more.

Feel free to contact me via the email in my profile for consultations on AI utilization and business automation in manufacturing.