The creator of Claude Code just revealed his workflow

> "If you're not reading the Claude Code best practices straight from its creator, you're behind as a programmer,"

The amount of “you are doing it wrong if you don’t get on our bandwagon” rhetoric I see surrounding AI coding has me convinced that this is a bandwagon I don’t want to be on. That level of insecurity is just not for me.

When these companies run out of VC hype money, what's the actual cost of running 10-20 instances of this at all times, going to be?

I appreciate the validation that he doesn't customize it much. I see a lot of people creating really complex agents/workflows that I tried to replicate and always came across to me as more trouble than they are worth. Kinda like 10 years ago when people would create complex workflows for storing their notes.

He uses 5-10 terminal tabs and 5-10 web tabs of … you got it - Claude Code.

Ah man! The only situation I can see this working out for me is, it’s a greenfield project and backward compatibility is never a bother in future versions. Even the 5-10? Maybe like 3.

This submission came later but is getting more traction, with 81 comments so far: https://news.ycombinator.com/item?id=46470017

I think it would be better to have one of each of the main AIs / coding tools running instead of all the same model and tool. That way you can keep tabs on latest developments while doing your work. Also if you use multiple providers / tools, you can take advantage of more free credits that way, because some providers have a few free credits each month.