sawickipedia

APIs so that we're not violating TOS (scraping at scale will get you blocked)

Yep pointing out features of someone’s site that aren’t optimized for LLM’s is part of it. But the bigger piece is figuring out what 3rd party content is being used as sources for a relevant topic.

SEO agencies use software like this to help their clients. We are already working with some of them. Someone has to do the work of upgrading one’s website & content & it’s usually an agency. We won’t be offering the services - we’ll partner with agencies to do that.

We’re playing with a combination of things (frontier models, open weight/open source models we’re maintaining, data) but those two items are key focus area for development.

If you click on the AI Recommendations button when signed in (click signin in the upper right to create an account), then you’ll see the recommendations on how to improve. Hope that helps!

Thanks for the feedback - we will soon allow users to input their own topics/questions - so I’m not surprised to read your preference for it at all. That being said - one of the challenges is how do you ask questions in the voice of the target persona. That’s the tricky part & why we started with the suggested questions first.

We fully expect a paid ad model to appear - perplexity seems to be the first in that direction. That being said - we see this more about managing your reputation ala credit score per the other comment than the absolute ranking game that traditional search has become.

With the move to RAG and models being open to the Internet now - results are changing more quickly. Also we ran an analysis, and for ChatGPT 4o, 51% of the sources/citations were posted in the last 90 days and 82% from the last 12 months.