I'm working on making it possible to feed LLMs live data from websites. It is challenging because injecting webpages into LLM prompts directly has a mixed resul

I'm working on making it possible to feed LLMs live data from websites. It is challenging because injecting webpages into LLM prompts directly has a mixed result, especially if there's a smidge of ambiguity or complexity in the webpage HTML, so part of my work is on automatic methods to economically prune, format and feed web data to LLM RAG pipelines. You can follow my progress on twitter if you're interested (I tweet about other things, don't expect LLM/Data stuff only :) https://twitter.com/omarkamali https://twitter.com/omarkamali https://twitter.com/themonitoro https://twitter.com/themonitoro