We just made our entire AI-search dataset public. Every ranking, citation and crawl across 8 models, refreshed daily. Free at trakkr.ai/data.
Trakkr is open-sourcing AI search. 22 million citations. 16.4 million crawler visits. 325,000 cited domains. Eight models, refreshed every day. I've been building this to understand what AI assistants actually recommend, cite and crawl. If AI search is becoming a real discovery channel, it should have the same measurement layer that regular search has been built upon. So that's what's in here: rankings, citation logs, crawler behaviour, query rewrites. It's free and open. Take it, build on it, check our work. Link's in the comments.