So, to the question, what have you used them for, how and with what outcome? Curious to hear what the setups look like, what models and what considerations went in setting up these solutions.
Thanks!
Half the time when my colleagues ask me stuff now I think "I wish they would have asked an AI first"
Also, I am getting ready to release the new version of my website which will use GPT-4 to control containers to help with installing/configuring software or creating simple web applications. It is intended to be used by non-programmers mainly for experimentation. Little worried about the recent slowness of the API for the gpt-4 model though. I think they are rate limiting the API per the user field but don't know what the limit is. My application might not work with gpt-3.5.
I have done a few contracts for other companies also. One wanted a bot in their Slack which could answer questions based on uploaded PDFs. That used Llamaindex.
On another contract I had it write KQL queries for a specific database table based on the user's question, then display the resulting data. It could optionally generate and insert graphs on the fly if that was requested.
https://neurotechnicians.com/p/generative-ai-and-anki-part-1...