Hi, I'm the founder of Overseer Labs. Companies that are data driven and care a lot about reliability seem to suffer from the "too much data" problem during an incident. Thus, we wanted to develop a tool that would help them sift through their data and help them root cause the problem faster.

Overseer Labs leverages machine learning to model system behavior during normal periods of operations. Afterwards, the trained model is applied in real-time to find strange behavior and then providing a ranking of all your metrics. By correlating these insights with your PagerDuty alert, you will be able to resolve your production problem faster.

We put up a demo of one of our algorithms so you guys can play with it on a small dataset. No algorithm will work perfectly out of the box for all the datasets out there, but if this looks useful for your company, we can tune it for you.

I'd love to get feedback from all the data driven companies here on whether or not something like this could help.