Your reviewers already decide what the detectors got wrong. Cockpit turns those decisions into a score per detector, shows the false alarms and suggests the threshold that cuts them.
Health from €149 a month after a 14-day trial with no card.
Reviewers spend hours on false alarms, and nobody can say which detector causes them.
Right, false alarms and missed, per detector, from your own review decisions.
Apply the suggested threshold, add a second opinion for unsure cases, watch the false alarms drop.
Made-up data on a test account. No real customer.
Only the final decision for each reviewed item: the content hash (or the request id), the check and the label your reviewer chose. No media.
Cockpit shows a score for a detector once at least 50 of its verdicts have been reviewed, so a handful of cases never decides it.
Send the new decision for the same item and it replaces the old one, so overturned verdicts count the right way.