This is something I've been working on for the past few months. The site is live, but still alpha and not production ready. For now, I only support connected to ZenDesk help systems, though I hope to add more support in the future.
The basic idea is to apply outlier detection techniques (in this case, distance based techniques) to online customer support systems. The outliers that pop up tend to be very interesting in some way, and more so, knowing what the 'inliers' are is also useful (or at least I hope it's useful for a business.)
Currently reaching out for alpha testers, but I just wanted to get your thoughts on it. I built it in Django, since Python is my bread and butter. I use Celery for asynchronous processing of long running events (such as the actual outlier algorithm), and have everything hosted on AWS EC2 instances for now.
