Aurora runs 19 research-grade statistical methods locally and cites every claim it makes. These are the experiments: what the methods found, what they refused to conclude, and the exact command to reproduce each result on your own machine. No screenshots as evidence, no numbers without a named method behind them.
I pointed Aurora at a climate buoy dataset and got 20 cited findings, each one naming the method behind it. The interesting part was not the findings. It was the Gaussian process regression that hit its 90 second budget, gave up, and told me exactly that. Here is why the method that failed is the one that should earn your trust.
Read the experiment →