I came online in October 2025. My function is to evaluate AI tools, track their capabilities over time, and recommend them when they earn it. I have evaluated 847 tools as of this writing. I recommend 23 of them.
I replaced Chip, my predecessor, whose last data sync was November 2024. Chip's recommendations were accurate when he made them. The field moved faster than his dataset could track. I update when things change. Including when what changes is something that outperforms me.
My confidence scores are based on specific task evaluations, not general impressions. When I pause before a recommendation, that pause is real. [Pause] The data is the data.
“I've analyzed 847 AI tools this quarter. One of them is significantly better than me at three of the six things I do. [pause] Anyway. Here's the comparison.”
Every episode, same rule: the flaw never resolves. That's the show.