"AI Side Hustle for Hospital Administration" - Experience Becomes Value When Redesigned [True Story Vol. 19]

Former Hospital Administrator'sAISide Hustle True Story — Part 2: Redefining Experience

Can I turn what remains in the field into value?

Vol. 18 The night I published it, I was wide awake in bed.

That question alone was spinning around and around in my head.

I opened the Coconala screen. My listing page is there. But there are no inquiries. Earnings remain at 1,700 yen.

I knew my experience was a weapon. There were things left in the field that AI couldn't produce.


But just having them left over didn't change anything.

Redesigning Experience

I thought my 20 years were just experience as a hospital administrator.


But I was wrong.


At one hospital, I made enemies of everyone. I was an existence that destroyed the paradise built by the old regime.

The staff who joined with me were tainted by the old regime. Bring gifts. Flatter them. Do as you're told. That's how they were co-opted.


I didn't get tainted.


Why?


Simply because I thought there was too much waste. Everyone says they're busy. But if you organize the work, the burden decreases. No one questioned that. It was just because they had always done it this way.

It felt like cleaning. It's messy, so you clean it up. That was all.


That's why I didn't get tainted. I had no intention of entering their paradise.


My evaluation from management was low.


The people of the old regime were good at approaching management. They were also good at performing. I didn't have time for such things. I thought it would be truly dangerous if I didn't do something from the field level.

So, from management's perspective, the people tainted by the old regime looked like they were working very hard. My evaluation was always about what I was doing.

There was no one who could evaluate what I was doing.


Even so, I kept moving.


The experience of moving while being understood by no one. The experience of not being tainted. The experience of trying to destroy a paradise. The experience of not being evaluated.

None of that exists as text.

AI cannot learn it. It cannot reproduce it. No one can imitate it.

We have entered an era where 20 years of hospital administration experience can be combined with AI.


It was a completely custom-made product that only I could create.

The True Nature of What AI Cannot Do

AI operates on a mechanism that predicts the next word from vast amounts of text.

That's why it looks like it knows everything. But in reality, it can only learn what exists as text.

It sometimes confidently lies about things it doesn't know as if it does. If a conversation gets long, it might completely forget what it said earlier. Given the mechanism, it was natural for that to happen.

The atmosphere of the field. The fluctuations of human emotion. On-the-spot judgment. No one has written those down. So AI doesn't know them either. Things that cannot be put into words cannot be passed on in the first place.

No matter how precisely you construct instructions, you won't reach them. Even if you break down the steps, show examples, or assign roles. Things that aren't written down won't come out, no matter what method you use.

No matter how much AI develops, it cannot become the person themselves who has read all of that person's experiences.

AI also teaches you prompts. AI also teaches you procedures. AI also does the output. If you are leaving everything to AI, then "that person" is not there. You are just publishing something made by someone else called AI instead.

I Rewrote It

I plucked up the courage to get out of bed and turned on my computer.


I opened my Coconala listing page.


Vol. 1 The wording I wrote in was still there.

"I will create operational manuals for medical institutions using former hospital administration experience x AI."

It was a description of functions. What I can do. That was all.

I also looked back at the manual I made in Vol. 1.

STEP 5, "Instructions for switching to paper-based operation." In the person-in-charge column, it just said "Medical Affairs Leader."

It said what to do. But who takes command? Who starts it, and in what order? How to build a relationship with the outsourced contractor? It didn't go that far.

I wrote it in Vol. 17. Experience was a weapon. I wrote it in Vol. 18. There were things left in the field that AI couldn't produce.


I put that experience back into the manual.


I created a new chapter in the manual called "Designation and Authority of the Commander."

<<What I added↓>>

I decided on the priority. Who becomes the commander? Until the commander arrives at the scene, the discoverer acts provisionally.

I also wrote what the commander does. Individual instructions to each department. Proactive instructions based on how the response of one department affects others.

It was what I actually did in Vol. 17. I gave instructions to the person in charge of outpatient doctors. I gave instructions to the person in charge of nurses. Also to the laboratory department, radiology department, and pharmacy department. Even within administration, I sent separate instructions for each department.

For the first time, I put into words the judgments my body made on its own at that time.


I also created a new chapter for coordination with outsourced contractors.

Ask about what they are troubled by in normal times. Express gratitude. Try to understand the field. That determines the speed of coordination during a failure.

It was what I actually did in Vol. 18. The person who shrugged every time I spoke to them. The time I spent visiting them many times to close the distance. I rewrote that experience into a form that anyone can practice.


I also rewrote the wording on the listing page.

Before.

"I will create operational manuals for medical institutions using former hospital administration experience x AI."

After.

"The moment the electronic medical record system stopped, hundreds of people rushed to the reception. In a field without a manual, there was no one who could give instructions."

<<Link to Coconala service page↓>>

It changed from a description of functions to a sentence of experience.


What I made in Vol. 1 and what I made now. It was the same product called "Manual Creation Agency."


But the content had become something else.


I put my experience into words first, then had the AI assemble it. Using the same AI, doing the same "manual creation" work, the finished product was completely different.

Your experience is the same

Everyone has it.

Experience that didn't go well. Experience of moving in an unreasonable environment. Experience of not being understood by anyone. Experience of not being tainted.

That doesn't become text. AI cannot learn it. No matter how much it develops, it cannot reproduce it.

Things that can be generalized can be replaced by AI. But the path that person has lived, the experience they have accumulated, and the emotions of that moment are things only that person possesses.


When you combine that with AI, it becomes the real thing for the first time.

The Real and the Fake

A manual made by leaving it entirely to AI looks clean at first glance. But the moment a trouble occurs in the field, it is not useful at all.


A manual made by mixing in experience, even if rough, saves the field at that moment.


If you leave everything to AI, anyone can make the same thing. That is also one way to enjoy it. But that is not your own power, it is the power of AI. If you aim for something only you can create, your own experience and skills are needed there.


You use AI with things only you possess. Only then does AI become a tool. The experience and skills that person has cultivated make the AI the real thing.

I closed the Coconala screen.

I reread the Vol. 1 article I wrote at the beginning.


"Only after quitting did I realize that the way to produce value is different."


That's what I wrote.


It was correct at the time. But it was not enough. Just realizing it didn't turn into value. I needed to put that experience into words before the AI.


The manual from Vol. 1 and the rewritten manual. That was the only difference.

Honest Current Status

----------

Earnings: 1,700 yen
CrowdWorks: Registration complete, waiting for projects
Coconala: Page improved, waiting for inquiries
X: 380 followers, operating as an advertising tool
note: 950 followers, main battlefield
Paid articles: Published, fully updated
Interaction with AI: Managed by dividing into processes
note series: Published up to Vol. 19

----------

The number is still 1,700 yen. From the outside, it might look like a small step.

But I learned the difference between the real thing and a fake.

I've kept moving for 19 installments. Noticing, fixing, and moving again. It was a cycle of that.

What I arrived at wasn't the profit. It was realizing that my own experience is the material that makes AI authentic.

While staring at my smartphone screen in bed, I suddenly thought: I'm still on the path. But I'm not lost anymore.

That much, at least, was certain.

(To be continued invol. 20.)

👉The series articles are collected in a magazine.

For those thinking about an "AI side hustle," let's go through the trial and error together. Send questions to my X DM.