I work on a few product methodologies, so my job is to create new knowledge. I’ve been doing it almost every day for 8 years.

Before AI, it took a few hours to get a new insight. It was slow, painful, exhilarating and extremely rewarding. I got used to it. Every insight felt like a birthday present to an 8-year-old and a great achievement at the same time.

I had my workflows: an A4 notebook, a lot of thinking while drawing, searching for information, reading papers, reading business, economics and psychology books. Building out a new insight or a new area of knowledge was slow and steady.

Now, with Claude Code spinning in a few tabs, I can create ENORMOUS amounts of insights per day. This May I planned a one-month vacation, which turned into a non-stop, crazy work sprint. I described in detail around 200 previoulsy vague ideas I had about my methodologies and found 100+ more. That’s A LOT. In a good year, I could find or describe around 100 ideas.

The sprint began with an IDEA: “What if I could validate or discover methodology theses with AI?” A day later, I found myself working 14 hours a day.

I was extremely excited.

But also overstimulated and frustrated.

In my regular ideation/validation process, building out an idea was slow, and the insight was born in my own brain and psyche.

With Claude Code, building out an idea is often instantaneous, and the insight is often born inside Claude’s machinery. I just validate and tweak it. The amount of information passing through my brain during ideation/validation with Claude Code is huge.

It ends up like this:

  • I create a lot of insights in a day
  • I’m overstimulated by the amount of thinking and emotion
  • I don’t feel that they’re the result of MY intellectual work
  • I forget 2/3 of these insights a week later, which NEVER happened when insights came to me the old way

It’s an odd thing to feel and to do. But still crazy effective. 10/10, highly recommended.

I think part of the problem was Opus 4.8, the SOTA model at the time, which was infamous for its “Claudish” writing that fried readers’ brains.

But I also think we need to rethink the routines and psychology of creative work for the AI era.

[btw, I find Anthropic models much better suited to creative work]