Sometimes I feel like I am not using AI efficiently or I feel like I lack skills to use AI the way others are claiming to use AI on Twitter.

There are countless such comments and I don’t think they are particularly wrong or are lying.

However, I am definitely finding it hard to let go completely. For instance, I was trying to fix a bug on my app, basically the chat interface had a large blank space after every response from the AI agent. I had Claude fix that bug but it broke something else…

The interface shows list of historical sessions on the left (screenshot below) and when a user clicks on any of the sessions the scroll position lands the user on the latest response received from the AI agent. Claude broke this behavior (which is something that it had fixed earlier).

I just guessed it without looking at the code or the behavior on the browser. Once I checked the browser and then I looked at the code, I knew what exactly was broken. I pointed it out and Claude promptly fixed it but this is the reason I find it hard to let go.

Not sure how others are doing it. I might get there someday but definitely not today.