Cockpit answers one question: is your AI detection doing a good job? This demo shows the whole loop in three steps, on a test account called Northwind Media with made-up data.
First, detections come in. Every check your platform sends through withConflux (AI-generated images and video, moderation, deepfakes, audio ownership) is counted per detector, with its speed, its errors, its cost and how often it flags something.
Next, the confidence chart. Most detectors return a score between 0 and 1, and a verdict flips at a threshold. The chart shows where your scores fall. The band just around the threshold is where the detector is unsure: those are the verdicts worth a second look, and the place where a small threshold change moves the most results.
Then an alert fires. In the demo the share of videos flagged as AI-generated jumps within a day. Cockpit sends an email (and, if you set one, a webhook) and explains it in plain words: which check, which provider, what changed and since when. You decide what to do next.