3 Comments

Heather, did you find Claude getting lost in the weeds at any point? How was its retention of information as the conversation went along? Was it evident that it "understood" the doc if you really dove in?

Expand full comment

No, I didn't experience that in this instance. I have used it in the past to ask questions about a freelance contract and I was comfortable with the analysis and the answers to my questions. I felt that it understood my documents in each case - but it depends on your questions. Sometimes, we need to consider what we're asking (clarity, intent, etc.) to get the answers. I think LLMs understand legal and complex text quite well, but we don't :)

I don't care for the suggested responses for something like composing an email or other things that I want to sound more like me. I use ChatGPT or Perplexity for those types of requests.

Expand full comment

Same, those are more annoying than anything (suggestions for your voice), although I suspect that'll improve a great deal over the next few years.

Thanks for reporting back! I appreciate having a line of sight into some of the LLMs I haven't experimented with yet. I'm interested in analyzing some spreadsheets my business uses, but I'm not quite comfortable sharing that data with OpenAI or anyone else just yet.

Expand full comment