Some things that impressed me particular are:
* It can often give you working urls of images related to your query. (e.g. give me some images to use with this blog paragraph).
* It can list relevant publications corresponding to paragraphs, chatgpt often halucinates new papers but claude consistenly gives highly cited and relevant suggestions.
* It can work with pdf inputs.
I just tried again now and asked it about HTMX and Django since I had just been reading an article about it. Claude invented a package called `htmx.django` and a decorator that could be imported called `@hx_request`. This is typical of my experience with Claude.
Out of 1000 people flipping coins over and over again, on average people get a mix of hits and misses - there will always be someone who just get 10 heads in a row and someone else with 10 tails in a row, this corresponds to some users being wowed by the superintelligence nailing the answer every time and other users getting made up citation after made up citation.
But I am using it often to chat with PDFs and it works great for that. I would not want to give this ability up.
Side note: this would get you a crackpot moniker a year ago. Chat with pdfs? Dude what are you smoking I need some too.
Perfectly normal today, in fact if you aren’t doing that you’re falling behind.