How my output skyrocketed when I went from "using" AI to "running" it

Hello everyone, this is Ponzu.

The other day, on a Sunday morning, I tried running three AIs at the same time, and I noticed something quite interesting.

Don't you think AI is something you use one by one in order? I thought so too until six months ago. This time, I'll write about how that assumption changed.

Sunday morning: I tried running three AIs simultaneously

There were two things I wanted to do that day: analyze my X posts and outline a note article.

First, I gave Claude my X analytics data and had it analyze my posting trends. Since it takes a little time to load the data, I opened another Claude window in the meantime and started brainstorming for my note article.

On top of that, I had NanoBanana make thumbnails on the side. In other words, I was running three things at once: two Claudes and one NanoBanana.

What I realized after trying this is that there is "no need to wait" for AI output.

Six months ago, I would give Claude one instruction and then stare at the screen. I would wait while watching YouTube until the output was finished. Then, I would look at the results and think about the next instruction.

Looking back, that was a really wasteful way to use it.

I think the trigger was subscribing to the paid plan. The feeling of "I'm paying a monthly fee, so I have to use it more" kicked in, and before I knew it, I had multiple windows open.

While one AI is thinking, I give instructions to another. By the time those results come back, the first AI has finished its output. By repeating this, I finished both the X analysis and the note outline almost simultaneously, just on a Sunday morning.

The moment it changed from "using" to "running"

What I felt from this experience is that my way of interacting with AI has changed.

When you say "using AI," you have an image of facing it one-on-one. But in reality, you can run multiple AIs at the same time and move between them. If anything, it might feel more like being a conductor.

Instead of doing the work yourself, you assign roles to each AI and oversee the overall flow. Since I grasped this feeling, the amount of my output has clearly changed.

I looked it up later, and it actually had a name

This feeling of "running AIs simultaneously"—when I looked it up later, it was close to a concept in the IT world called "parallel processing."

Originally a computer term, it refers to a mechanism that reduces overall time by processing multiple tasks at the same time.

What's interesting is that there is research showing that when humans do this, it has the opposite effect. According to research by the American Psychological Association, human multitasking reduces productivity by about 40%. Even though it looks like the brain is processing things simultaneously, it's actually just switching at high speed, so it's tiring and leads to more mistakes.

However, it's a different story if you have AI handle tasks in parallel while you just check the results. The brain gets tired because 'you' are switching tasks, but having 'AI running simultaneously' doesn't put a load on the brain.

In short, my understanding is that instead of a human who is bad at multitasking, you should just have AI handle the parallel processing.

And another thing: the psychology of 'I paid for it, so I need to get my money's worth.' I think this is close to what is called the 'sunk cost effect' in behavioral economics. It's usually discussed in a negative context, like 'making irrational decisions due to a waste-aversion mentality,' but for me, it was the opposite. Paying for it became an accelerator for my actions.

I realized through my own experience that the 'waste-aversion' mentality can sometimes work in the direction of 'starting to do something' rather than 'not being able to stop.'

By the way, for me, it was like this.

I might be writing arrogantly about 'running AI ragged,' but six months ago, I wasn't utilizing AI at all for my side hustle.

I was using it reasonably well in my main job. But when it came to my own side hustle, I didn't know what or how to ask AI. I would try it for free, think 'Hmm, that's amazing,' and that would be the end of it. It was just a repetition of that.

The turning point was restarting X in December. Once I decided to post every day, the tasks—coming up with ideas, refining text, and analyzing—increased all at once. It's an amount that would definitely not be manageable if I did it manually one by one.

That's when I subscribed to a paid plan and started using it in earnest. Furthermore, I realized that 'waiting is a waste,' so I started running multiple AIs at the same time.

As a result, I have been posting on X every day for over 60 days, and I have been updating my note every day for over 10 days.

If the me from six months ago heard this, I probably wouldn't believe it. I think the reason someone who couldn't use AI at all for a side hustle became able to produce this much output every day is not because AI performance improved, but because the rhythm of how I use it changed.

Well, the amount of output has increased, but monetization is still a work in progress, so I'm still only halfway there.

If you're interested, try starting with these things.

For those who read this far and thought, 'Maybe I'll give it a try,' here are just two things you can try right away.

The first is, once you give an instruction to an AI, try doing something else without waiting for the output. You can open another AI, reply to an email, or even brew some coffee. The important thing is to grasp the feeling of 'not just sitting and waiting in front of the AI.'

The second is, if you are only using a free plan, I think it's worth trying to pay for just one. The feeling of 'wanting to get my money's worth' changes your behavior more than you think. I have experienced for myself the feeling of shifting up a gear in how I use it with an investment of a few thousand yen a month.

Neither takes more than 5 minutes, and you won't lose anything if you fail. Why not try at least 'not waiting for the AI' first?

Finally.

To summarize, I used to think AI should be used carefully one by one, but it's fine to run them simultaneously. In fact, the story was that output increases if you run them without waiting.

When humans multitask, productivity drops, but if you have AI handle parallel processing, your brain doesn't get tired. You just need to look at the whole picture like a conductor.

Of course, I am still in the middle of trial and error myself. But just by changing my rhythm from 'using AI' to 'driving AI,' my daily output has definitely changed.

If you'd like, please try starting by stopping the habit of just sitting and waiting in front of the AI. I'm sure it will be a hint to finding your own rhythm.

Well then, I'll be heading off now. 🐧