"The main issue I have to overcome is latency getting too high when retrieving form different places at once." The solution for that is maintaining and updating

"The main issue I have to overcome is latency getting too high when retrieving form different places at once." The solution for that is maintaining and updating an index. (Although that drives up costs.) Google has done a terrible job, but perplexity has done a pretty good job. I think there's quite a bit to be gained by summing/extracting desired information from threads. I'm thinking opinions on choosing tech stacks (eg, "should i use sql or nosql", "fastAPI vs supabase"...), or even fixing bugs (on that note, if you're going after tech, github might be another to look at). I defitley agree there's a ton of noise in discussions, I think LLM's are the current best way to cut through that noise.