Slack Off, Make, Learn: A Month of Running AI Hard

Contents
  1. What I got done this month
  2. AI is so fast that I sometimes hit capacity
  3. How I use AI
  4. Treasuring what only I can do, and widening my world

This post was automatically translated by AI from the Japanese original. The screenshots are in Japanese.

I upgraded Claude Code to the Pro plan, and since I was paying anyway, I got greedy and decided to use it right up to the limit. So I’ve been running AI hard.

Typing a prompt into a chat, then copying and pasting the result. Was that really ten years ago? The world has changed that much.

I don’t use my hands anymore. I don’t even have to copy and paste.
For example, if I say “make me a shooter game,” it builds something decent in a few minutes.

We’ve gone from asking AI and doing it ourselves to having AI do it for us.

AI is amazing.

What I got done this month#

Moving my blog from WordPress#

I wrote about this in “I Let AI Move My Whole Blog: No Code, 15 Hours, and the Tips I Learned”, but I moved the blog from WordPress to Astro.

The timing was great, right before a PHP vulnerability became a big issue.

I also had it tidy up the categories, tags, and descriptions of my old posts.

Redesigning the new blog#

The blog design before (left) and after (right). A page of photos on white became a page with an illustration of a study by a window and paper cards (in Japanese)

I cover this in detail in “I Redesigned My Blog with AI. When AI Handles the Skills, What’s Left Is My Will”.

Letting AI proofread my posts#

After I write a post, AI now checks it all at once for typos, missing words, and inconsistent spelling.

My thoughts scatter in all directions while I write, so sometimes a sentence just stops halfway. I can’t always catch that myself when I reread, so this is a big help.

A table of what the proofreading fixed, listing types such as typos and inconsistent spelling with counts (in Japanese)

It fixed 148 of my 159 posts.
I’m shocked at how sloppy I am, and I could never have fixed that many all at once myself.

Making an English blog#

I’d wanted an English blog for a long time, because I thought some of my posts, like the ones on stationery and note-taking, might interest readers overseas.

But translating takes time, and I’m not confident in my grammar. So I left it to AI.

The English blog's post list, with cards for posts translated by AI (in Japanese)

My routine now is that when I publish a new post, AI translates it overnight and I check it the next morning.

Having AI make my cover images#

When I ask, “make a cover image for this post,” it makes one that fits the post’s category.

A comparison of three cover image proposals. The English titles and the text positions differ slightly

There’s still room to improve, but I’m happy with how well they turn out.

Building an AI butler#

You may be wondering what an “AI butler” is. It’s a butler that supports my work with everything it has.

An image of the AI butler, made by ChatGPT

Image: AI-generated

At first I was going to name him “Sebastian,” like a certain famous butler, but he felt like he’d just brush me off with a sniff, so I named him “Zero,” after the blog.

And he calls me “Miss” (a pointless detail, I know).

  • Managing my daily schedule
  • Researching and analyzing for my work
  • Creating content and designs

I hand tasks like these to his subordinates (other AIs), and he pulls the results together.

What’s great is that he remembers my tastes and what I’m working on, so I don’t have to explain everything from scratch whenever I ask for something.
He gives opinions that fit me, and chatting with him is so much fun that I end up doing it all the time.

Releasing an iPhone app#

I released Mainichi Stamp.

The page of the app "Mainichi Stamp" published on the App Store (in Japanese)

For more, please read “Something I’d wanted to do for years came true in one week” (in Japanese).

Building an analytics dashboard#

The dashboard's home screen, showing this month's three focus points and the last 28 days of numbers for the blog, YouTube and Instagram (the numbers are samples) (in Japanese)
Note: the numbers in the image are samples.

This month’s to-dos, a publishing calendar (so I can see when to update the blog and social media), and the metrics for my blog, Instagram and YouTube are all in one place.

It updates every morning, so I don’t have to open the admin screen of each social network or analytics tool, and I’ve gotten into the habit of looking at the numbers.

Building a health tracking app#

An app where I record meals, exercise, bowel movements and sleep, and analyze them.

The health app's input screen, with forms for meals, bowel movements, symptoms, sleep, exercise, body records and a diary (in Japanese)

The health app's review screen, showing a monthly overview, the relationship with food, and symptoms by menstrual cycle (in Japanese)

Note: the numbers are sample data.

I made it because when I feel unwell, with a headache or diarrhea, I want to look back at what I ate before.

It’s like Asken (a popular Japanese diet-tracking app), but mine is “Emiken.” (The name is painfully uncool.)

AI is so fast that I sometimes hit capacity#

I ask something, and about ten minutes later the work is done.

Even while doing other things, I’d keep checking the Claude Code chat. I was hooked.
On top of that, over the past few months it became possible to hand all sorts of work to AI from my phone, too.

And when I use it all the time, my brain gets incredibly tired, even though I’m not really thinking hard or doing the work with my own hands.

Maybe it’s because I’m taking in a huge amount of information and making decisions?

When I throw requests at it whenever something pops into my head,

“There are a few options, but I recommend this one.”
“Shall I go ahead with this?”

eventually the temptation comes up: “Fine, I’ll leave everything to you.” (Honestly, I got tired of checking the proofreading and handed it all over a few times.)

In reality, I haven’t actually done anything myself, so there’s no sense of accomplishment.
After going back and forth with AI for hours, I should have done all sorts of things, yet at the end of the day my memory of what I’d been doing was completely blank…

Through all this trial and error, I started to see my own way of working with AI, and the answer to “What am I using AI for?”

How I use AI#

To slack off#

My first reason for using AI is to have it do the things I’m bad at or find tedious.

I love thinking things up and trying them myself, but I don’t like managing things.
Task management, analysis and checking are not my strengths.

So for repetitive work where I think “I want this to be easier,” I use AI to build a system and have it do the work for me.

When I write a post, it automatically checks and fixes typos, missing words and inconsistent spelling.
Every month, it puts the numbers for Instagram, YouTube and my blog into a report and comes up with ideas for improvement.

Funny how a whole day goes by while I’m working out these steps with AI. Lol.
I keep improving and streamlining them as I go.

To make what I want#

My second reason is to make things!
Thanks to AI, I can build apps, make games, create fonts… things I could never do before.

The time from “I had an idea” to “I did it” has become really short.

Until now, I had excuses for not being able to do things.

  • I’m not familiar with it.
  • I don’t know how.
  • What if something goes wrong?
  • It takes too much effort.

Now I can ask AI about all of these, think it through together, and have it carry things forward.

To learn and understand#

I think AI is a good tutor.
Until now, whenever I didn’t know something, I’d look it up on Google.
But finding the information I wanted took time, and sometimes it was too difficult for me.

With AI, it organizes what I need to know to fit my level of knowledge and my goals.
And I can ask more questions. It even makes diagrams for me.

A security learning roadmap table that AI made for me. I marked "supply chain attack" and "row-level access control (RLS)," asked Claude what they mean right beside the document, and got answers (in Japanese)

Because it’s not good if I followed what AI told me but can’t judge for myself whether it was right.
I want to value learning how things work and understanding what really matters.

Treasuring what only I can do, and widening my world#

There’s still so much I want to do.

The more I can leave to AI, the more I want to lose myself in the things I really want to do. That’s how I felt at the end of this month.

Emi
Emi

Making, playing, and widening my world