Analyzing a huge amount of emails

[earnest sphinx]

Hey all! Quick question. I am looking to use batch api to analyze a huge amount of emails. I have emails in a file. I want the end result to be a csv file, with a line for each email with email_id, email_category (categories predefined by me, LLM has to choose) and clients_questions(what were the main client questions if any). So in other words the llm should find questions from each email and put the correct category tag. I am sure this is pretty straight-forward and doable. I would love if anyone could reassure me and say that yes this is possible and it is relatively simple, as I see it, before I accept this data analysis task. I am very new to the API. Thank you!

[potent swan]

yes fully possible. the only issue i can think of is that you need to have your emails in a file that make sense. json,csv something that make it possible for the llm to know wher an email start and ends.

[earnest sphinx]

thanks so much. Do you happen to know if letting the LLM read confidential info is going to cause trouble?

[potent swan]

they dont do any learning on API but you should really have a company account aka enterpise account. otherwsie contact support for 100% correct info

but in general yes it should be totaly safe to use "secret" stuff with api

[ornate knot]

Use Azure OpenAI.

There’s like a 50/50 shot your company either already trusts Microsoft or Google with your email content, so personally I don’t see why trusting them to use AI on your emails is much of a stretch.

[tacit plaza]

Unless you are openai competitor it's fine