Hello! A general question here, in case someone who had the same use case read this: I want to analyse (semantically) a 14.000 rows csv files. It’s about products. To be categorized in different ways, whose type and ingredients must be identified etc. LLM tend to be lazy. How will you integrate the csv with the assistant/agent to ensure an exhaustive analysis?
Hi! I'm going to ship something tomorrow that might help with your use case. Do some column in your csv have long content? Let's say above 500 chars?
When we use the table query tool, the output of the query is a csv file. Basically what we want to try is: if this csv file contains some long text, semantic search might also give some good results and in that case we will generate an additional file that looks like a json and is optimized for semantic search into the conversation.
Regarding mine some cells could reach this amount as I concatenate all of them to create an index. I have another use case concerning user research - interview quotes - where it could definitely be the case.
Ok so I should ship it tomorrow morning, I can ping you here when it's live and you tell me if that works better for your use case?
great, thanks
It's shipped, let me know if it makes a difference for you 🤞🏻