Additionally, I chose to avoid AI as I have picked up from a lot of people that Google’s AI features have more or less ruined it. Do you think there is a way that I could use AI here that wouldn’t ruin things?
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.
This seems to be an overcrowded market so I would be afraid to just "make a search engine". It seems difficult to me at the moment how exactly I can make ThreadQuilt a unique value proposition if I take that route.
If AI is the differentiator, then I would absolutely want to integrate it.