Thinking about AI costs led me to organizational theory

I recently increased my Claude usage limit.

"Alright, now I can use it as much as I want!"

……or so I thought for a brief moment.

I still find myself worrying about the cost (lol).

If you have the money and can reliably recoup the investment, I think you should go ahead and use the top-tier models as much as you like.

But honestly, I'm not the type who can recoup that much.

I want to increase my budget... a little more (lol).
I want more time too (crying).

In the midst of this, I recently asked an AI to analyze a session, and it told me,

"That way of using it is a bit wasteful."

It taught me that.

I see.

In that case,

how can I maximize results with minimal cost?

I started thinking about nothing but that.

Improving token efficiency.
Thinking about division of labor.
Figuring out how to use different models effectively.
And so on.

・・・

Then, it suddenly occurred to me.

"Wait, isn't this the same as building an organization?"

Hiring expensive talent doesn't solve everything.
Spending a lot on outsourcing doesn't guarantee results.

Within a limited budget,

Who do you entrust with what?
How do you combine them?
And what do you do yourself?

This is something business owners think about almost every day, isn't it?

It is surprisingly similar to what I think about when using AI.

In the past,

"Doing it yourself"
"Hiring people"
"Outsourcing"

These were the options.

But now, there is one more option added to that,

"Entrusting it to AI"

That is the option that has been added.

That is why lately,

I have been viewing AI less as a "convenient tool."

Rather, it's a new team member.
No, a somewhat unusual external partner (laughs)

That is the feeling I get.

When you think about it,

Prompts are how you give instructions.
Context is handover and information sharing.
Model selection is personnel selection.
Agents are team composition.

It's not everything, but there are parts that overlap with human organizations.

That is why, even though I am supposed to be working with AI, I start thinking about organizational theory and management.

I think this is quite an interesting phenomenon.

Perhaps that is why,
I have been working more with companies that handle human resources over the past few months,
and my connections with the people behind 'Ultimate Team Building' (though the name has changed now) have been gradually returning.

I have a feeling that everything is starting to connect into a single line.

'Ah, this is the time for me to be thinking about this.'

That is what I have been thinking lately.

And, a small dream of mine recently.

I want to build a 'Playful Agent Team' as soon as possible (lol).

A kid who just keeps doing research.
A kid who keeps expanding the conversation.
A kid who pokes at me, asking, 'Is that premise really correct?'
A kid who organizes documents beautifully.
A kid who praises me a lot.
Conversely, a kid who gives me a kick in the pants to keep me moving.
And a kid who occasionally goes rogue (lol).

And I, right beside them,

'Oh, that's interesting. Let's play around with that direction a bit more.'

I feel like I'll be grinning while saying things like that.

I often hear talk about AI taking away human jobs.

But to me, it feels like the more AI increases, the more opportunities I have to think about 'what an organization is' and 'what only humans can do'.

Perhaps, while I think I am learning about AI, what I am actually learning most is how to connect with people—that is what I have been thinking lately.